- ORNL-TM-3963

 

 

ZERO-POWER EXPERIMENTS WITH
233 IN THE MSRE

J. R. Engel
B. E. Prince

a7 430 OIL A8 GINZOHE

)

i

-

QISTRIBUTIOR OF TRIS BOTUNENT (S UNLIMITED

 

. {
S e 7

X/ -

o/l \’.'k
* [ " Ly
= _ .- IS

o oy
&

o -

OAK RIDGE NATI

OPERATED BY UNION CARBIDE CORPORATION e FOR THE U.S. ATOMIC ENERGY (OMMISSION
 

 

This report was prepared as an account of work sponsored by the United
States Government, Neither the United States nor the United States Atomic
Energy Commission, nor any of their employees, nor any of their contractors,
subcontractors, or their employees, makes any warranty, express or implied, or
assumes any legal liability or responsibility for the accuracy, completeness or
usefulness of any information, apparatus, product or process disclosed, or
represents that its use would not infringe privately owned rights,

 

 
ORNL~TM~3963

Contract No. W=7405-eng-26

REACTOR DIVISION

7FRO-POWER EXPERIMENTS WITH 2*2°U IN THE MSRE

J. R. Engel B. E. Prince

DECEMBER 1972

NOTICE

This report was prepared as an account of work
sponsored by the United States Government. Neither
the United States nor the United States Atomic Energy
Commission, nor any of their employees, nor any of
their contractors, subcontractors, or their employees,
makes any warranty, express or implied, or assumes any
tegal lizbility or responsibility for the accuracy, com-
pleteness or usefulness of any information, apparatus,
product or process disclosed, or represents that its use
would not infringe privately owned rights,

 

 

 

ODAK RIDGE NATIONAL LABORATORY
Oak Ridge, Tennessee 37830
operated by
UNION CARBIDE CORPORATION

for the
U.S. ATOMIC ENERGY COMMISSION

PARES A e L
IRV O e e
=D koY I JEE R

v o -
. CLi e R e
Wl el ey ) At

= n YT
I oed & Dniaa RO e
iii

TABLE OF CONTENTS

ABSTRACT . v & 4 v o+ & o s o o s o o o o s o o
FOREWORD . 4 4 o o s o 5 o o s o o s o o s o »
INTRODUCTION . ¢ & ¢ + « o o o o o o s s o s s
TEST PROGRAM . . v 4 ¢« o ¢ ¢« v ¢ o s+ o o s+ o
System Preparation . . . + « ¢« & ¢ & o o
Chronology of TeSts . « o & o o« ¢ o« o o« &

Test Procedures . . . + ¢« s o o o & o & o
Critical Experimemt . . . . . . . .
Control-Rod Calibratiom . . . . . .

Uranium Concentration Coefficient of

Temperature Coefficient of Reactivity

Effective Delayed=Neutron Fraction .
Dynamic Characteristics . . . . . .
Noise Analysis . . + « + & ¢ ¢ o o

RESULTS AND INTERPRETATIONS . . & & ¢ ¢ « o &

Critical Loading . . . v + ¢ ¢ « o o o« &
Observed « +« « « 2 ¢ o o o « s o« o =
Predicted - & a » - 2 ® - - @ . - ®

Variations in Predictions ., . . . .
Neutron Multiplication in Drain Tank . .
Control-Rod Calibrations . « « + + « .« &

Data Analysis o+ 4o v « o o o o s s

Results . & o o o o s o o o o o o o
Concentration Coefficient of Reactivity .
Temperature Coefficient of Reactivity . .
Effective Delayed Neutron Fraction . . .
Dynamics TeStS . &+ « ¢ ¢ o o o o o ¢ o
Noise Analysis . . &« 4« + o o o « o s o« &

CONCLUSIONS . & & s o s o ¢ o s o o s o o o

Reactivity

vi

W W ~ W W =

12
13
13
14
14
15
16
16
16
20
24
31
33
33
41
46
46
48
48
52
53
ZERO-POWER EXPERIMENTS WITH 223U IN THE MSRE

J. R. Engel B, E, Prince

ABSTRACT

Zero-power nuclear tests were performed in the MSRE with
U as the principal fissile isotope in October and November
1968, The initial critical loading was 1.9 + 1% lower than the
corresponding calculated value while the measured control-rod
worths, the fuel concentration coefficient, and the temperature
coefficient of reactivity were all within 77 of their predicted
values, Dynamics tests indicated good nuclear stability and
demonstrated the adequacy of the predictions and testing tech-
niques. Neutron noise associated with circulating gas bubbles
prevented measurement of the effective delayed neutron fraction
with circulating fuel. Uncertainties in the precise condition
of the reactor, as a consequence of prior operation with 235y
fuel, and a strong dependence of the neutronics calculations on
the treatment of neutron leakage prevented any refinement of
the basic nuclear data for *°°U from the experimental results.

233

Keywords: #MSRE + #*Nuclear Analysis + *Uranium-233 + *Criticality +
*Testing + #*Startup + *Experience + Neutron Physics +
Control rods + Reactivity
vi

FOREWORD

The zero~-power physics experiments with ?®?U fuel in the Molten
Salt Reactor Experiment (MSRE) were performed within a two-month period
from September to November, 1968. Prior to that time a general outline
of the test program had been published and detailed test procedures had
been developed and approved. Because the results of these tests were
required for subsequent operations, the data were promptly analyzed and
reported in a variety of internal memos and semiannual progress reports
of the MSR Program. However, since the analytical techniques and the de-
tails of their application to this reactor had been published previously,

233y zero-power tests, no single, comprehensive re-

in connection with the
port of the 2°°U tests was prepared. After the successful conclusion of
MSRE operation, it became apparent that a summary report would have some
value in bringing together the results of the individual tests and in

identifying the techniques and procedures that were applied. The report

that follows provides such a summary,
INTRODUCTION

The MSRE was first made critical with 2°°U fuel in June, 1965 and
that event was followed by a month-long series of zero-power tests. Be-
cause this was the first operation of this unique reactor and special
analytical techniques were applied to some of the data, a comprehensive
report® of those experiments was subsequently published. The reactor was
operated with the 23®°U fuel at power levels up to 7.5 MW through March,
1968, accumulating some 9000 equivalent full-power hours (EFPH). During
that time a decision was reached to extend the operation of the reactor
to include a period with *®3U as the primary fissile isotope. Accordingly
the salt charges were processed on-site by fluoride volatility in August,

235-238y pixture as UFg (ref. 2). During

1968, to remove and recover the
the next two months 35 kg of 233U fuel as UF,~LiF eutectic was added to
the fuel carrier salt. The reactor was subsequently operated until De-
cember, 1969, accumulating another 4000 EFPH. Comprehensive summaries

of the overall operating experience with the MSRE have been published
elsewhere. (See, for example, refs. 3 and 4.)

The ?°3°U loading in the MSRE was the first use of this isotope in a
molten-salt reactor and the first critical loading in any reactor capable
of producing substantial nuclear power. Thus there was considerable in-
terest in comparing the observed critical loading and reactivity coef-
ficients with predicted values. Because of the nature of the reactor and
the uncertainties resulting from the residual fission products and other
effects of prior operation, however, the experiments could not be expected
to yield more precise values for the nuclear characteristics of 2*°3U,
Nevertheless, the data were carefully analyzed and published internally

because of their direct applicability to the analysis of subsequent power

operations.

 

'B, E. Prince, et al., Zero-Power Physics Experiments on the Molten-
Salt Reactor Experiment, USAEC Report ORNL-4233, Feb. 1968.

?R. B. Lindauer, Processing of the MSRE Flush and Fuel Salts, USAEC
Report, ORNL-TM-2578, August, 1969.

°P. N. Haubenreich and J. R, Engel, Experience with the Molten Salt
Reactor Experiment, Nucl. Appl. & Tech., 8, 2, pp. 118-136, Feb. 1970.

“M. W, Rosenthal, et al., Recent Progress in Molten-Salt Reactor De-
velopment, Atomic Energy Review, IX, 3, pp. 601—49, (Sept, 1971).
It is the purpose of this report to present, under one cover, a
reasonably complete picture of the zero-power tests that were performed
with 233U in the MSRE. To this end, the report begins with a section
that outlines the test program and briefly describes the procedures that
were used. This is followed by a presentation and interpretation of the
results., Since the techniques for performing and analyzing the experi-
ments were essentially the same as those applied to the *°°U tests, ex-
tensive use is made of references to previously published reports for
detailed discussions of these aspects. Finally, some general conclusions

are drawn from the results of the tests.
TEST PROGRAM

The program of zero-power tests that was planned for the *°3%U

operation is described in Ref., 5, along with a general discussion of
plans for subsequent operation of the reactor. The objectives of these
tests were essentially the same as those for the initial critical loading;
i.e., to establish the actual properties of the system and to provide the
data required for proper analysis of power operation. The first step was
to establish the critical uranium concentration under the simplest at-
tainable conditions — with the fuel salt static in an isothermal core and
all three control rods withdrawn to their upper limits. This was followed
by more additions of fissile material for control-rod calibration and to
provide the excess reactivity required for operation at power. Measure-
ments of the isothermal temperature coefficient of reactivity, the fuel
concentration coefficient, and the dynamic properties of the system were

also performed.
System Preparation

235
h

The prior operation of the MSRE wit U and the condition of the

233U that was used in the reactor both imposed special considerations on

233y tests. These led to differences in detail

the performance of the
between the **°U and the *°°U tests.

The first step in preparing for the critical experiment was to re-
move as much of the original uranium mixture as possible. This was ac=~
complished by contacting first the flush salt and then the fuel salt with
fluorine in a facility especially designed and installed for that purpose
at the reactor site.® The entire charge of flush salt, containing some
6~1/2 kg of uranium, was transferred to the processing tank where it was

fluorinated for 6.6 hr to reduce its uranium concentration to 7 ppm. (The

UF¢ product that was produced was collected on NaF traps with an activity

 

®J. R. Engel, MSRE Design and Operations Report, Part XI~-A, Test Program
for 223U Operation, USAEC Report, ORNL~TM-2304, Sept., 1968,

°R. B. Lindauer, MSRE Design and Operations Report, Part VII, Fuel
Handling and Processing Plant, USAEC Report, ORNL-TM-907 revised,
Dec. 28, 1967.
decontamination factor near 10°.) The salt was then treated with hy-
drogen and zirconium metal to reduce to the metallic state the corrosion
products produced in fluorinationo* These were filtered’ out of the salt
as it was returned to the reactor system. The fuel salt was then treated
as a single batch with 47 hrs of fluorination required to remove 217 kg
of uranium.

In addition to removing uranium from the MSRE salts, the processing
operations also removed some of the fission products. For example Mo, I,
Te, and Sb would be completely removed, and Nb and Ru would be partly re-
moved in the fluorination step. Some of the more noble metal fission
products might have been precipitated in the reduction step but these had
probably already been plated out on reactor surfaces before the salt was
processed., One important chemical species that was practically unaffected

by the processing was the plutonium that was produced by neutron absorptions

238 235

in U during the U operation. Within the limits of analytical ac-
curacy, all of the 589 gm of Pu calculated® to have been present remained
with the salt,

After recovery of the uranium, a portion of the fuel salt was left
in the processing tank for use in a salt distillation experiment,® (Dis-
tillation has been considered as a means of separating fission products
from carrier salts in breeder reactors.) This provision, along with the
volume reduction produced by the uranium removal, would have left the in-
ventories of both the flush and fuel-carrier salts below their desired
levels. Therefore, before the processing operations were started, two

increments of salt were added to the fuel and flush tanks to compensate

for the anticipated changes.

 

%
This reduction step also reduced the residual uranium to metal.

’R. B. Lindauer and C. K. McGlothlan, Design, Construction and Test-
ing of a Large Molten Salt Filter, USAEC Report, ORNL-TM-2478, March 1969.

8MSR Program Semiannu. Progr. Rep., Aug. 31, 1969, ORNL-4449, pp. 98—101.

®J., R. Hightower, Jr., et al., Low-Pressure Distillation of a Portion
of the Fuel Carrier Salt from the Molten Salt Reactor Experiment, ORNL-
4577, Aug. 1971.
Although the processing operation was very effective in removing

235=2387 nixture was

uranium from the salt that was treated, some of the
left in the system. The physical configuration of the piping at the MSRE
was such that about 17 liters of salt was left in the drain tank when a
batch was transferred out for processing. Since the last material to be
processed was fuel salt, this heel contained significant amount of uranium.
(The flush-salt heel was mixed in with the fuel salt for the final trans-
fer.) This uranium residue was measured before the *3?U critical experi-
ment was begun.

The operating plans with 2°3U included a study of the ratio of cap-
tures to fissions in that nuclide in a typical MSR spectrum. This study
depended upon precise isotopic assays of the uranium as a function of
burnup and a particular initial isotopic composition was desirable for
optimum results. The compositions and anticipated amounts of the residual

238

heel and the new fuel charge were such that additional U was required.

Therefore, a separate addition of about 0.9 kg of *?°U was made before the

233

start of the U loading. This addition was also used as an isotope di-

lution measurement to evaluate the amount and isotopic composition of the
235-23%7 heel mentiomned above. (In addition, it provided an opportunity
to check out the equipment to be used in loading the %2°U.)

The high level of fission-product radiocactivity that remained in the
fuel carrier salt after the uranium recovery made it highly desirable to
keep the reactor cell closed and shielded during the 2?2°U critical experi-
ment. Consequently, the provisions in the reactor thermal shield for ex-
tra neutron detectors were inaccessible and only the normal reactor instru-
ments could be used to follow the approach to critical and the subsequent
experiments. The external neutron source, also in the thermal shield, was
equally inaccessible, so the critical experiment excluded any measurements
based on source movement. The external source was completely overshadowed
by the very intense (a-n) source inherent'® in the fuel salt so that move-
ment of the external source probably would not have been detectable. In

addition, there was a substantial (y-n) source in the fuel from fission-

product decay gammas.

 

‘°R, C. Steffy, Jr., Inherent Neutron Source in MSRE with Clean
233y, USAEC Report, ORNL-TM-2685, Aug. 10, 1969,
The *3°U feed material to be used in the MSRE was made available as
UO0s containing 39 kg U of the isotopic composition shown in Table 1.
This material was converted at ORNL'' to a eutectic mixture of LiF-UF,
and loaded into containers suitable for use at the reactor site. The
presence of 220 ppm of 232" .04 the fact that the material had last been
purified some 4 years prior to its use in this application made the oxide
a strong source of gamma and high-energy alpha radiation. Conversion to
the eutectic fluoride mixture then produced a strong neutron source from
(¢-n) reactions with Be, Li, and F. Because of these radiation sources,
preparation of the feed material had to be carried out in heavily shielded
equipment, as described in detail in Ref. 9. In addition, all subsequent

operations with the enriching salt required heavy shielding.

Table 1

Isotopic Composition of ?°°U Feed Material

 

Abundance

U Isotope (atom %)

232 0.022

233 91.49

234 7.6

235 0.7

236 0.05

238 0.14

 

117 M. Chandler and S. E. Bolt, Preparation of Enriching Salt
’LiF-2?3°UF, for Refueling the Molten Salt Reactor, USAEC Report, ORNL-
4371, March 1969.

*Uranium-232 decays with a 74-year half-life through 4 successive
short-lived, a—emitting daughters to 2*?Pb. Subsequent B decays of **Pb,
212p; and 2°°T1, the latter formed by a-decay of *'?Bi, produce the gamma
radiation.
23%y loading, most of the enriching salt re-

As in the case of the
quired for criticality was to be added to the fuel carrier salt in the
drain tanks. However, the original procedure — melting the enriching
salt in large furnaces atop the drain-tank cell and transfer to the drain
tanks in liquid form — could not be used because of the shielding re-
quirements. Instead, the special arrangement of the reactor portable
maintenance shield®* and the core-graphite sampling shield shown in Fig. 1

was set up to permit the insertion of cans containing up to 7 kg of

uranium as the solid eutectic directly into the drain tank.

Chronology of Tests

 

233U critical experiment was begun on Sept. 10,

Fuel loading for the
1968, Over the next 11 days, about 33 kg of uranium was added through the
drain tank equipment, and one 95-g capsule was added through the sampler-
enricher on the fuel pump to recheck that system. By that time the uranium
loading was within 1/2 kg of the critical value and plans called for making
all subsequent additions through the pump bowl. The progress of the ex-
periment was then interrupted to permit removal of the special equipment
from the drain-tank cell and closure and testing of the containment.
Uranium loading was resumed on Oct. 2 and initial ecriticality with *33%U
was attained on that date after the addition of 3 more capsules of uranium.
The schedule of uranium additions (23 more capsules) and various zero-power
tests continued for several weeks, ending with a reactor drain on Nov. 28.

Although the physics tests were the major activity during this period,
the operation was not devoted exclusively to those tests. Shortly after
the start of salt circulation with *23U fuel, a change was experienced in

the amount of undissolved cover gas in the primary loop and this was in-

vestigated extensively.*® Various aspects of the fuel chemistry were also

 

*2R. Blumberg and E, C. Hise, MSRE Design and Operations Report, Part
X, Maintenance Equipment and Procedures, USAEC Report, ORNL-TM-910, June 1968.

*®J. R. Engel, P. N. Haubenreich, and A. Houtzeel, Spray, Mist, Bubbles,
and Foam in the Molten Salt Reactor Experiment, USAEC Report, ORNL-TM-3027,
June 1970.
ORNL-DWG 68-967R

  
  
  
 
 
 
    
 
    
 

il —~-—— ENRICHING SALT
1 TRANSPORT CASK

~ CORE GRAPHITE SAMPLING
/ SHIELD

PURGE GAS
SUPPLY -~

 

,f~TOOL EXTEN&ON -
SEALS

" porTABLE
MAINTENANCE SHELO Y
N

  
  

TURNTABLE AND .~ - HIGH EFFICIENCY FILTER

STORAGE WELLS —4

 

 

- EXHAUST BLOWER

CONTAINMENT ENCLOSURE ;
AND STANDPIPE ASSEMBLY = = -

[ R

| t - -FDT ACCESS FLANGE

|l . — FUEL DRAIN TANK
i (FDT)

Fig. 1. Arrangement for Adding *°°U Enriching Salt to
Fuel Drain Tank.
studied and changes were made in the oxidation-reduction potential of the

4 An unscheduled drain of the fuel loop part way through the se-

salt.
quence of tests necessitated some minor adjustments in the planned loading
program. Just before the conclusion of this phase of the operation, the
reactor power was raised briefly to 1.2 and 5.5 MW to check out the heat

rejection system,

Test Procedures

The general procedures to be followed for the various zero-power
tests were outlined in Ref. 5. In addition detailed descriptive pro-
cedures with step-by-step check lists were supplied to the reactor oper-
ators for each operation or measurement., (The analysis group responsible
for evaluating the data were on hand for consultation and to oversee the
performance of the tests, but the actual manipulations were performed by
the operating staff under the immediate direction of the operating shift

supervisors.)

Critical Experiment

 

233y required for initial criti-

As was indicated earlier most of the
cality was added to the fuel drain tanks using the special equipment shown
in Fig., 1. This equipment permitted the transfer of single cans of en-
riching salt (containing no more than 7 kg of U) from the shielded trans-
port cask in which they were delivered to the reactor site into the drain
tank under shielded, controlled-ventilation conditions. After the en-
riching salt had been melted out, each empty can was stored on a turntable
within the equipment for removal as a group at the end of the drain-~tank
loading operations.

The nuclear reactivity of a drain tank containing ?°?U fuel was some-
what higher than the same drain.tank containing the *?°U-*°°U mixture.

Although calculations had indicated that the drain tank would be far sub-

critical under all normal storage conditions, careful observations were

 

4R, E. Thoma, Chemical Aspects of MSRE Operations, USAEC Report,
ORNL-4658, Dec. 1971.
10

made during the fuel additions to ensure that criticality was not ap-
proached in the tank. To accomplish this, two neutron-sensitive cham-
bers — a sensitive BFs; chamber and a less-sensitive fission chamber to
cover a wide range of counting rates — were installed just outside the
drain tank for the loading operations, Since both (a,n) and (y-n) sources
were present in the fuel, no external source was required for neutron
monitoring.

The intense internal neutron source made counting rates with the ex~
ternal source and no fuel unreliable as a baseline in this critical ex=-
periment. Therefore, the counting rates measured during the first loop
fill with uranium-bearing salt were used as the first points on the usual
inverse-count-rate plots. Thus two loadings of predetermined size were
required before extrapolations could be made to establish the size of
subsequent loadings. The first two rounds of additions consisted of 21
and 7 kg U, respectively, and the subsequent additions were based on ex~-
trapolations of count-rate data obtained from the preceding additions with
the salt in the reactor. The objective was to bring the uranium loading
to within 1/2 kg of critical in this manner. The enriching salt was
available in cans of various sizes so that arbitrary amounts could be
added in 1/2-kg increments.

The initial charging operation required the addition of three 7-kg
cans of uranium to the fuel drain tank (FD-2). These cans were delivered
to the reactor site individually, and inserted into the charging equip-
ment. The cans were remotely suspended in the gas space of FD-2 above
the liquid carrier salt, In this position the enriching salt slowly
melted and dripped into the carrier salt below it. During this time the
can was suspended from a weighing device so that progress of the melting
could be followed. At the same time the increase in neutron count rate
was observed as the neutron source and subcritical multiplication in-
creased, After the addition, the empty can was weighed more accurately,
to ensure that it was empty, and stored on the turntable for later disposal.

Prior to the addition of each can of enriching salt, one-half of the
carrier salt was transferred to the adjacent fuel drain tank (FD-1) to
provide room in the top of the tank for suspending the cans without con-

tacting the salt. After the withdrawal of the empty can, the remaining
11

salt was returned to FD-2 for mixing and to provide neutron count-rate
data on the full tank. Extrapolations of ratios of these count rates

were used in conjunction with observations during additions to ensure that
the drain tank remained subcritical. After the addition of the first can
of enriching salt, the transfers to FD-1 also removed some uranium to keep
keff very low during subsequent additionms.

After three cans of enriching salt had been added and cross-mixed
between the drain tanks, the core and primary loop were filled with salt
to obtain the initial set of subcritical count-rate data. Filling of the
reactor vessel (the first component to fill with salt) proceeded in several
steps with count-rate data being collected at each level to verify that
the full vessel was not going to be critical. The control rods were held
in a partly withdrawn position during the fill to allow for the rapid in-
sertion of some negative reactivity and initiation of a fuel drain if
criticality should be attained prematurely. The same procedure was fol-
lowed for two subsequent additions of uranium that were made in the drain
tanks. The second addition was 7 kg of uranium in one can; the third,

5 kg in two cans. Count-rate measurements with the primary loop full
after each addition and extrapolation of the data indicated that after
the third addition, the system loading was within 500 g of the amount re-~
quired for criticality.

The remaining uranium required for criticality was added in 95-gram
increments through the sampler-enricher. (These operations were inter-
rupted for removal of the drain-tank loading equipment and sealing and
testing of the containment cell.) Each capsule was added with the fuel
salt circulating and one control rod partly inserted to prevent criti-
cality. When the contents of a capsule had been thoroughly mixed in, the
fuel pump was stopped and the control rod withdrawn to obtain neutron
counting rates. The series of data points thus obtained was used to pro-
ject the critical loading. After sufficient uranium had been added,
static criticality was established by stepwise withdrawal of the final
control rod and verified by brief operation at various nuclear power

levels up to 5 kW,
12

Control-Rod Calibration

The basic experimental procedures used to calibrate the control rods

2
350 zero-

for the #??°U loading were the same as those employed in the
power tests. In essence, data were collected at a variety of control-rod
configurations as uranium was added to establish the operating concentration.

23%y tests, we anticipated that a major

From the experience in the
source of information would be the measured differential worth of the regu-
lating rod as a function of position in the core with the other two rods
withdrawn to their upper limits. Accordingly, these measurements were made
after each of the 23 uranium enrichments that followed the attainment of
criticality in the static system, When the concentration was high enough
to allow criticality with the salt circulating, measurements were made both
in static salt and with full-flow circulation. Each measurement involved
two separate determinations of the differential worth using the rod-bump,
period technique., In this approach the reactor was first made just criti-
cal (or very slightly supercritical) at about 10 W of nuclear power by
manual adjustment of the regulating rod. The rod was then withdrawn a
short distance and the increase in power was recorded as a function of
time for about a 2-decade rise to determine the stable reactor period.

The reactor power indication from the two compensated ion chambers was
recorded digitally on magnetic tape at precisely 1l/4-sec intervals by the
on-line computer. By manually switching the amplifier gain and recording
the range with the output, a precise, unambiguous linear power record was
obtained. This approach was adopted to eliminate some of the uncertainties
associated with the extraction of period data from strip-chart records of
the output of log-count-rate meters connected to the fission chambers.

2357 tests because the on-line com-

(The latter approach was used in the
puter was not fully operational then.) In addition to the measurements
just described, two series of differential-worth data were collected with
the shim rods inserted various distances into the reactor.

Rod-drop data were collected at two points in the uranium loading
sequence, for the purpose of supplementing the differential-worth data.
In these tests the reactor integrated power (fission chamber counts ac-

cumulated on an electronic scaler) was photographed as a function of time
13

from a few seconds before a rod drop to about 30 seconds thereafter.

233 tests and described

(The technique was the same as that used in the
in Ref, 1.) The negative reactivity associated with the rod drop was
obtained by integrating the reactor kinetics equations and matching the
calculated curves to the observed data. Data were obtained for rods
dropped singly, in pairs, and as a gang of three, with the fuel salt
stationary and circulating.

Additional worth information was obtained from control-rod shadowing
measurements in which the critical position of the regulating rod was re-
corded as it was withdrawn to compensate for insertion of first one and

then both shim rods. These data were collected at 4 points in the fuel

loading sequence.

Uranium Concentration Coefficient of Reactivity

Information relating to the uranium concentration coefficient of re-
activity was accumulated throughout the zero-power tests, The only data
required were the uranium loading of the system and the critical control-
rod configuration; the latter giving the reactivity worth by way of the
rod calibration results. These data were obtained after each fuel en-

richment, both with the salt circulating and stationary.

Temperature Coefficient of Reactivity

The isothermal temperature coefficient of reactivity of the reactor
between 1175 and 1225°F was measured on two occasions before the start of
power operation with 2?°U, (The first measurement was made during the
zero-power tests themselves and the second was made at the beginning of
the next period of reactor operation.) Data were obtained for temperature
variations in both directions within this range. Each time, the tempera-
ture was varied slowly by adjusting the external heaters while critical
control-rod configurations and system temperatures were recorded with the
fuel salt circulating. At several points in each test, fuel circulation
was stopped and additional data were taken with the salt stagnant. These
measurements were subsequently judged unsatisfactory because of the ef-
fects of circulating gas bubbles in the system. A third, better determi-

nation was made later in the program when installation of a variable-
14

frequency power supply for the fuel pump made it possible to operate

the pump at reduced speed so as to circulate void-free salt.

Effective Delayed Neutron Fraction

In the %°°U zero-power tests the loss of delayed neutrons due to fuel
circulation was evaluated from reactivity changes associated with starting
and stopping the fuel pump. That determination was made possible by the

233y operation, there were no circulating voids

fact that, early in the
under normal system conditions. With the ?°°U fuel, a large reactivity

effect caused by circulating voids made this type of measurement imprac-
tical. However, some indirect data were obtained from the evaluation of

the control-rod drop tests.

Dynamic Characteristics

 

The zero-power dynamic characteristics of the MSRE with 223U fuel
were studied in a series of measurements of the neutron flux-to-reactivity

® Some tests were performed in which

frequency response of the system.'
single step or pulse reactivity perturbations were imposed on the reactor
at very low powers (<100 W). However, the most useful studies were those
in which the nuclear reactivity or neutron flux was perturbed in a peri-
odic manner. The periodic signals used were either pseudorandom binary
or pseudorandom ternary sequences.'® These are particular series of
square wave pulses that were chosen because they have particular charac-
teristics which permitted determination of the frequency response over a
wide spectrum with only one test. The frequency range over which we ob-
tained frequency-response results was from about 0.005 to 0.8 rad/sec.
The lower limit was set by the length of one period of the test pattern
and the high-frequency limit was determined by the time width of the

square wave pulse of shortest duration which the standard equipment would

 

'5R, C. Steffy, Jr., Experimental Dynamic Analysis of the MSRE with
293U Fuel, USAEC Report, ORNL-TM-2997, April 1970.

'®R. C., Steffy, Jr., Frequency-Response Testing of the Molten-Salt
Reactor Experiment, USAEC Report, ORNL-TM-2823, March 1970.
15

adequately reproduce, The shortest basic pulse width used in these tests
was 3.0 sec. The frequency range covered by these tests was essentially
the range over which thermal feedback effects are important in power
operation.

The on-line computer (Bunker-Ramo, Model 340) was programmed to
generate the sequences by opening and closing a set of relays. Voltage
was fed through the relays from an analog computer (Electronic Associates,
Inc., Model TR-10). This voltage was used to determine the movement of
the control rods, which were forced either to follow the pseudorandom test
pattern themselves or to cause the flux to follow the test pattern. The
control-rod position and the neutron flux were digitized and recorded
every 0.25 sec on magnetic tape. The data were retrieved from the tape
and stored on punched cards which could then be processed to yield the

frequency-response information.

Noise Analysis

Several sets of neutron-flux noise data were collected during the
zero-power test program with the salt stagnant and circulating. The tests
consisted simply in recording the inherent fluctuations in a neutron-flux
signal from the unperturbed reactor for subsequent analysis. Objectives
of these tests were to supplement the data on the dynamic behavior of the
reactor and to gain some information about the effective delayed-neutron
fraction. For the equivalent measurements in the ?°°U zero-power tests,

a chamber had been installed in one of the thermal-shield thimbles to ob-
tain the maximum possible sensitivity. Since these thimbles were inac-
cessible for the 2?°3U tests, the detector had to be installed in the same

facility that housed the rest of the nuclear instruments.
16

RESULTS AND INTERPRETATIONS

The zero-power tests with ?°°U established the conditions and em~
pirical information required for subsequent power operation of the reac-
tor. Comparison of observed and predicted values showed the general
validity of the calculations for the MSRE, but uncertainties in both the
calculations and the measured quantities forestalled refinement of values
for 23U nuclear characteristics. The experimental results and the sig-
nificance of any comparisons that can be made are discussed in the fol-

lowing sections,

Critical Loading

Observed

233

Most of the U required for reactor criticality was added to the
fuel salt in the drain tanks in three major steps which increased the
uranium inventory by 21, 7, and 5 kg, respectively. After each addition
neutron count-rate data were collected with the salt in the fuel loop to
follow the approach toward criticality. Since the neutron source intro-
duced into the fuel with the enriching salt was a major factor in the
count rates during the early stages of the operation, the first increment
of uranium could not by itself provide any reliable indication of the
neutron multiplication. Consequently, the first two increments of fuel

were specified on the basis of the theoretical calculations and the re-
sultant data were extrapolated to determine the size of the third.
Normally an approach to criticality is monitored by plotting an in-
verse function of observed count rates against the uranium loading. This
procedure was followed and the results from one of the neutron detectors
are shown on Fig. 2 for three different treatments of the data. For each
treatment, a straight line was drawn through the last two points and ex=-
trapolated, first toward a projected critical loading and then backward
to illustrate the deviation of the data from linearity. The solid points

are simply the inverse of the observed counting rates, an approach that
17

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

(x 10-2) ORNL-DWG 7212848
22 T i |
v
‘ O 1/CR
A Mu/CR (x 10~3)
20 |—4 10
\ V Mu/CR (CORRECTED)
o
18 9
16 8
14 N 7
12 6
o
o
3
=
ot
10 5 8
8 \\ 4
6 \\ 3
4 ' \\ 2
2 \\ 1
N
\
0 0
20 22 24 26 28 30 32 34

MASS OF URANIUM ADDED (kg}

Fig. 2., Subcritical Count—Rate Data in MSRE Core During
233y Loading.
18

would be appropriate if the fission-independent neutron source were con-
stant., The strong curvature shows that such a plot would have, in the
early stages, projected a smaller uranium loading than that required for
criticality., While this type of error would be in the safe direction with
regard to premature criticality, it would also increase the number of
steps required to reach the critical loading. The open circles in Fig. 2
resulted from multiplying the observed inverse count rates by the uranium
loading, a procedure that is correct only if the neutron source is pro-
portional to that loading. The slight downward curvature in these points
indicates that, in fact, part of the neutron source was independent of the
uranium loading. It may be noted that these data tend to project a criti-
cal loading somewhat greater than that actually required. The data actu~-

233y critical loading in the MSRE are repre-

ally used to project the
sented by the X's in Fig. 2. (Data from 3 separate detectors were used
with reliance on the most conservative projections.) To obtain these
points a correction was applied to account for that portion of the neutron
source produced by y~n reactions in the fuel salt., These corrections were
derived from information gained during the uranium additions to the drain
tanks and were, therefore, only approximately valid for the reactor ves-
sel, In addition the corrections neglected the effect of the external
source in the thermal shield. Nevertheless, some added conservatism was
gained by using the corrections to project the critical loading. Figure 2
also indicates that, while the above source considerations were important
during the early stages of the loading, their importance decreased rapidly
as criticality was approached.

After the three major uranium additions shown on Fig. 2, one capsule
of enriching salt was added through the sampler-enricher to check out the
associated equipment and procedures. Then the fuel loop was drained to
permit removal of the charging equipment and closure of the containment.
Subsequent uranium additions to attain criticality were then made to the
fuel loop with 0.081 of the fuel-salt inventory remaining in the drain
tank. |

Initial criticality with *3®3U was attained after the addition of 3

more capsules of enriching salt (286 of U) to the primary loop. This
19

produced a loading equivalent to 14.98 grams of U* per liter of salt.
At criticality, the fuel salt was stagnant (the circulating pump had been
off for 33 min), one control rod was inserted a distance equivalent to
about 0.015 of its total worth, and the temperature was about 1202°F. The
core temperature was inferred from the output of thermocouples on the con-
necting piping shortly before circulation of the fuel salt was stopped.
Experience had shown that prolonged stoppages of the fuel pump had almost
no effect on the mixed average temperature of the salt, but no data could
be obtained on the relation of the stagnant core temperature to the mixed
mean., However, deviations of more than a very few degrees are considered
unlikely.

Another minor source of uncertainty in the initial critical condition
is the amount of undisscolved cover gas in the core., Since the circulation

2337 fuel salt was accompanied by a void fraction around 0.5 vol %,

of the
some voids were inevitably present in the core when the pump was stopped.
During the first 20 min after a pump stop some of these voids escaped to
the free liquid surface, as evidenced by a gradual decline and levelling
off of the salt level in the fuel pump. An increase in reactivity during
this time indicated that at least some of the voids were migrating out of
the core. However, subsequent tests showed that: (1) after the salt had
been still for 1-1/2 hr, restarting the pump at low speed removed more gas
from the loop and (2) after natural-convection circulation for 12-~1/2 hr
the core retained a positive pressure coefficient of reactivity. We con~-
cluded, from these observations, that a small, but unknown void fraction
remained in the core when initial criticality was attained.

Since the core calculations were based on stationary salt at 1200°F
. with the control rods at their upper limits and no voids in the core, the
deviations and uncertainties in the observed critical condition are most
conveniently expressed as effects on the critical concentration at that
condition. The actual concentration depends on the volume of salt to
which the uranium was added, as well as on the amount of the additions.

The uranium loading was 33.26 * 0.015 kg and the fuel-salt inventory was

 

%
Uranium of the isotopic composition of the enriching mixture.
20

77.6 + 0.5 ft®, Combining these numbers (with a small correction to ac-
count for the fact that the last three capsules of enriching salt were
added to only part of the salt inventory) leads to a critical concentra-
tion of 15.15 + 0,10 g U per liter of salt. Corrections must be applied
to this figure to account for the excess core temperature (1.7 = 2°F),
control-rod poisoning (0.04 = 0.02% 8k/k), and core voids (0.1 + 0.1 vol %).
The corresponding effects on the uranium concentration are, respectively,
0.006 * 0,007, 0.016 £ 0.008, and 0.018 * 0,018 g/f%. These corrections
and treatment of the uncertainties as independent errors lead to an.ef—
fective critical concentration of 15.11 + 0,10 g U per liter of salt at
the nominal standard conditions. Even if all the uncertainties accumulate
in the same direction, the maximum error is only 0.13 g U/2. The nominal

uncertainty is equivalent to 0.677% in concentration or 0.25% in keff'

Predicted

23 . A
30 ecritical

The original theoretical calculations to predict the
loading were performed in the summer of 1967.%7 A major application of
the results was the specification of the amount of 22U feed material that
had to be prepared for the actual experiment, Although we were aware of

several effects of the 2?3°

U operation that would influence the ecritical
loading with *23U, the reactor was still operating so we could only esti-
mate their effects. These calculations led to a predicted critical con-
centration of 15.82 g/liter (uranium of the isotopic -composition of the
feed material). While this result provided an adequate basis for speci-
fying material requirements, it could not be regarded as a "best effort"
at predicting the critical concentration for the conditions that actually
prevailed at the time of the experiment.

Some of the factors that affected the critical uranium concentration
were not fully evaluated until after the critical experiment had been

completed. When tese data became available, we attempted to evaluate

the corrections to the original calculations that we knew were required

 

Y"MSR Program Semiannu. Progr. Rep., Aug. 31, 1967, ORNL=-4191, p. 54,
21

to make the conditions for the computational model conform to the actual
conditions at the time of the experiment. The most important of the neces-

sary changes and updating were as follows:

1. In the original calculations, we had assumed that the operation with
2357 would be terminated at approximately 60,000 MWhr exposure., At
the time of shutdown from MSRE run 14, the actual exposure was
72,440 MWhr.

2, In connection with item 1, the residues of plutonium and samarium re~-
maining in the fuel salt had to be reevaluated.

3. For the earlier calculations the concentrations of the most important
neutron-poisoning impurities, ®Li in the salt and *°B in the graphite,
were assumed equal to the concentrations at the start of operation
with ?®°U., Hence corrections had to be introduced for the burnout
of these nuclides,

4, The calculations had been based on the assumption that no residue, or
heel, or uranium would remain in the reactor system after the transfer
and processing of the fuel salt. From isotopic dilution measurements
made in the course of the uranium additions, we concluded that 1.935
kg of uranium (isotopic assay: 32.97% 2°°U, 66.23% *3°U) from the
first loading had been left in the drain tanks. This became a part
of the fuel salt when the uranium-free salt was returned from the
storage tank for the start of the 2?3U operation.

5. A special addition of 0.89 kg of depleted uranium was made in order

to obtain the isotopic abundances necessary for the planned experiment

o

to measure the capture~to-absorption cross-section ratio.in

This addition had not been considered in the earlier calculations.
6. Certain changes had been made in the evaluated library of cross-

section data for the fissile isotopes since the earlier calculations

were performed.

The above modifications were introduced into the theoretical model,
and the associated changes in the multiplication constant, as well as the
changes in uranium required to compensate for the multiplication changes,
were evaluated. The results are summarized in Table 2. It is apparent

that the largest single correction is that associated with the uranium
22

 

 

Table 2. Changes in Theoretical Calculation of ?®3U Critical Loading
Reactivity Effect Uranium Equivalenta
Correction (% Ak/k) (% AU/U)
Isotopic changes at 72,440 MWhr
239p, +0.095 —0.243
1%p +0,379 —0.973
2495 + 3igp’® +0. 061 —~0.156
°Li +0.217 —0.558
Uranium heel
22%y +0.554 —1.424
238y —0.147 +0.376
Depleted uranium addition —0.102 +0.262
Cross—-section changes +0,212 —0.545
Net +1.269 —3.261

 

“ranium with isotopic composition of enriching salt (91% *°°3u).

Assumes no removal during chemical processing.

Net change is positive since effect of dilution by the drain
tank salt heel was not accounted for in original calculations.
23

heel, but all the corrections are significant. Application of these cor-
rections to the originally predicted value led to a revised prediction of
15.30 g/liter at the reference conditions. This value differs from the
nominal observed loading (15.11 g/&) by 1.25% in concentration or 0.46%
in keff'
In addition to the effects listed in Table 2, there are two other
reactivity effects that were not explicitly included in either the origi-
nal calculation or the revised prediction. One of these is the negative
reactivity contribution of the low-cross-section fission products remain-
ing from the first fuel loading. The magnitude of this reactivity effect
depends on the extent to which various fission products left the salt during

the 235

U operation and also on any separation which may have occurred during
fluorination to remove the uranium. The effect could be as large as —0.5%
Sk/k if all these fission products remained in the salt.

The second reactivity effect is that associated with the distortion
of the moderator graphite due to neutron irradiation. Results of a de-
tailed analysis of this phenomenon in the MSRE'® indicate that the reac-

233

tivity effect at the time of the U loading was probably between —0.04
and +0.07% 8k/k. The spread in probable values results from variations
in the observed distortion of individual specimens of grade CGB graphite
(the material in the MSRE core) as a function of neutron exposure. Al-
though the variations are small compared with the total dimensional change
which occurs at high exposures, they represent a substantial fraction of
the change associated with the neutron exposures experienced in the MSRE.
If median values for these two effects and uncertainties to cover
their possible extreme ranges were included in the calculated loading,
the predicted concentration would be 15.40 £ 0.11 g/%. The difference
between this value and the observed concentration is 1.9% + 1.0Z, where
the uncertainty includes the effects of uncertainties in both the calcu-

lated and observed quantities. The equivalent effect on keff is 0.7 &
Oe4%.

 

'8MSR Program Semiannu. Progr. Rep, Aug. 31, 1969, ORNL-4449, p. 5.
24

Variations in Predictions

Since the 223U loading in the MSRE was the first high-temperature
critical experiment with that fuel, there was considerable interest in
its outcome and some hope that the results might contribute to an improve-

233y, However, the precise analyses

ment in the basic nuclear data for
required for such improvement are severely restricted when, as in the case
of the MSRE, the reactor is not designed as a nuclear physics experiment.
In addition to uncertainties in the measured critical loading and in the
parameters used to define the reactor conditions, significant variations
could be produced with small changes in the calculational procedures or

in the cross-section data.

Several computations were performed in an effort to elucidate the
effects of cross-section data and of calculational techniques on the
criticality predictions. Although some of the computations were done for
the original (*°°U) fuel loading, many of the results of this study are
also applicable to the situation with %3°U,

During the actual nuclear operation of the MSRE, the standard compu-
tational tools employed for neutronics analysis were the GAM-II (ref. 19)
and THERMOS (ref. 20) multigroup spectrum averaging programs, together
with the EXTERMINATOR-2, multi-group diffusion program.?' Subsequently,
another generation of programs designed to perform these analyses and
oriented for use of the IBM 360/75 and 91 computers became available,
These are the XSDRN (ref. 22) and CITATION (ref. 23) programs currently

being employed in MSBR studies. The former program performs the combined

 

'®*G, D. Joanou and J. S. Dudek, GAM-IT — A By Code for the Calcula-
tion of Fast-Neutron Spectra and Associated Multigroup Constunts, GA-4265
(Sept. 1963).

2°H. C., Honeck, THERMOS — A Thermalization Transport Theory Code for
Reactor Lattice Calculations, BNL-5826 (Sept. 1961).

*'T. B. Fowler et al., EXTERMINATOR-2: A Fortran IV Code for Solving

Multigroup Neutron Diffusion Equations in Two Dimensions, ORNL-4078
(April 1967). '

?2N. M. Greene and C. W. Craven, JR., XSDRN: A Discrete Ordinates
Spectral Averaging Code, ORNL-TM-2500 (July 1969).

*°T, B. Fowler and D. R. Vondy, Nuclear Reactor Core Analysis Code:
CITATION, ORNL-TM-2496 (July 1969).
25

spectrum-averaging function of the GAM-THERMOS sequence, and the latter
replaces the EXTERMINATOR-2 code. This section describes the comparative
analyses of the MSRE neutronics using various combinations of these
programs.,

The particular approximation of the MSRE core chosen for this set of
calculations was a seven-zone model in cylindrical (R-Z) geometry with
azimuthal symmetry. The compositions of the salt corresponded to the
minimum critical loadings of 22°U and ?°°U at 1200°F with the control rods
withdrawn to their upper limits. For a given fuel loading, neither the
composition nor the geometry was varied in the comparative calculations
summarized below.

The simplifications of the core geometry required the cluster of
three control rods and sample holder to be represented as an annular
ring about the core axis. While this undoubtedly tends to make the re-
lation between absolute calculations of the multiplication factor and
critical experiment observations more indirect, the purpose of a com-
parative study of computing methods, such as described below, could just
as well be served by taking advantage of the economy of performing the
group—diffusion calculations in two-dimensional geometry.

Table 3 summarizes the various combinations of input data and com-
puter programs used in this study and compares the results in terms of
relative changes in multiplication factor and absolute spectrum—-averaged
capture-to absorption ratios for the fissile components. These two
quantities can be used as simple figures of merit for comparing the vari-
ous cases, Since both are functionals of the distribution of neutron flux

in energy and position over the reactor core. Cases 1, taken from earlier

235 d 233

calculations for the U an U loadings, were arbitrarily chosen as
reference values for comparing the changes in multiplication factor. The
details of the comparison between cases 1 and 2 for the *°°U loading have
been described elsewhere.®® The differences are due to the use of revised
cross-section data. In case 3 we have used the GAM-THERMOS-generated

broad-group cross sections as input for a CITATION calculation. In this

 

2“MSR Program Semiannu. Progr, Rep., Aug. 31, 1969, ORNL-4449,
PP. 22—26.
 

 

Table 3. Comparison of Computation Models for MSRE Multiplication Factors and Capture-to-Absorption Ratios
Ratio of Ratio of
Multigroup Extrapolation Radiative Radiative
MSRE Approximate Spectrum- Few-Group Number of Number of Distance at Cantures Captures
Fissile Case No. Date Averaging Diffusion Broad Slowing- Broad Thermal Reactor Vessel Akeff in **°U to in *??0 to
Loading Program Program Down Groups Groups Boundary Absorptions Absorptions
. 2358 - 233
in U in U
223y 1 November 1968 GAM-THERMOS® EXTERMINATOR-2 3 1 0 0 0.2067 (0.1107)b
2 July 1969 GAM-THERMOS® EXTERMINATOR-2 3 1 0 —0.0053 0.1977 (0.1111)
3 November 1969 GAM-THERMOS® CITATION 3 1 0 —0.0026 0.1977 (0.1111)
4 November 1969 XSDRNd CITATION 3 1 0 +0.0135 0.1991 (0.1128)
5 November 1969 XSDRNd CITATION 3 1 0.71 *er +0.0226 0.1997 (0.1129)
6 November 1969 XSDRNd CITATION 2 4 0.71 Atr +0.0235 0.1998 (0.1130)
233y 1 December 1968 GAM-THERMOS® EXTERMINATOR 3 1 0 0 (0.1858)d 0.1071
2 November 1969 XSDRN CITATION 3 1 0.71 rey +0.0290 (0.1866) 0.1086

 

a . . . .
Cross—-section data libraries for uranium

isotopes based on pre~1965 evaluations.

b . . . . . . . .
Ratios given in parentheses refer to a fissile compeonent present in the calculation either in small concentrations or at infinite
dilution, relative to the primary fissile component.

a , . , . . .
New evaluations of uranium isotope cross-section data {(see discussion in ref. 24).

Cross-section library now used in MSBR studies.

(Data for uranium isotopes same as in case 2.)

9c
27

latter calculation the basic geometric mesh describing the problem wés
constructed to correspond as closely as possible to the EXTERMINATOR
model. (In both cases the mesh consisted of 37 and 56 intervals in the
radial and axial dimensions respectively.) Case 3 indicates that the
variations in the finite differencing and computing schemes used in these
two programs do give rise to a small change in the calculated multipli-
cation factor. However, this change is of no consequence in this appli-
cation.

In case 4 we have introduced the XSDRN multigroup spectrum calcula-
tion, with its associated cross-section library. One finds in this case
a significant change in the multiplication factor (v1.67% Ak, compared with
case 3, which uses the same basic cross-section data for *°°U and ?°°U).
By use of the CITATION perturbation calculation, this difference was shown
to be due almost entirely to increases in the calculated broad-group
transport cross sections for the various nuclide constituents in the re-
actor. The largest increase observed for the graphite-moderated region
was about 5%, occurring in the thermal group. This appears to originate
as follows: In this version of the XSDRN library, the scattering cross-
section data for the thermal energy range include only the P, component
(which is equivalent to assuming isotropic neutron scattering in the
laboratory system). In our earlier calculations we had introduced trans-
port corrections for the thermal group in an ad hoc manner using a stand-

25  Thus it is

ard recipe derived from monoenergetic transport theory.
possible that the thermal transport cross sections used in the earlier
calculations may be more reliable in this épplicationn If these wvalues
had been used in the calculation for case 4, the perturbation results
indicate that the difference in multiplication constants between cases 3
and 4 would have been reduced to about 1.17% Ak.

The remaining differences in the calculated transport cross sections
occurred in the slowing~down energy range. Of particular significance were
the changes for the nuclide constituents in the Hastelloy N reactor vessel,

which accounted for about 0.55% Ak. Approximately 60% of this change ap-

pears to be due to reevaluations of the nuclear data for nickel, chromium,

 

2°R, V. Meghreblian and D. K. Holmes, Reactor Analysis, p. 342,
McGraw-Hill, New York, 1960.
28

iron, and molybdenum since the earlier calculations were made. The re-
maining 40% arises from inherent differences in the methods of calculating
transport cross sections in the slowing-down range used in the GAM and
XSDRN codes.

In case 5 of Table 3, we have modified the zero-flux boundary con-
dition on the CITATION-calculated neutron flux at the outer surface of
the reactor vessel, replacing it with the Milne approximation®® of 0.71x
for the vessel composition. The apparent importance of this correction in
calculating the MSRE neutron leakage is reflected by the increase of about
0.9% Ak relative to case 4,

In the final case 6, for the 2?3°

U loading, we examined the effect of
modifying the broad-group energy structure, placing more groups in the
thermalization range. (An examination of the influence of the number of
slowing~down groups on the calculations has been reported earlier,)?’

Most of our calculations have been made for the few-group energy structure
listed in column 2 of Table 4. Here a single broad thermal group was used,
based on an effective upper cutoff energy for thermalization of 0.876 eV.
We modified this structure as shown in columm 3 of Table 4, placing four
"thermal" groups below 1.86 eV (the highest energy to which thermal up-
scattering can occur in the XSDRN model). Although there was a corre-
sponding slight increase in the calculated multiplication factor between
cases 5 and 6 of Table 3, this difference is of no consequence for this
application.

233

For the U fuel loading, only two cases were studied, which corre-

233y loading. No new

spond to the comparison of cases 2 and 5 for the
feature was exhibited by these calculations, although the net difference
in multiplication factor was somewhat larger because of the increased

233

neutron leakage for the U loading.
The cumulative results of Table 3 display the fact that the sensi-

tivities of the calculated effective multiplication factor and the fissile

 

261bid, p. 180,
*"MSR Program Semiannu. Progr. Rep., Jan. 31, 1964, ORNL=3626, p. 54.
29

Table 4. Broad-Group Energy Structure Used in
CITATION Calculations for MSRE

 

Broad Group

 

No. Cases 1-5 (Table 3) Case 6

1 15 MeV—13.71 eV 15 MeV—13,71 eV
2 13.71-3.93 eV 13,71-1,86 eV
3 3,930,876 eV 1.86—0.881 eV
4 0.876—0 eV 0.881—0.180 eV
5 0.180~0.060 eV
6 0.060—0 eV

 
30

capture-to~absorption ratios to certain details of the computation models
are quite different. The particular sensitivity of Akeff to those details
which most strongly affect the leakage calculation is not very surprising,
for viewed as a critical assembly, the MSRE has a large neutron leakage.
(For the 22°U fuel loading, about 36% of the neutrons born within the
graphite-moderated region leak from that region, and 31% of all neutrons
born within all the salt-containing regions leak to the reactor vessel;

for the 2°%°

U loading, these numbers are increased to 44 and 397% respec-
tively,) The general behavior of the spectrum-averaged capture-to-
absorption ratio for ?°°U should be considered on a separate basis,
however. Here the largest change of about 5% occurred between cases 1

and 2 of Table 3 and corresponded to a substantial revision in the basic

235 d 238

cross-section data for U an U, as described in ref. 24. Between
cases 2 and 6 of Table 3, the maximum variation in this ratio is about 1%.
Such differences still leave this quantity within the standard deviation
of direct measurements of the average ratio for ?°°U in the MSRE,?®
Comparison of the relative variations in the capture-to-absorption

ratios with those of Ake also suggests the caution which should be taken

in attempting to correlaii differences between calculated and measured
multiplication factors from critical experiments, with any particular
changes in nuclear cross—section data (such as those for the fissile nu-
clides), unless the critical assembly has been designed to minimize or
factor out neutron leakage effects, For example, if the changes in multi-
plication factor were due only to the effective changes in capture-to-
absorption ratios listed in Table 3, the differences in Akeff between
cases 2 through 6 would be much smaller in magnitude and would, in fact,
be opposite in sign. Consequently, we must refrain from drawing any con=-

233

clusions about basic nucleatr data for 2°°U or U from the results of

these critical experiments.

 

28MSR Program Semiannu. PPng; Rep., Aug, 31, 1969, ORNL-4449,
pp. 70—72.
31

Neutron Multiplication in Drain Tank

 

Although it was not an integral part of the core critical experi-
ment, the neutron counting rate in the drain tanks was monitored through-
out the 223U loading operations. Calculations had indicated that the
drain~tank neutron multiplication would be higher with *®°U than with *3°y,
but that the tanks would still be safely subcritical with the anticipated
loading.2?® Thus, the purpose of this monitoring was to demonstrate that
criticality was not being approached in the drain tank as fuel additions
were made,

As was the case in the core critical experiment, two neutron sources
had to considered: the fission-product photoneutron source which was es-
sentially constant throughout the loading, and the (a,n) source which in=-
creased with uranium concentration. In fact, the photoneutron source was
more important in the drain tank than in the core because the effective
concentrations of the photoneutron producers were higher. (The core con-
tained 77.5 vol % graphite while the drain tank was almost all salt.)
Thus, neither the simple inverse count rates, nor the ratios of uranium
mass to count rate could be expected to produce data from which reliable
extrapolations toward criticality could be made. We, therefore, again
resorted to compensating the observed mass-to-count-rate values for the
constant photoneutron source. The results are shown graphically in Fig. 3
for one of the two detectors, along with the uncompensated values and the
simple inverse count rates. These data show, even more clearly than the
core values, the advantage of improved treatment of the source term, All
three curves indicate that the drain tank was well below critical at the
final loading, but the indication appears earlier and is more obvious with
the compensated values. While the results can not be translated into
values for keff in the drain tank, they do, at least qualitatively support

the conclusions reached in the calculations.

 

29P, N, Haubenreich, et al., MSRE Design and Operations Report,
Part V-A, Safety Analysis of Operation with 22°U, ORNL-TM-2111, Feb.
1968.
32

ORNL—-DWG 72--12847

(x 1073 (x 1072)

] 9
C\\Qtfn(cona)

 

 

 

 

6 \\/CR \\ 6
._\\ . \8\ °
A
//

 

1/CR
M/CR

 

 

 

 

/ \v

 

 

 

 

 

 

 

 

 

 

 

 

 

0 4 8 12 16 20 24 28 32 36

MASS OF URANIUM ADDED TO FD-2 (kg)

Fig. 3. Subcritical Count-Rate Data in Fuel Drain Tank
During 2°?U Loading.
33

Control-Rod Calibrations

Recalibration of the control rods was necessary for operation with
2337, because of the increased importance of neutron leakage effects and
the more "thermalized" energy spectrum relative to the ?°°U loading. The
primary purpose of this calibration, as before, was to provide the data
required to accurately evaluate control-rod poisoning as a function of

rod configuration for inclusion in the on-line reactivity balance.®°

Data Analvsis

235

For the earlier,“””U, loading we had performed both period—differential-
worth experiments and rod-drop—integral-worth experiments on the regulating
rod.? The former measurement determined the slope of the reactivity vs

rod position curve at a fixed uranium concentration and an initially criti-
cal condition. By adding excess uranium and varying the initial critical
rod positions for these experiments, we were able to calibrate the rod

over its entire length of travel. In the rod-drop experiments, we de-
termined the total negative reactivity corresponding to the scram of the
rod from its initial critical position at some specified uranium concen-
tration. However, because the rod-drop experiment required a more elabo-
rate recording technique and analysis, it was convenient to perform it

only a few times during the course of the uranium enrichments., Hence, in

233y, we used the results of integrated

the earlier experiments with
period—differential-worth data as the basis of our final evaluation of
the rod worths. The drop experiments were used mainly to cross check the
integrated period data, as well as to test the technique of the rod-drop
experiment itself. These results were then combined with rod-shadowing
data to produce the general description of comntrol-rod poisoning for all

control-rod configurations.

 

®%p. E. Prince, J. R. Engel, and C. H, Gabbard, Reactivity Balance
Calculations and Long-Term Reactivity Behavior with ?*°U in the MSRE,
ORNL-4674, Feb. 1972,
34

When we attempted to apply this same general approach to the *3?°U

calibration experiments, we encountered some obstacles, however. In
brief, after the data from the differential-worth measurements were col-
lected and analyzed, the precision of the results proved to be too poor
for application in the manner described above. Although the exact reasons
for all of the scatter were not identified, increased uncertainty in the
control-rod position, as a result of wear and aging in the assemblies
probably was a major contributor. (Difficulties in positioning the rods
precisely also affected the dynamics tests.) Since the rod travel re-
quired to establish a given stable reactor period was much smaller with
2337 than with 2®°U, this uncertainty also had a greater effect on these
results. Another potential source of scatter is minor variations in the
core void fraction during individual period rumns. In several cases abrupt
changes in the "stable" period were observed, but the cause could not be
identified. The occasions on which the period decreased could have been
caused by an escape of gas bubbles from the core.

Because of the poor precision of the differential-worth experiments,
it became necessary to rely more heavily on the results of the rod-drop
measurements to obtain the information required to calibrate the rods.

In the course of the zero-power experiments with ?°U, we had performed
two sets of drop experiments, one near the start of the excess uranium
additions and one when the critical position of the regulating rod was
near the position of maximum differential worth. The results of analysis
of these experiments are described below. Unlike the period—differential-
worth measurement, the reactivity determined from the rod-drop experiment
is quite insensitive to small errors in the initial and scram positions

of the rods, The principal new problem we had to consider was how to in-
terpolate the information from these two sets of experiments to other

d 2?30 concentration to ob-

values of the initial critical rod position an
tain the shape of the rod worth curves.,

The first set of experiments was performed after the addition of cap-
sule No. 23, This corresponded to very nearly 30.78 kg of added uranium
contained in the fuel loop, or 33.50 kg of total uranium added to the salt
(including the drain tank residue). At these conditions the critical po~-

sition of the regulating rod at 1200°F was approximately 41.8 in. with the
35

pump off and 47.3 in. with the pump on. In the ensuing description the
regulating rod will be designated as rod 1 and the shim rods as rods 2
and 3.

As described in ref. 1, in these experiments the integral count is
measured starting with the reactor critical a few seconds before the scram
and ending about 30 sec after the scram. The fission chamber is positioned
so that the initial count rate at the critical condition is about 30,000
counts/sec. The attempt was made to start each experiment at a neutron
level equivalent to about 50 W. Since it was impractical, however, to
try to obtain exactly the same count rate at the time of rod scram in each
experiment, for convenience in analysis we first renormalized the measured
counts for each case to correspond to an initial rate of 3 X 10“ counts/sec.
This should not introduce any error since the system is linear in this range.

The results of the measurements and analyses for the first set of ex-
periments (designated as series A) are shown in the bottom curve in Fig. 4
and in all curves in Fig. 5, Figure 4, bottom curve, corresponds to the
scram of rod 1 from its initial critical position of 41.8 in, with the
pump off, The circled points are the normalized integral count data, and
the curve is the result of numerically integrating the reactor kinetics
equations in the manner described in ref. 1. The magnitude of the nega-
tive reactivity inserted is used as an adjustable parameter to obtain a
close fit to the measured count data. The average accelerations of rods 1
and 2 used for these analyses were 150 and 190 in./sec?®, calculated from
average drop-time measurements made subsequent to termination of operation
with ?3°U. 1In addition, all analyses were normalized to the values of the
delayed neutron fractions given in Table 5. These are effective values
based on actual yields®' with approximate evaluations of (1) the relative
abundance of ?2°U and 2??Pu in the fuel salt (source of data for measured
delay fractions for all fissile isotopes was ref. 5), (2) the relative

energy effectiveness of delayed neutrons in the reactor spectrum,®? and

 

21G. R. Keepin, Physics of Nuclear Kinetics, p. 90, Addison-Wesley,
Reading, Mass., 1965.

22MSR Program Semiannu. Progr. Rep., Aug. 31, 1968, ORNL-4344,
PpP. 45—46.
36

ORNL-DWG 69-5477

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

S <
g | © DATA FOR FUEL PUMP OFF,| _ | ] ,,.‘/
SERIES A | .V;
ol e DATAFORFUELPUMPOFF,| . | | '/fl_o_o11b o
SERIES B | L SERIES B
6 |_.— INTEGRATION OF REACTOR | _ _ __4-° , 31.80 kg OF ADDED |
= KINETICS EQUATIONS / URANIUM IN FUEL
5 - LOOP
8 5 T /..,/Hv [ — [
-
2 s
a1 et - =-0.0235 -
O 7T P
K o — SERIES A
Z | Rtd R | | 30.78kg OF ADDED |
e T URANIUM IN FUEL
L — LOOP
2 e e 4 - . b e —— b ]
f =
1 L j i ___l,_ _
0 a |

 

 

 

 

 

 

 

 

 

 

 

 

 

 

0 2 4 6 8 10 12 19 16 18 20 22 24 25 28 30
TIME AFTER SCRAM (sec)

Fig. 4. Results of Rod-Drop Experiments for Regulating Rod (Rod 1).

ORNL-DWG 69-5478

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

|
S| SERIES A:3078kq OF ADDED ' - T T
URANIUM IN FUEL LOOP .
5 © FUEL PUMP OFF e | -
z & FUEL CIRCULATING p=-00250 i
3 — INTEGRATION OF REACTOR |
© 4P~ KINETICS EQUATIONS // 5=-0.0260 |
a ;
o 5 |
g 3 ffif p=-0.0250 — I : . A
z - = = |2 | ° p=-00260 '
¢ e fi A 2 5 |
A .
%’fl : :
1 - 3 ' ;
/ T
o | 1 |

0 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30
TIME AFTER SCRAM {sec)

Fig. 5. Results of Rod-Drop Experiments for Shim Rod 2.
37

(3) calculated importance-weighted reductions in delay fractions for the
conditions of steady fuel circulation.®® The background for treating the

effects of fuel circulation is discussed in detail in ref. 34.

Table 5. Effective Delayed Neutron Fractions for
the **°*U Loading Used in Analysis of
Rod~-Calibration Experiments

 

 

Stationary Fuel Circulating Fuel
Group (neutrons per 10° (neutrons per 10°
fission neutrons) fission neutrons)
1 2.376 1.137
2 8.576 4,188
3 7.190 4.370
4 8.214 6.654
5 1.579 1.544
6 1.003 0.998
Total 28.938 18.891

 

The data points for the top curves in Fig. 5 correspond to the nor-
malized integral counts for shim rod 2, scrammed from its upper limit of
travel. The pump was off, rod 3 remained set at 51 in., and rod 1 re-
mained at 41.35 in., its initial critical position for this experiment.
This case is of particular interest because, except for the slight '"shad~
owing' effect caused by the 9.65-in. insertion of the regulating rod, it

is a direct measurement of the total worth of a single shim rod. (In

 

33MSR Program Semiann. Progr. Rep., Aug. 31 ,1967, ORNL-4191, pp.58-60.

°%B. E. Prince, Period Measurements on the Molten Salt Reactor Ex-
periment During Fuel Circulation: Theory and Experiment, ORNL-TM-1626,
Oct, 1966.
38

addition, the shim and regulating rods were found to have the same poi-
soning effect by comparison of their critical positions,)

In order to better illustrate the sensitivity of this measurement
to the magnitude of the reactivity insertion, in Fig. 5 we have shown two
calculated curves corresponding to —2.,50 and —2.60 % 8k/k. These curves
closely bracket the count data, and we can assign a reactivity insertion
of =2.55 *+ 0.05% Sk/k as the "measured" reactivity. Therefore the rela-
tive sensitivity or fractional uncertainty in this method should be better
than 27.

The effect of the slight shadowing perturbation caused by rod 1 during
the first 10 in. of fall of rod 2 can be estimated from results of rod-
shadowing measurements, to be described later in this section. This cor-
rection is quite small and should increase the measured reactivity magni~
tude by about 1% of the nominal value., Thus we can assign 2.58 * 0.05%
Sk/k as the total worth of a single rod at the concentration of *°°U ob-
taining in this first set of experiments.

The bottom curves of Fig. 5 represent the data and results of analy-
sis for the same shim rod, taken with the fuel circulating. As in the
experiment with the pump off, rod 2 was scrammed from its upper limit of
travel. Rod 3 remained set at its upper limit, and rod 1 remained at its
critical position with the pump running, 47.32 in. The solid curves are
the results of integrating the kinetics equations for the same reactivity
magnitudes shown in the top curves of Fig. 5, but with the delayed neutron
fractions for each precursor group replaced by effective fractions calcu-
lated with the theoretical model used previously for MSRE analysis (Table
5). These results suggest that the calculated losses in delayed neutron
fraction due to circulation may be slightly underestimated by the theo-
retical model. The observed difference is rather small in comparison with
the total effect, however, and we concluded that the calculated delayed~-
neutron losses are within 157 of the actual losses in the MSRE. (Note
that the presence of any entrained gas in the circulating fuel salt would
not be significant in interpreting this experiment, since these conditions
would have remained essentially constant throughout the experiment.)

The second set of experiments (series B) was performed after 31.80 kg

of uranium had been added to the fuel loop. The top curve in Fig. 4 shows
39

the result of dropping rod 1 from its new critical position at this
loading (22.25 in. with the pump stopped). The negative reactivity ob-
tained from the analysis of this experiment was —1.10% 8k/k. As will be
shown below, the results in Fig. 4, together with the total measured worth
of one rod, provide information sufficient for normalizing and interpo-
lating the rod-calibration measurements.

Another curve of special interest in the control-rod calibration is
the variation of excess reactivity with the uranium concentrationo* Each
point on this curve corresponds to the excess reactivity of the system if
the rod were withdrawn.from its critical position to the upper limit of
travel, at a fixed uranium loading., This information is directly appli-
cable to the on-line reactivity balance calculations, since the uranium
is depleted and readded in the course of operation. The curve is also of
theoretical interest, however, because the uranium additions without com-
pensating rod insertions represent a uniform variation in nuclear proper-
ties over the volume of the fuel salt. Of the characteristics calculated
in core physics studies, this effect should be one of the most reliable,
whereas quantities depending directly on the calculation of the rod poi~-
soning effect are probably the least reliable. Therefore for this prob-
lem we have made use of theoretical results concerning the shape of the
excess~uraniumreactivity curve to aid in interpolating the experimental
data described above,

By means of theoretical derivations and arguments based on pertur-
bation theory,+ one can arrive at the following conclusions:

1. The variation in the excess reactivity p with 2®3%U loading for a
fixed rod position is closely approximated by the formula

= K(C—=Co) _ K(M — M)
p = o - M » (l)

 

 

%

Much of the discussion which follows is also pertinent to the evalu-
ation of the uranium concentration coefficient of reactivity, but because
of its direct relevance tc the rod calibration, it is presented here.

+These aspects are discussed in detail in ref. 30.
40

where X is a constant, C and M are the concentration and mass of ?3°U in
the salt, and subscript 0 refers to the values at the minimum critical
loading.

2, If a reactivity measurement is made which involves the motion
of the control rod away from its initial critical position at some fixed
uranium loading and we wish to interpolate this measurement to another
uranium concentration, the interpolation factor is approximately inversely

proportional to the uranium concentration:

¢

Ap(Co) -~ ¢
R (0) - Co (2)

We first tested the accuracy of approximation (1) by applying it in
the case where an experimental curve was attainable from independent meas-

urements. This was the 23°

U mass-vs-reactivity data determined by inte-
gration of period measurements and reported as Fig. 7 of ref. 1. We found
that fitting the data by a theoretical curve of the form (1) required only
a very slight change from the "empirical" curve reported in ref. 1.
Approximations (1) and (2) provide all the necessary relationships to
interpolate the rod calibration data for the *°3?U loading. The value of
Mo, the loading of the fuel loop at the reference conditions for the ini-
tial critical experiment, was 30.59 kg of uranium with isotopic composition
of the enriching salt. (Note: The heel of *°°U and ??°Pu remaining in the

salt from the 23°

U loading constitutes a base-line effect in all these
measurements and can be shown to have negligible influence on the analysis
following.) The approximations were applied as follows. From Eq. (2) the
total reactivity effect of one rod at a loading of 31.80 kg of uranium is

approximately
2.58(30.78/31.80) = 2.497% Sk/k .

Subtracting the rod-drop reactivity from the total rod effect gives the

excess uranium reactivity at 31.80 kg of uranium:

excess U reactivity = 2.497 —1.10 = 11,3977 Sk/k .
41

This result was then applied to generate the reactivity curve corre-

sponding to Eq. (1),

_ oM (0.01397)(31.80) _
K = ¥ —=m = ~31.80 =30.59 - U-369.

The curve of Eq. (1) based on these values of the parameters X and M, is
shown in Fig. 6.

These results may be cross-checked for consistency with the rod-drop
measurement given by the bottom curve in Fig. 4. In this case the reac-
tivity effect of dropping the regulating rod from its critical position

at 30.78 kg of uranium was 2.35% 6k/k. Then

excess U reactivity = 2,58 —2.35 = 0,237 8k/k .

This is observed to be in close agreement with the value obtained from the

interpolated curve in Fig. 6.

Results

To synthesize the results of the preceding sections, we used the
simple procedure of combining (1) experimental measurements of the vari-
ation in critical rod position with uranium loading and (2) the information
given in Fig. 6 to obtain the reactivity effect corresponding to arbitrary
insertions of the regulating rod. The result is the calibration curve for
one control rod with the other two rods withdrawn to their upper limits.
These data are shown as the open circles along the lowest curve in Fig. 7.
(The small amount of statistical variation in these data probably corre-
sponds to slight uncertainties in the temperature and the rod position.)
To facilitate the use of this information in the on-line reactivity bal-
ance calculation, we used Eq. (2) to put all the numerical values of re-
activity in Fig. 7 on the basis of one uranium concentration, arbitrarily
chosen as the initial critical concentration. Figure 7 also presents a
summary of the complete control-rod calibration including the effects of
partial insertions of the shim rods. These additional results were ob-

tained by the following procedure. During the course of the uranium
MAGNITUDE OF REACTIVITY POISONING EFFECT {% B4/k)

Fig.

42

ORNL-DWG 69-5479
2.4 ‘ |

 

 

20

 

EXCESS REACTIVITY (% 8k/k)

 

 

 

 

 

 

i |

 

030.5 340 315 32.0 32.5 330
MASS OF URANIUM ADDED TO FUEL LOGP
(kg of uranium ; isotopic composition N4 % 233)

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

- - - - 2 »

Fig. 6. Effect of Uranium Mass on Reactivity in 37 Loading;
Theoretical Curve, Normalized to Results from Rod-Drop
Experiments.

58 ORNL-DWG 69-5480

. I I T ]

SHIM RODS 2 AND 3 v
] | AT 34 in. — /r o /
! «"J
2.4 ( l; — + — ,bv S —
| | V\Q 'bg
o ¥ #
° 4 0
; | 2 o w>
i @ N O
20 b—— t e b —eeee b e -& L - - e~ ot —
' & a Q2 |0 <
i 1 & -\ > 4;p .
! | (<) 1. O
i .I\Q ‘\\\“ ()c’ ‘;0 »
| i‘ o ° | o 2 &
16— b e @ — St e PN
1 O—\\ N % &
é‘ \:\\
~ < o
el — e & — A - —
. quv 6‘\\
o
Q |
O
0.8 e e —— e} —_— .
|
_ N K AR S N B SN S SR
! i L
°‘ | L
40 36 32 28 24 20 16 12 8 4 0

 

POSITION OF ROD MOVED {(in. withdrawn}

Results of Fit of Rod Calibration Data by Least Squares Formula.
o Data obtained from measurement of critical rod positions and
use of Fig. 6. A Data interpolated from rod-shadowing measure-
ments (Fig. 8). Reactivity effects normalized to initial
critical loading of 30.59 kg of uranium in fuel loop.
43

additions, we performed several rod-shadowing experiments, in which the
change in critical position of the regulating rod was measured as one shim
rod and then two banked shim rods were inserted. Results of measurements
carried out at four different uranium concentrations are plotted in Fig. 8.
Here the ordinate and abscissa are the shim~ and regulating-rod positions,
and each plotted point represents a critical configuration of the rods,
The particular conditions for each experiment are also indicated in the
figure. A sequence of critical positions taken at a fixed uranium concen-
tration and temperature would represent a contour line of constant reac-
tivity relative to the reference condition if the reactivity were being
plotted in the dimension perpendicular to the plane of Fig. 8.

The data taken from the experiments in which the two shim rods were
moved in a bank was of special interest in application to the on-line
reactivity-balance calculations because this is the configuration used in
the operating reactor., This interest is emphasized in Fig. 8 by the smooth
curves drawn through these data points. The curves are terminated on the
line of equal insertion of the shim and regulating rods because the area
to the right of that line is the region in which the rods were operated
in the MSRE. From the earlier work with the 2*?°U loading, we developed
an analytical formula for the rod reactivity variation with shim- and
regulating-rod insertion.®® This formula contained adjustable parameters
which were fitted by least~squares analysis to the measured data in the
region of interest. The analysis procedure described in ref, 35 was ap~-

233U experiments, with the resulting

plied without modification to the
least-squares curve fit shown in Fig. 7. The triangular points in this
figure are check points for the least-squares formula, interpolated from
the smooth curves in Fig. 8. The least-squares formula provided a good
fit to the measured data of Fig. 7 for most of the range of rod travel

but deviated somewhat from expected values near the upper and lower limits
of the range. The total regulating rod reactivity determined from the

formula was too large, and the curves probably needed to "bend" more near

 

*3MSR Program Semiannu. Progr. Rep., Feb. 28, 1966, ORNL-3936,
pp. 82—87.
POSITION OF MOVED SHIM ROD(S) (in. withdrawn)

n

ORNL-DWG 69-5481

 

 

 

 

 

8 1 1 i ! T T 1
p e ROD 3 FIXED AT 51 in., ROD 1 COMPENSATING
. FOR INSERTION OF ROD 2

12 . a RODS 2 AND 3 AT EQUAL ELEVATIONS, ROD { ]

. COMPENSATING FOR INSERTION OF BOTH

. . SHIM RODS
16 —
® o
e °
20
| ] | J
9 L J

24 =% A . . ° /| e

a A ®
28 -

 

 

32 * \
36 [ T . \ .

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

48 \* ° ,\
EXPIA EXPIB EXIPC EXP D

52

 

52 44 36 28 20 12 4
CRITICAL POSITION OF ROD { {in. withdrawn)

Conditions:
After Capsule i ) Reactivity Effect
EXp.  Addition No.  Fuel Circulating (% Bk/k)*
A 9 Yes 0.32
B 13 No 1.10
C 23 No 2.05
D 27 Yes 2.52

Fig. 8. Change in Critical Position of Rod 1 as Rods 2 and 3 are
Inserted into Core.
45
these limits. (This same situation was encountered earlier in the 2%°3°y
measurements.) These deviations did not constitute a problem, however,
for normal operating procedures kept the tips of the rods well away from
these extremes.
Estimates of the reactivity worth of the control-rods with *°3U fuel
had been made®’ along with the early calculations of the critical loading.,

2953 to predict the

These estimates were used in a variety of analyses
characteristics of the reactor system. Although it was recognized that
the control-rod worths were subject to some uncertainty, the experience

with the *3°

U loading suggested that the differences between projected
and actual values would not substantially affect the conclusions of these
studies. In addition, it was anticipated that empirical rod-worth data
would be used for power operation of the reactor. Thus, there was little
incentive to produce '"best possible'" estimates of control-rod worth,
Nevertheless, a comparison of the measured rod worths with the values

that were calculated, Table 6, shows that acceptable agreement was obtained.

Table 6. Comparison of Observed and Predicted
Control-Rod Worths in MSRE with *°°U Fuel

 

Value (% Sk/k)

 

 

Configuration Observed Predicted
1 rod 2,58 2.75
3 rods 6.9 7.01

 

 

*¢0, W. Burke and F, H. S. Clark, Analyses of Transients in the MSRE
System with 2°°U Fuel, ORNL~4397, June 1969,
46

Concentration Coefficient of Reactivity.

As we indicated earlier, much of the information relative to the
uranium concentration coefficient of reactivity was developed in the evalu-
ation of the control-rod calibrations. The extension of that information
to the concentration coefficient proceeds very simply. Differentiation of
Eq. (1), p. 39, shows that XK is in fact equal to the concentration coef-

ficient at the minimum critical loading; that is,

dp - -
(¢ dCJC—C = K = 0.369 . (3)
—v“0

(Extension of this process to higher uranium concentrations with the aid
of Fig. 6 yields a value of 0.355 at the operating concentration.) The
value obtained from theoretical calculations'? for the critical loading

was 0.389 which compares quite favorably with the observed quantity.

Temperature Coefficient of Reactivity

 

Although only one was made during the zero power tests, all three
temperature coefficient measurements that were made in the early part of
the ??°U operation are discussed here in order to permit presentation of
the final results. Each measurement was based on variation of the iso-
thermal system temperature between 1175 and 1225°F.

The first measurements were made near the end of the zero-power tests
when the uranium concentration was approximately at the value for power
operation. Data on critical rod position and temperatures were logged on
the computer and manually, both with the pump running and after it had
been off for 15 min, at each of nine temperatures. Temperature coeffici-
ents were computed by plotting the rod reactivity effects (as determined
from the empirical calibration) against system temperature. The points
with the pump off fell very close to a straight line with a slope of
—~7.75 X 10™® (8k/k)/°F., The points with the fuel circulating were more
scattered but gave a distinctly different slope: —6.9 x 10™° (8k/k)/°F.
47

The difference was not unexpected, since we knew there was an appreciable
amount of entrained cover gas circulating with the salt, and past experi-
ence’® had demonstrated that the circulating void fraction increased with
decreasing temperature. Thus, the fluid density change due to the depen-
dence of gas entrainment on temperature would tend to offset the salt
density change with temperature, reducing the magnitude of the reactivity
effect of slowly varying temperature, But even the value with the pump
off was significantly less than that predicted on the basis of no gas in
the salt. Therefore the series of measurements was repeated later with
great care, The results with the pump off were practically the same as
before, but with circulation the measured value was —7.4 X 10=° (8k/k)°F.
The inference was that the variation of the void effect (bubble fraction)
with temperature had changed.

After the variable-~frequency power supply was put in service and the
loop bubble fraction was observed to be very low at reduced fuel circu-
lation rates, the temperature-coefficient measurement was repeated. With
the pump running at reduced speed and practically no bubbles circulating
with the salt, the data gave a value of —8.5 X 107° (8k/k)/°F. The higher
value suggests that the effects of gas had not been completely eliminated
in the earlier experiments by simply stopping the pump. This was in
agreement with other observations of the gas behavior discussed earlier
in this report.

The predicted value of isothermal temperature coefficient of reac-
tivity was —8.8 X 10™° (8k/k)/°F at operating fuel concentration (=9.4 X
10~° at critical concentration). The only directly comparable observed
value is the one measured with no circulating gas, which was 3.47% smaller.

The individual contributions of the fuel and the graphite to the
overall temperature coefficient were not measured experimentally because
of the uncertainties introduced by the circulating gas bubbles. Instead,
the calculated values were used on the basis that the close agreement in
the overall coefficient could be expected to extend to the separate

components.
48

Effective Delaved Neutron Fraction

Neutron noise measurements were made du ing the zero-power tests in
an effort to evaluate experimentally the effective delayed neutron frac-

37  The noise from two identical

tion, Be, with 222U fuel in this reactor.
neutron-~sensitive ionization chambers was reduced to a frequency spectrum
using a digital cross-correlation method that eliminates unwanted noise.
The resulting f ®quency spectrum of the reactor cor rlated noise was
least-squares fitted to a point-~kinetics model of the MSRE to evaluate

the ratio of Be to the prompt neutron generation time A, The calculated
value of )X was then used to infer Be. The value of Be obtained from the
data for noncirculating fuel was 0.0031 + 0.0001, which compares favorably
with the calculated value of 0.0029 (see Table 5). Data were also taken
with the fuel circulating but these did not yield a value of Be, because
neutron noise caused by helium voids in the circulating fuel masked the
effect of Be' Thus, the only experimental support for the calculated de-
layed neutron fraction with circulating fuel was the indirect observation

from the rod-drop experiments (p. 38).

Dznamics Tests

The results of the zero~power dynamics tests have been reported else-
where.'® 1In general the results were found to agree favorably with theo-
retical predictions.38 The step and pulse reactivity tests, however, con-
tributed little to this conclusion., Since the signal strength in these
aperiodic tests is distributed over the entire frequency range, there
was insufficient signal at any one frequency to rise above the very high
noise level associated with the high void fraction in the ?3°U-fueled

system. Thus the results contained too much scatter to be useful.

 

®*7’D. N. Fry, Instrumentation and Controls Div., Annu. Progr. Rep.,
September 1, 1869, ORNL-4459, p. 100.

°%R, C. Steffy, Jr., and P. J. Wood, Theoretical Dynamic Analysis
of the MSRE with *°3U Fuel, ORNL-TM-2571, July 1969,
49

The frequency response tests in which periodic perturbations were
introduced were analyzed by several different methods. Although the re-
sults of these analyses were not always the same, the agreement with the
predicted response was acceptable. Figures 9 and 10 show the experimentally
determined zero-power frequency response by two analytical methods in com-
parison with the predicted response. In both of these tests, one with
stationary and one with circulating fuel, the reactivity was perturbed by
manipulating a control rod and the response of the neutron flux was meas-
ured.

With the fuel stationary, the frequency response of the zero-power
MSRE was essentially the same as that of any stationary-fuel, zero-power,
233U~fueled reactor. The measured frequency response with the fuel not
circulating is shown in Fig. 9. The magnitude ratio, &n/No*dk, is seen
to be in general agreement with the theory, but the phase angle is not
in particularly good agreement,, At the higher frequencies, the magnitude
ratio and the phase angle were lower than the theoretical at other power
levels as well as here. This is thought to have been caused by the con-
trol rod not adequately following the test pattern yet giving the indi~-
cation that it was. The indicators, which are physically located with
the drive assembly, accurately display the action of the rod-drive motors;
however, the flexibility of the control rod makes it doubtful that the
tip of the rod, which is about 17 ft from the drive assembly, reproduces
the high-frequency component of the rod-drive movement.

The results of a zero-power test with the fuel circulating are shown
in Fig. 10. The shape of the magnitude ratio curve is in excellent agree-
ment with the theoretical curve, but the results have been normalized by
multiplying each experimental value by 1.75. The phase angle data was in
better agreement with the theoretical predictions than was the case for
the non-circulating data, but there is scatter in the results. The need
to normalize some results and not to normalize others is also considered
to be caused by poor control rod indication. The normalization was sub-
sequently shown not to be power-~dependent since some data did and some
did not need normalization at each power level, and the normalization

factors, when they were required, were different for different tests.
50

ORNL -DWG 69-12050

FUEL STATIONARY
S POWER LEVEL - ZERO

ANALYSIS METHODS
« CPSD

2 s FOURCO ENSEMBLE
— THEOQRY

 

~N

10

 

O

 

 

 

|
W
o

 

 

 

 

 

 

 

 

 

PHASE (deg)

|
D
&)

 

 

 

 

 

 

 

 

 

 

 

 

 

-90

o3 o2 10°

 

FREQUENCY (rad/sec)

Fig. 9. Neutron Flux-to-Reactivity Frequency Response of the
233_pyeled MSRE at Zero-Power with Stationary Fuel.
31

4 ORNL-DWG 69~ 12044

POWER LEVEL -ZERO
ANALYSIS METHODS

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

o FCQURCO
« CPSD
-~ THEQRY
’ |
!
S |
Q ’30 i Lt
© ‘ . o’
= J i-wl.l-a '-43 ) /50 K
L 1T ece % %gs .j ‘."fifl
% 60 ' ! Mi P Dt *
_ | 1 nan L
90 -2 - O
10> 2 5 107% 2 5 10 2 5 10

FREQUENCY ({rad /sec)

Fig. 10. Neutron Flux~to-Reactivity Frequency Response of the
223y-Fueled MSRE at Zero-Power with Circulating Fuel.
52

The zero-power results shown here were supplemented by other tests
at higher powers which did much to verify the satisfactory dynamic char-

acteristics of the system,

Noise Analysis

The nominal purposes for examining the neutron flux noise at zero-
power were to complement the dynamics studies and to evaluate the effect
of fuel circulation on the delayed neutron fraction. Useful data had been

235

obtained in both of these areas':>® during the U zero-power test pro-

233 tests, the neutron detection efficiency was

gram. However, in the
reduced by a factor of n15 by the need to locate the detector in the in-
strument penetration rather than in the thermal shield and the neutron
noise contribution from circulating gas bubbles was much higher. Con-
sequently, no useful information could be extracted from the data in these
two particular areas., On the other hand, evaluations of the neutron-flux
noise had been shown to be of substantial value in studying the behavior

of circulating voids in the system“® and these techniques were employed

throughout the operation with *°°U,

 

3°D. N. Fry and D. P. Roux, Results of Neutron Fluctuation Measure-
ments Made During the MSRE Zero-Power Experiment, Unpublished internal
memo, ORNL-CF-65-10-18, Oct. 29, 1965.

“°D, N, Fry, R. C. Kryter, and J. C. Robinson, Measurement of Helium
Void Fraction in the MSRE Fuel Salt Using Neutron Noise Analysis,
ORNL-TM~2315, Aug. 27, 1968.
53

CONCLUSIONS

The primary purpose of the zero-power tests with 222U fuel in the
MSRE was to establish the neutronic characteristics required for subse-
quent analysis of the operation of the reactor at power. That this was
successfully accomplished is demonstrated by the fact that there was no
suggestion of anomalous nuclear behavior while the system was operated
for 4000 EFPH with this fuel. Some changes from the behavior of the sys-
tem with ?®°U fuel were observed (particularly with regard to circulating
voids) but these were not due to unexplained differences in the nuclear
behavior.

In general, the agreement between the observed and calculated nuclear
properties was sufficiently close that the analyses and conclusions that
were based on calculated values remained valid. Table 7 presents a com-
parison of the basic results of the zero-power tests with the corresponding
predicted quantities., The conclusions that may be drawn from these com-
parisons and the foregoing discussions are summarized as follows.

233
h

1. The critical loading wit U was predicted less accurately

than was the initial loading with 2®°U. Two possible reasons are recog-
nized: (1) the reactor conditions were less well defined for the 233°U
case because of the prior operating history with *2°U, and (2) the calcu-
lation for 22°U was more sensitive to uncertainties in the treatment of
neutron leakage processes.

2, The measured control-rod worths were within 7% of the calculated
values; the fuel concentration coefficient was within 6%; and the tempera-
ture coefficient was within 5%. Thus the predicted safety characteristics
of the reactor that were based on the calculated values were valid,

3. Although the higher circulating void fraction with ?2°U compli-
cated the interpretation of some data and prevented some direct measure-
ments (separate fuel and graphite temperature coefficients and loss of
delayed neutrons due to fuel circulation), indirect observations provided
the confirmations necessary for safe operation of the reactor.

4, The dynamics tests at zero-power which were only a small part
of the continuing surveillance of the system's dynamic behavior, satis-

factorily demonstrated the adequacy of the predictions and of the testing

procedures.
54

 

 

 

Table 7. WMSRE Nuclear Characteristics with 22°U
Value
Property
Calculated Observed
Base-line critical concentration,a 15.30 15.11
grams of uranium per liter
Control rod worth, % 8k/k
One rod 2,75 2.58
Three rods 7.01 6.9
e d
Temperature coefficient of reactivity —8.8 x 10°° —8.5 X 10'56
(8k/k)/°F —7.4 x 10-°
Concentration coefficient of reactivity, 0.389 0.369
(8k/k) / (8¢/C)P
Change in Beff due to fuel circulation —1.005 x 10-?

 

“At 1200°F with fuel stationary and all control rods withdrawn to
their upper limits.

Uranium of the isotopic composition of the material added during
the critical experiment (91% 2°°U)

c ’ - » *

Assumes no gas in the fuel and the operating uranium concentration.
The predicted temperature coefficient at the initial critical concen-
tration was —9.4 X 1077,

dWith very little gas.

®With about 0.5 vol % gas circulating.
30.
31,
32,
33.
34,

- - .

»

RPUHAYIOODRDEPIOOPrtooGs oGSO B & -

B o

- » * - - - - °

-

HLO D2 DU g dudCr e wonda g

=
-

Alexander
Anderson
Baes
Bauman
Beall
Bettis
Bohlmann
Briggs
Compere
Cottrell

. Crowley

Culler
DiStefano
Ditto
Eatherly
Engel
Ferguson
Ferris
Fraas

. Gabbard

Grimes
Grindell
Guymon

. Haubenreich

Hoffman
Kasten
Kerlin
Keyes
Krakoviak
Lundin

55

ORNL-TM=-3963

Internal Distribution

35.
36.
37.
38,
39.
40,
41.
42,
43,
44,
45,
46.
47,
48-49,
50,
51,
52.
53.
54,
55.
56,
57.
5€,
59.
60.
61.
62-63.
64,
65-67.
68.

Lyon

MacPherson
MacPherson

McCoy

McCurdy

. McNeese

Meyer

Miller

Moore

Nicholson

Perry

Prince

Ragan

Rosenthal

Dunlap Scott

Myrtleen Sheldon

J. Skinner

L. Smith

Spiewak

A, Sundberg

E. Thoma

Trauger

Weinberg

Weir

White

. Whitman

Central Research Library
Y-12 Document Reference Section
Laboratory Records Department
Laboratory Records (RC)

»

= g:u:?-nini?-m-r*fliHizin:ud
S HEECCCGnE O e

QUG oORROHOR

.UO',TJFUJ
 

69.

70.
71.
72,
73.
74.

75,
76-77.
78-79,.
80-81.
82-98,

99,

.100~101,

56

External Distribution

D. F. Cope, Atomic Energy Commission, RDT Site Office, ORNL,

Oak Ridge, TN 37830

W. H. Hannum, USAEC, Washington, D.C. 20545

Kermit Laughon, AEC, RDT Site Office ORNL, Oak Ridge, TN 37830

J. Neff, USAEC, Washington, D.C. 20545

M. Shaw, USAEC, Washington, D.C, 20545

R. C. Steffy, Tennessee Valley Authority, 540 Market St.,

Chattanooga, TN 37401

M. J. Whitman, USAEC, Washington, D.C. 20545

MSBR Program Manager, USAEC, Washington, D.C. 20545

Director, Division of Reactor Licensing, USAEC, Washington, D.C. 20545
Director, Division of Reactor Standards, USAEC, Washington, D.C. 20545
Manager, Technical Information Center, AEC(for ACRS Members)

Research and Technical Support Division, AEC, ORO

Technical Information Center, AEC -