Control of Nuclear Reactors and Power Plants (M.a. Schultz)
Control of Nuclear Reactors and Power Plants (M.a. Schultz)
M. A. SCHULTZ
Pittsburgh, Pennsylvania
1955
TK
. 2L
United States of America. All rights reserved. This book, or parts thereof,
bomb. The problems they faced were ones of safety and of complete
distrust for refined control elements. The legend is told in the industry
that in the first reactor, constructed under the west stands of the Uni-
nuclear control from this point to where useful power could be safely
handled has been monumental. Now, in this new age, the problem is
simply stated: Given a nuclear power plant, what is the best way of con-
and now a nuclear power plant, a jet engine, or a guided missile is treated
system for a nuclear power plant. The entire field of reactor and power-
plant control is far from covered in this book. Only one specific type of
design of control systems for other types of reactor plants may be obtained
in the field and to ignore the remainder as being too numerous to men-
vi PREFACE
tion. Because of the pioneering efforts of the small and somewhat closed
First, since all the present activity in this country in the field of nuclear
power plants is under the direction of the United States Atomic Energy
Commission, most of the references originally came from the basic work
for the use of some of the illustrations used in this book. Finally, I must
reactors. He, of course, was responsible for the automatic control design
was their initial work on the oscillation of a reactor that gave the servo
system.
M. A. SCHULTZ
CONTENTS
Preface v
CHAPTER 1. INTRODUCTION 1
Scope of Text.
2-4. Reactor Period. 2-5. Reactor State. 2-6. Prompt Critical. 2-7.
Input in Sk. 3-3. Solution of Kinetic Equations for Ramp Function. 3-4.
Sk.
Horsepower Requirements.
vii
viii CONTENTS
sis. 6-8. Coolant Mixing. 6-9. Flow Changes. 6-10. Analysis for Multi-
Instrument Circuits.
trol Systems.
Generated on 2011-10-24 11:37 GMT / Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google
Control.
Problems 303
Index 309
CHAPTER 1
I NTRODU CTION
Each reactor plant that has been built thus far contains a different con-
but many common theoretical problems and basic design concepts have
points.
and nuclear-plant control for the new group of control engineers now
entering this field. Historically, nuclear power plants grew from nuclear
reactors, which in turn grew from basic nuclear physics. The detailed
cerning nuclear reactors, which in some cases may create concern on the
Generated on 2011-10-24 11:38 GMT / Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google
part of the nuclear physicists that they and the control designers are not
transition stage from the physicists to the engineers. The plants that
have been constructed are as complex in their own way as are the basic
physical equations upon which the reactors are founded. The engineer
ing terms which are compatible with the rest of the system. The phys-
1
2 CONTROL OF NUCLEAR REACTORS AND POWER PLANTS
icist in turn has been more concerned with the intricate details of the
The problem of reactor control has existed since the first reactor and
has been the subject of extensive study for many years. The problem
of nuclear plant control is a newer one, and the answers are not as well
known.
Tying a load onto the reactor and extracting power from it would corre-
COOLANT
PUMP CONDENSATE
PUMP
FIG. 1-1. Block diagram of elementary nuclear power plant containing pressurized water
spond to tying a load onto the d-c generator. In the case of the generator
it would easily be anticipated that the load would affect the generator
plant would affect the characteristics of the reactor. This reaction could
devices could find their performances reflected back on the basic reactor
from an over-all point of view, the reactor plant, d-c generator analogy
t Superior numerals in the text correspond to the numbered References at the end
of each chapter.
INTRODUCTION 3
pressure water is used to cool the reactor and extract heat from it. This
plant contains a steam turbine, condenser, and all the necessary auxil-
an electric generator. Both the primary coolant and steam systems are
closed loops.
parameters can be set up for a plant of this sort, depending upon the
between the water temperatures of the primary loop and the steam tem-
specified from either loop. For the purpose of this discussion the plant
operation will be specified from the primary loop in terms of the coolant
expansion
Let us assume that our control program is such that the average tempera-
ture of the primary loop coolant is held constant regardless of the load
used. The flow of water created by the pump is at a fixed rate and does
control are shown in Fig. 1-2, for the arbitrary condition of T.v = 500°F.
It can be seen in this plant that the steam temperatures fall off very
the power output of the reactor is proportional to Th — Tc. The con-
4 CONTROL OF NUCLEAR REACTORS AND POWER PLANTS
stant primary coolant flow is assumed such that, again using arbitrary
let us again assume that at unity power T.v — T, = 60°F, which will be
designated as rated full power. The numerical values of all these tem-
700
800 r
600 700
500 600
o o:
8I
g2
Q. UJ
300 400
<
UJ
uj
200 300
100
200
100
Generated on 2011-10-24 11:38 GMT / Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google
23456
REACTOR POWER
FIG. 1-2. Temperature and pressure control conditions for plant having constant-Tat,
program.
peratures are the ones given in Fig. 1-2. In this elementary plant the
steam leaving the steam generator is of high quality, but it is not super-
dry and saturated steam at all power levels. The steam pressure then
depends only on the steam temperature and may be obtained from steam
tables.
the steam loop, the power delivered to any component becomes simply
F, &H, where F, (Ib/hr) is the steam-flow rate and &H (Btu/lb) is the
enthalpy drop across the component. Then F,, Hw, and Hx can be cal-
INTRODUCTION 5
reactor power. The normalized output power and efficiencies are plotted
in Fig. 1-3. The curves represent fully the over-all steady-state perform-
product / AV. The d-c circuit that is roughly analogous to the nuclear
power plant is shown as Fig. 1-4. This circuit has a source of power, a
senting shaft power. These three components are in series since the
components of the steam loop are in series. The reactor and steam gen-
erator are simulated by the d-c generator. The generated voltage corre-
The battery is used because the power lost in the condenser is more nearly
in Fig. 1-5, and they roughly approximate the plant characteristics shown
in Fig. 1-3.
From the above analogy it can be seen that the steady-state over-all
plant performance of a nuclear power plant is not too dissimilar from that
control and the role of reactor control is apt to be a subservient one. The
design a specific reactor plant, the designer of the control system must have
ophy has been the one in which the control system as well as all other
power plants has been such that if one were to blow up inadvertently,
the resulting publicity would severely harm the entire program of nuclear
safe variety.
occur in a nuclear power plant are not so severe as might first be imagined.
bomb" and a nuclear power plant is one of control—in the first case the
CONTROL OF NUCLEAR REACTORS AND POWER PLANTS
energy is given off instantly, in the second case it is given off slowly. In
the event of a failure of the control system the obvious thought is that the
power plant might become a bomb. Hurwitz2 has indicated that a con-
1200
1000
90
40
30 y
20 i
<
10
23456
REACTOR POWER
TURBINE
POWER
CONDENSER LOSS
losophy, the generator and other expensive items in the circuit must, in
the load, and no great harm is done by taking any one machine off the
120
100
>
Q 60
Ul
a:
UJ
§40
20
50
40
30
o
Generated on 2011-10-24 11:38 GMT / Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google
UJ
20|
10
1 33456
GENERATED POWER
under this philosophy the reactor would be shut down as fast as possible,
not only for the above reason but also because if it were not shut down,
engine cannot be shut off for any external cause or the entire structure
depends upon the specified operating situations of the plant. That these
8 CONTROL OF NUCLEAR REACTORS AND POWER PLANTS
example, the desire to protect the reactor may be so strong that all of the
used.
system is the degree of automatic control desired. Here again the ulti-
mate usage of the plant greatly affects the decision. For example, a
station nuclear power plant, on the grounds that operating costs for labor
over-all safety, the proposition has been advanced for extensive use of
automatic control rather than manual control, on the basis that, although
automatic systems have been known to fail, they can be made to fail
a mistake. The same cannot be said for the human being as a control
element.
the basic conditions that he has to meet. The reactor designer will
specify the amount of heat that can be taken from the reactor and how
much overload for how long a period of time will be permitted. The
pump designer will specify how much cooling fluid can be circulated.
Generated on 2011-10-24 11:38 GMT / Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google
The boiler designer will specify given steam temperatures and pressure
out the plant, which must not be exceeded if excessive corrosions and
The control-system designer must tie all these factors together in indi-
vidual and over-all control loops in such a manner that everyone is sat-
isfied. Then he conceivably has time to ask himself the inevitable ques-
power plants are inherently stable, but it is well known that even the
most stable basic system can be upset by improper detailed control design.
1-6. Scope of Text. The ultimate scope of this book is to lead new
processes of reactor physics.3 This text will first present a review of the
this review, and very little hint as to the complex basic physics problems
sented, and the transfer function of the reactor is derived for future use
ularly interesting and critical element of the reactor control loop is the
of present-day practices. The stage has now been set for a short glance
at over-all plant control, and some of the basic problems of plant response
The details of some of the special reactor components that are not
often exists on the part of any designer to try out his device, Chap. 11
machines the hazards existing with attempts to try out a new device on an
REFERENCES
June, 1953.
2. Hurwitz, H., Jr.: Safeguard Considerations for Nuclear Power Plants, "Proceed-
material is used to slow down fission neutrons to where they possess the
The moderator may also be circulated through the core and serve as the
used to conserve neutrons and "bounce" them back into the core in an
both neutrons and gamma rays. Inside the core or the reflector are
located the control rods, the basic purpose of which is to regulate the
2-2. Fission Process. Inside the core there exists an initial source of
ity of splitting this nucleus into two or more fragments. This process is
process (approximately 200 Mev per fission). In the act of fission two
Generated on 2011-10-24 11:38 GMT / Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google
to three neutrons are released from the fragmentation, and these neutrons
are capable of creating more fissions in other uranium nuclei under the
proper conditions. The two to three neutrons that are produced in fis-
sion may take part in several reactions, all of which are competitive. A
be lost to the core by leaking out of the system. In any event, for a chain
reaction to take place, for each uranium nucleus capturing a neutron and
These fast neutrons have high energies. Inside the reactor these neu-
by the various materials in the system or be lost through escape from it.
Depending upon the relative amounts and nature of the moderator, fuel,
If most of the fissions result from the capture of neutrons which have
thermal reactor. When most of the fission processes are caused by the
about 1,000 ev. If the main source of fissions is the capture of fast
neutrons directly by the fuel without the neutrons having suffered any
energy losses, the system is called a fast reactor. Power reactors in gen-
eral are of thermal and intermediate types; fast reactors are usually used
tion can take place. If k is less than unity, the chain reaction cannot
For the chain reaction in the core to keep going, the production of
Generated on 2011-10-24 11:38 GMT / Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google
neutrons must equal the leakage plus the absorption of the neutrons.
infinite pile, in some texts referred to as &,-n/. The term k — 1 is defined
an infinite pile. The term kei represents the amount the multiplication
differs from unity, and as most reactors operate around unity, kex gives
P=
In this text the symbol Sk will be used for reactivity, meaning the amount
Sk = p = keff, ~ 1 (2-2)
Keff
ELEMENTARY PHYSICS OF REACTOR CONTROL 13
From an engineering point of view these terms are generally used in the
of a general nature k and keff have been used in the past interchangeably,
and k,x and Sk have similarly been loosely interchanged. The cause of
this confusion stems from the fact that for control purposes the internal
structure of the reactor and the details of the multiplication in the core
are not important. The control designer obtains his information from
which elapses from when neutrons are produced in fission until they return
again to fission or are lost to the reaction. The term /* may be con-
sidered as
I* = I (2-3)
one generation over the preceding generation is then Sk. If there are
initially n neutrons per cubic centimeter present in the core, the rate of
succeeding generations,
dn Sk ,0 .,
di = I* n (2-4)
n = noe(«*"*" (2-5)
is the number after a lapse of time t. On this basis the number of neu-
fissions occurring, and for 3 X 1010 fissions per second 1 watt of power is
number of neutrons in the core in any given time interval, and the symbol
level is involved.
2-4. Reactor Period.
defined as
Period =
(l/n)(dn/dt)
(2-6)
with the inverse reactor period (l/n)(dn/dt) being the quantity usually
measured. We can solve for the period of the reactor of Eq. (2-5),
Period = - =
OK
sec
(2-7)
reactor is that amount of time which the reactor would take to change
tion at a fixed power level, the period is infinite. Only when the reactor
all the neutrons created in the fission process were given off instantly
in fission are given off at discrete amounts of time after the actual fission
process occurs. These neutrons which are produced after fissioning rep-
resent approximately 0.75 percent of the total neutrons produced and are
delayed neutrons. Table 2-1 indicates the properties of the delayed neu-
Mean life
ti, sec
Decay constant
Xi, sec"1
0.071
0.62
2.19
6.50
31.7
80.2
14.0
1.61
0.456
0.151
0.0315
0.0124
0.00025
0.00084
0.0024
0.0021
0.0017
0.00026
14
ELEMENTARY PHYSICS OF REACTOR CONTROL
irons which are given off in the uranium 235 fission process by thermal
neutrons.4'6'7 The delayed neutrons are given off in six distinct groups
denote the total fraction of the delayed neutrons with ft being the frac-
tion of the delayed neutrons in the iih group of delayed neutrons. Sim-
ilarly, Xj represents the decay constant of the ith group of delayed neu-
MULTIPLYING MEDIUM
NEUTRON SOURCE
neutron lifetime /.
n = n0(l + k + fc2 +
or in closed form
1 - k"
(2-8)
(2-9)
degenerates into
n_ _ 1
(2-10)
no 1 — fc
CONTROL OF NUCLEAR REACTORS AND POWER PLANTS
This ratio is known as the subcritical multiplication factor, and all reac-
tors exhibit this effect. The result of this equation is shown graphically
were to change k to 0.9 by removing part of the control rod from the
4 8 12 16 20
24
-4 -2 0
change.
factor becomes higher and higher, more time is always taken for the
Generated on 2011-10-24 11:39 GMT / Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google
medium to settle out at a given level. And finally it does not settle out
at all but continues to rise. This situation holds only if there is a source
16
Let us now examine how the power level changes in this subcriticar
removing the control rods. The simplest case is to consider pulling the
rods at an infinitely slow rate, and in this case the total multiplication is
so slow, the decay times of even the longest-lived delayed emitters are
Therefore all the delayed neutrons have ample time to be emitted before
Eq. (2-10). Figure 2-4 shows the neutron-level build-up under this
level is 1.
2-8. Subcritical Period. The period that results from this very slow
pulling rate can be obtained from the definition of period of Eq. (2-6).
Period =
From this expression it can be seen that if the rate of change of k is con-
stant with time, then the period decreases directly as the reactivity
slow rod pulling at a constant time rate of change, the period approaches
inferred that the power level of the medium was constant with k = 1, it
is obvious from Sec. 2-7 that the source neutrons continue to add in and
Generated on 2011-10-24 11:39 GMT / Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google
create a rising power level. From a practical point of view this phenom-
neutrons. The usual reactor source strength may vary from a few neu-
trons per second to possibly a few million neutrons per second. The
reactor behaves when k > I if all the neutrons were prompt. An illus-
trative example of the change of level under this condition may be given.
10~3 sec. This is roughly the value of I* for a large graphite moderated
into the reactor. Equation (2-5) then indicates that at the end of 3 sec
Let us now examine the situation when delayed neutrons are present.
dn Sk ft .
=n~n
and the other symbols have the same meanings as previously described.
d is defined by
f -f»-^ <2-13)
delayed neutrons coming in from the past must be added to make up the
to solve the approximate equations that result when the delayed neutrons
are assumed all to be bunched in one group which has an average value
Let us also assume that Sk is small so that k and /ce// can be used inter-
dn Sk — 0 . . „ ,0 .
^ = -^-n + XC (2-14)
Generated on 2011-10-24 11:39 GMT / Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google
f-£»-XC (2-15)
of the forms
- c-a
—
c = S (2-19)
ELEMENTARY PHYSICS OF REACTOR CONTROL
simply
(2-20)
n(Q)
- Sk
- Sk
n(0)
= 1.67e°-067' - 0.67e-4-6'
(2-21)
A plot of this result is shown in Fig. 2-5. The effect of the delayed neu-
trons is at once apparent. It will be recalled from Eq. (2-5) that with
only prompt neutrons, the power level soared to 8,000 times the original
level in 3 sec when +0.0035& was inserted into the multiplying medium.
3.5
3.0
2.5
2.0
Generated on 2011-10-24 11:39 GMT / Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google
1.5
1-0
0.1
0.2 0.3
10
TIME, SECONDS
FIG. 2-5. Relative neutron level as a function of time for a step reactivity change of 0.003.
From Fig. 2-5 it can be seen that at the end of 3 sec the power level has
risen by a factor of only 2.1. Thus the effect of a mere 0.75 percent of
a simple feasible one rather than a most difficult if not impossible one.
what is called a transient period, whereas the first term creates a so-called
stable reactor period. The stable period of Eq. (2-20) would then be
T=
= 15 sec
(2-22)
- Sk
\Sk
20 CONTROL OF NUCLEAR REACTORS AND POWER PLANTS
for the example just presented as against the prompt neutron period of
Eq. (2-7)
o/c
must now be examined in more detail. It will be recalled that the pro-
duction of neutrons equals the leakage plus absorption for critical opera-
change the leakage one might mechanically put a hole or a window in the
monly used method, particularly for thermal reactors; and control rods
assumed that the control rods mentioned in this book are of the absorptive
rods may be shim rods, whose function is to affect the power level in a
power level.
beTSuiIt into the reactor and then some reactivity removed by inserting
control rods partially into the reactor in order to obtain critical operation.
control rods are completely extracted minus one will be termed excess
when the control rods are completely inserted will be called the shutdown
Let us now examine the method of changing the power level in a simple
reactor. Figure 2-6 illustrates the process. Ajssume that the reactor
level. The first step_is__tp extract a control rod a small amount and
Eq. (2-20)1 As the power level rises and approaches the desired ultimate
seen from Mg. 2-6 that the control rod must oscillate about the k = 1
position, but ultimately it will settle down at the original position from
ELEMENTARY PHYSICS OF REACTOR CONTROL
21
independent of rod position, and in order to change power level one moves
during the lifetime of a reactor because of fuel depletion. That is, as the
uranium in the reactor is used up, the number of fissions occurring will
III)
CRITICAL POSITION
FINAL LEVEL
TIME
Coefficient. Other causes exist for moving control rods, one of them
being the temperature of the reactor. Most reactors have what is termed
heats up, its reactivity is reduced. Reactors which have water or gas
system parameter, but for the present let us examine the basic operation
Generated on 2011-10-24 11:39 GMT / Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google
fixed position. Now, through some external means such as that of heat-
ing the normal coolant of the reactor, let us raise the average tempera-
This process of heating will reduce the reactivity of the reactor. Con-
critical. Control rods must then be extracted to make up for this loss
Cold critical then refers to the position of the control rods when the reac-
tor is critical at room temperature, and hot critical refers to the position
COOLANT OUT
CONTROL
ROD
'out
ToV
Tin
POWER LEVEL
POWER LEVEL
COOLANT IN
CONSTANT
TEMPERATURE
FIG. 2-7. Elementary operation of a reactor having negative temperature coefficient and
of the control rods when the reactor is critical at its normal operating
Generated on 2011-10-24 11:39 GMT / Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google
temperature.
trates this condition. For a simple example assume that the coolant
perature and the coolant out of the reactor is used in heating radiators
and then dumped. If the flow from the faucet is constant, the tempera-
ture versus power level is shown in Fig. 2-7. Here, as the power output
for this rise in average temperature, control rods must be extracted from
then that the control rods are equally effective at all positions in the
22
ELEMENTARY PHYSICS OF REACTOR CONTROL 23
reactor, it can be seen that under this condition the position of the control
tical situation many other types of programming can exist. The position
of the control rods will rarely be either of the two simple functions of
ate, fission products are created from the uranium. Many direct fission
and indirect nuclides may have large cross sections for the absorption of
neutrons, and therefore they can act as poisons. If these poisons are
formed by radioactive decay even after the reactor is shut down, the con-
nuclides are of particular interest, xenon 135 and samarium 149. Xenon
135 is formed as a result of the decay of the direct fission product tel-
Generated on 2011-10-24 11:39 GMT / Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google
manner:
Barium 135 is stable. Xenon 135 has an absorption cross section for
Sm149, on the other hand, is the stable end product of the chain
1.7 hr 47 hr
This reaction occurs in roughly 1.5 percent of the fissions, and the Sm149
5.3 X 104 barns. For these reasons it does not contribute as much to
the poisoning of a reactor as does the Xe136 and consequently the Sm149
becomes
rlX'
any time t
time t
time. Because the half life of Te136 is very short compared with that of
I136 (1 min against 6.7 hr), we can make the simplifying assumption that
where Ti is the fractional yield of I136 from the direct fission process and
It can be seen that two kinetic effects occur which concern reactor
control. The first effect is the so-called equilibrium poisoning, and the
Generated on 2011-10-24 11:39 GMT / Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google
reactor power operation, and the peak poisoning after shutdown of the
reactor from a high power level. From Eq. (2-24), the amount of xenon
present in the reactor builds up from the concentration of the iodine 135
and dies off both by the radioactive decay into cesium and from the
that, after long operation at a given fixed power level, an equilibrium can
exist between the build-up and the two decays, so that at the end of a long
period of time a steady amount of poison can exist in the reactor. Fig-
ure 2-8 shows the time scale to build up to equilibrium for such a process.
The amount of poisoning involved depends upon the steady power level
of the reactor and the design of the particular reactor involved. The
Let us now shut down a thermal reactor as fast as possible from a high
level is reduced to effectively zero, and consequently the decay term due
iodine 135 which has been previously formed. Ultimately the radio-
RELATIVE EQUILIBRIUM
O 0 o O r4
) 10 20 30 40 50 60 70 80 9
TIME, HOURS
FIG. 2-8. Equilibrium xenon poisoning build-up for an enriched thermal reactor
Generated on 2011-10-24 11:39 GMT / Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google
10
40
20 30
FIG. 2-9. Relative peak xenon poisoning reactivity as a function of time after shutdown for
a thermal reactor.
active decay of the xenon 135 to cesium 135 takes over and the total
xenon 135 concentration drops off. The time involved in this process is
depends upon the initial steady power level of the reactor and its specific
design. However, the peak xenon poisoning may be many times the
magnitude of the initial power level and the extent to which the power
Other variables of operation can enter. After a reactor has been shut
down and the xenon poison concentration partially built up, the reactor
may be turned on again and subsequently burn out the xenon back to
the equilibrium condition. This operation can call for a fast rate of
change of reactivity from the control rods to keep the reactor critical
amount of reactivity may be so much that the reactor does not contain
control rods are pulled out all the way. Under these conditions, where
will be impossible to start the reactor until the xenon decays down. As
an example from Fig. 2-9, let us assume that sufficient reactivity exists
from a given power level. The figure indicates that unless a startup is
made within this Y% hr, it may be 40 hr before the reactor can be started
up again.
into the control rods to hold the reactor safely shut down when all rods are
safely shut down differs from reactor to reactor. Approximately — 2 per-
that a given reactor changes its reactivity by 5 percent going from cold
ity of 17 percent is designed into the machine, and it is desired that the
as the amount of negative reactivity in a new cold reactor when all the
rods are fully inserted. For this particular set of conditions it can be
excess reactivity of 17 percent, the reactor will not fully override the
xenon poisoning on the last day of its rated life provided that it is started
ELEMENTARY PHYSICS OF REACTOR CONTROL
27
its life and in the last third it will be necessary to start up the reactor
cold, to operate the reactor in such a manner that the poisons do not build
down reactivity requires that the reactivity value or rod worth of the
rods fully inserted into the cold clean reactor must be at least 26 percent.
Many control schemes can be conceived so that poisons are not per-
ing is usually a startup made quite close to the peak of the xenon poison-
ing. Therefore, on occasion, even the type of working day must be con-
sidered in the design of how much reactivity shall be built into the reactor.
<v
CONTROL
>
%
Generated on 2011-10-24 11:39 GMT / Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google
J_
ALL
IN
ALL
OUT
ALL
OUT
rod position in the reactor. Actually this is not the case. Moving a
mately as the sine squared of the rod position. To make the matter more
complex, the position of each individual rod usually can affect the reac-
incorrect at the end points. It would not be expected that the worth of
a rod would have a zero slope at its end positions. Figure 2-10 indicates
having this type of rod worth. It can be shown that the peak-to-average
rate of reactivity change the reactor goes through critical during a startup
28 CONTROL OF NUCLEAR REACTORS AND POWER PLANTS
positions so that minimum rod worth does not occur when the xenon
inserting reactivity into the reactor at its maximum rate. For this con-
dition the control rods must put in reactivity fast enough to override the
xenon poisoning.
REFERENCES
York, 1954.
4. Hughes, D. J., J. Dabbs, A. Cahn, and D. Hall: Delayed Neutrons from Fission
6. Soodak, H., and E. C. Campbell: "Elementary Pile Theory," John Wiley & Sons,
7. Sun, K. H., et al.: Delayed Neutrons from U23s and Th232 Fission, Phys. Rev., vol.
Generated on 2011-10-24 11:39 GMT / Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google
8. Phillips, H. B.: "Differential Equations," John Wiley & Sons, Inc., New York,
1939.
CHAPTER 3
REACTOR KINETICS
The first two chapters have provided background from which we can
power plants will be discussed. Before the control of reactors and plants
ramp functions, and sine waves all fit into specific parts of reactor opera-
type of approach; i.e., the only significant reactor constants that are
ties can be made external to the reactor, and its performance calculated.
It is well known, of course, that spatial effects inside the reactor can
box type of treatment usually yields kinetic answers that are consistent
Then solutions will be provided for ramp functions in Sk. Finally the
this way the transfer function will be derived for future use in control
circuits.
Chap. 2.
29
CONTROL OF NUCLEAR REACTORS AND POWER PLANTS
•
dn Sk —
Hi
• i
(3-D
(3-2)
where the symbols have the meanings previously denned. The nature
of the solutions of Eqs. (3-1) and (3-2) for step function inputs is also
8k
0.010
0.008
0.006
0.004
0.002
t"=10"3 SEC
. l*=10"5 SEC
0.01
AI
0.1
P,, SECONDS-'
n—rm
Generated on 2011-10-24 11:40 GMT / Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google
10
FIG. 3-1. Chart for response of reactors to a step change in reactivity, positive step only.
n(0 =
(3-3)
where the first exponent PI has the same sign as Sk, the input disturbance,
30
values of Aj, Pji and Sk. This relationship can be shown graphically
REACTOR KINETICS
j=7 j=6 .
31
0.006
0.004
0.002
8k
-0.002
-0.004
-0.006
Pj, SECONDS"1
0'8[7TTTTT~I I in ! i M i T Mill i I I
-1.2
-1.6
-2.0
and is presented for reactors of I* = 10~3 sec, 10~4 sec, and 10~6 sec in
Figs. 3-1 and 3-2. Figures 3-1 and 3-2 are used when a positive 5/c step
These curves are used in the following manner: The size of the Sk step
is selected and the value of the exponent Pi is read from the top graph of
Fig. 3-1 or the farthest curve to the right top of Fig. 3-2. Then reading
Generated on 2011-10-24 11:40 GMT / Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google
vertically down from that value of PI to the bottom graph, the corre-
32 CONTROL OF NUCLEAR REACTORS AND POWER PLANTS
across the top curves of Fig. 3-2, from right to left, locates the value of
Pj opposite the ordinate of the selected Sk, and then reads down to the
Using these charts in an example, one might derive the equation for
0 Z 4 6 8 10 12 14
TIME, SECONDS
FIG. 3-3. Relative neutron level versus time for positive step function reactivity changes,
/* = 10-4sec.
— 0.0035fc is inserted into the reactor. Reading across and down for each
It can be seen from Figs. 3-1 and 3-2 that the positive exponent PI
very rapidly. These rapid rises take place at values close to prompt
critical.
I* of 10~4 sec for various step Sk inputs. The value of I*, as noted from
Generated on 2011-10-24 11:40 GMT / Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google
REACTOR KINETICS
33
Figs. 3-1 and 3-2, principally affects the front edge of the rise. After
an initial rise caused mostly by the prompt neutrons, the reactor settles
indicates the response of this reactor for longer periods of time, and it
20
40
100 120
140
60 80
TIME, SECONDS
FIG. 3-4. Relative neutron level versus time for positive step reactivity changes, /* = 10~4
I* of 10~4 sec, for negative 8k step inputs. Here it will again be noted
that after an initial drop, the rate of drop is determined by the delayed
emitters and the ultimate rate by the longest-lived delayed emitter, which,
be seen that the last term of this equation is the predominant one for very
short periods of time. However, the two terms preceding the last term
in Eq. (3-4) have comparatively small coefficients. The first four terms
Generated on 2011-10-24 11:40 GMT / Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google
have small exponents. Therefore, after the large initial drop, the neu-
4 6 8 10
TIME, SECONDS
FIG. 3-5. Relative neutron level versus time for negative step function reactivity changes,
/* = 1(T4 sec
20
120 140
40 60 80 100
TIME, SECONDS
34
FIG. 3-6. Relative neutron level versus time for negative step function reactivity changes,
35
KT
ur1
After the first dropoff, the reactor level cannot fall off at a rate faster
minimum of 30 min.
analyzed by an approximation
io*
KT
10 15 2O
TIME. MINUTES
25
30
shutdown.
1.0.
0.2
TIME, SECONDS
FIG. 3-8. Front edge shape for a step function change in reactivity showing results of using
approximate formula.
36 CONTROL OF NUCLEAR REACTORS AND POWER PLANTS
dn 6k - ft 8
dt I* P ° ( )
of — 8/(Sk — 8). How close the approximate expression matches the
actual case can be seen in Fig. 3-8. Here the initial rise created by the
pared with the approximate expression. The front edges are alike for
ramp function of the form Sk = a + yt into the black box produces use-
rod effectiveness is usually such that the reactivity does not change
intervals if desired.
The exact solution of the reactor kinetic equations for this type of
ramp input is a complex one, tedious to calculate, and will only be outlined
below, f Where many solutions are required, the use of a reactor kinetic
analytical method.
The analytical method follows these steps: the basic kinetic equations
(3-1) and (3-2) are first presumed to consist of only one group of delayed
dn Sk — 8 i \n \ g
^ = p n - \C (3-8)
equation in n whereby
t This solution was originally obtained by Sylvan Wallach, and the presentation
/72« •/«
n = ex'R(x) dx (3-11)
£
R (x) is denned as
P(x) = x2 + Bx + D (3-13)
now substitute Eq. (3-11) into Eq. (3-10) and find that
£
Substituting the definitions of Eqs. (3-13) and (3-14) into Eq. (3-15)
£
_p
\S = [R(x)Q(x}et]fl (3-20)
38 CONTROL OF NUCLEAR REACTORS AND POWER PLANTS
where Xi and x2 are the end points of the path of integration £. R(x)
may be found by substituting Eqs. (3-13) and (3-14) into Eq. (3-12)
= o (x + 0" '
where , - D ~ ~ (3-22)
£
these paths Eq. (3-11) must converge and Eq. (3-20) must be satisfied
integration £ can be taken as the positive real axis from zero to infinity.
when the reactivity exceeds prompt critical. Hence it is clear that as the
reactivity increases with time, the neutron density begins to grow very
rapidly at prompt critical, but below prompt critical the growth is com-
Generated on 2011-10-24 11:40 GMT / Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google
paratively small.
Figure 3-9 presents a solution to the ramp function input problem for
As can be seen from Eq. (3-23), the neutron-level response is not partic-
ularly sensitive to I*. It will be noted that below critical the form of the
The effect of different linear reactivity change rates may be seen from
Fig. 3-10. s The curves are plotted as a function of the reactivity remain-
39
It can be seen that as reactivity is inserted into the reactor at higher and
higher rates, the critical point comes at lower and lower neutron levels.
rates of change. Figure 3-11 indicates the periods that result as the
470 480
STARTS, SECONDS
-6 -5 -4 -3 -2 -1
rods rapidly in. order to see quickly a measurable period. Figure 3-12
seen that one actually has a somewhat more gradual operating approach
retical reactor operation is one in which the reactor is critical and reac-
1.25
6K CHANGE AT 0.0076/SEC
6k CHANGE AT 0.0038/SEC
6k CHANGE AT 0.0016/SEC
MULTIPLICATION FACTOR k
FIG. 3-11. Reactor period versus multiplication factor for various ramp function reactivity
change rates.
dn
dt
Sk —
now, however,
Therefore
I*
Generated on 2011-10-24 11:40 GMT / Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google
n+
(3-25)
(3-26)
(3-27)
40
Sk = At
dn _ At - ft
dt I* H
REACTOR KINETICS
41
2.5
*-6l
fik
CHANGE
CHANGE
CHANGE
6k CHAN
AT 0.00
AT 0.00
AT O.OC
GE AT 0
re/SEC
38/SEC
H6/SEC
00045/S
EC-*
10
//
'/
Generated on 2011-10-24 11:40 GMT / Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google
20
40
-sS
40
240
TIME, SECONDS
FIG. 3-12. Reactor period versus time for various ramp function reactivity change rates.
II
Then
a' C\
— 7!k
n0 r
2Al*
where
IA
Jo
and
7 = —
(3-31)
a ~ \2Z*
42 CONTROL OF NUCLEAR REACTORS AND POWER PLANTS
Substituting at + y = n, we get
rt
Jo
- dt = -
2"
r> /*o;(-T-'y
— /
VTT Jo
(3-32)
This form is that of the probability integral whose values are available
(3-33)
Plotted from this equation, Fig. 3-13 shows the results of suddenly apply-
8k CHANGE
'AT 0.06/SEC
100
8k CHANGE
0.012/SEC
400
500
Generated on 2011-10-24 11:40 GMT / Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google
200 300
TIME, MILLISECONDS
FIG. 3-13. Relative neutron level versus time for various ramp function reactivity change
n0
dn Sk ,„ ,
(3-34)
r\
\o-oo)
(3-38)
gives
Generated on 2011-10-24 11:41 GMT / Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google
»=i
where the initial condition transforms have been dropped. This is per-
n (* + A<)
Sn(s) _ no i = i
I* <.,
s [] (s + n)
(3-42)
-8
0.001
100
FIG. 3-14. Derivation of reactor transfer function amplitude using break frequency method
ri = 77.0 r4 = 0.336
r2 = 13.38 r6 = 0.080
sl f \ T*
Sk(s) I*
(3-43)
44
45
From this data it is now possible to plot the frequency response char-
-12
I*=5x10~5 SEC
l*=1.25x1CTs SEC
-V*=1Cr4SEC
-16
100
20
g-40
uj
< -60
UJ
<n
a -80
Generated on 2011-10-24 11:41 GMT / Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google
-100
-120
I I I I I Ml 1 1 I I I
/-V
-r=10-« SEC
l*-5x!0"*SEC
l*=1.25x10'3SEC-
0.001
0.01
III
10
0.1 1
100
the gain and the roots. Figure 3-15 gives the amplitude response of the
the gain factor no/l* by the gain of the frequency dependent portion,
•
n («+ ^)
s [I (s + r4)
t=i
lead term and those of the denominator as a single lag term. Thus
Sn(s) _ n0 . (s + X)
Sk(s) ~ ^ Al A,s(s + f)
is, the small signal sinusoidal gain of the reactor as a circuit element
intolerable one for a reactor control loop, and the dependence upon level
rod sinusoidally over a small amplitude, and fHe neutron output of the
REFERENCES
3. Soodak, H., and E. C. Campbell: "Elementary Pile Theory," John Wiley & Sons,
4. Goodman, C.: "The Science and Engineering of Nuclear Power," vol. 142,
13. Bode, H. W.: "Network Analysis and Feedback Amplifier Design," D. Van
Proc. Inst. Elec. Engrs. (London), vol. 100, pt. 1, p. 102, 1953.
Generated on 2011-10-24 11:41 GMT / Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google
CHAPTER 4
this chapter we shall first examine the reactor more closely as a control
and poisoning will be shown. When the various forms of the reactor
transfer function have been indicated, control loops will be tied around
the reactor and the system examined for stability and transient response.
later.
with Eq. (3-41) defining the reactor transfer function, completely describe
ji transfer function point of view, Eq. (3-41), the break point of highest
upon the value of fi/l*. It is the value of this root which determines the
control system operating clear out through the last break in the frequency
From another point of view, one can examine the open-loop transient
3-4. It will be noted that an attempt can be made to contain the initial
steep front edge of the rise by means of a control system. This initial
front, edge is also dependent upon ft/I*. However, if slower control can
48
AUTOMATIC REACTOR CONTROL
49
be tolerated, one can design a control system that ignores this front edge
and operates on the flatter portion of the open-loop response which occurs
viewpoint that a thermal reactor in which the break point is less than
ity problems exist, and they will be discussed later in this chapter.
There are two distinct types of feedback loops which must be con-
sidered. These are external loops and internal loops. The external and
"We~shall
Chap. 6 and for the present shall confine ourselves to two internal or
ing. The internal loops appear as indicated in Fig. 4-2, and each effect
>
1 —
LOCAL
TEMPERATURE
COEFFICIENT
FEEDBACK
KTCGTC(s)
POISONING
FEEDBACK
KXGX(S)
REACTOR
KRGR(S)
back paths.
re^1roT~a"s~a'~ruiictipn of temperature. In
135 atoms are created both directly and indirectly in the fission process,
and these atoms change the reactivity of the reactor in a complex manner
with time. The effect of the local temperature coefficient feedback loop
upon the reactor transfer function will be developed first. Then the
effect of the poisoning feedback loop upon the performance of this com-
been using for the reactor indicates that the reactor by itself is unstable.
That is, at zero frequency the reactor has infinite jjam. Physically, when
this~process weTshTrll consider tKeblock diagram of Fig. 4-3 and shall use
50
AUTOMATIC REACTOR CONTROL
KBGR(s)KTCGTc(s)
51
(4-1)
LOCAL TEMPERATURE
COEFFICIENT EFFECT
KRTCGRTC(S)
COMBINED REACTOR
AND
LOCAL TEMPERATURE
COEFFICIENT EFFECT
TRANSFER FUNCTION
(a) (b)
FIG. 4-3. Combination of reactor transfer function with local temperature coefficient feed-
back transfer function, (a) Individual transfer functions, (b) Combination transfer
function.
reactor coolant is also present in the reactor, this coolant would be fixed
Generated on 2011-10-24 11:41 GMT / Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google
temperature level. The heat from the fuel then causes a change in the
Tm = ASn (4-2)
Sn = neutron-level change
coolant.
describe the elementary heat flow from the fuel to the moderator. In
52 CONTROL OF NUCLEAR REACTORS AND POWER PLANTS
Laplace notation
To ~T~ i
That is,
The gain term of the transfer function may be lumped into a new gain
values of r and KTc, and the combination of this transfer function with the
Figs. 4-4 and 4-5 for a reactor having an Z* of 10~4 sec, T = 0.159 sec, and
amplitude curves of Fig. 4-4 indicate that at very low frequencies the
longer infinite at zero frequency, but a finite value of gain equal to 1 /KTC
Generated on 2011-10-24 11:41 GMT / Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google
transfer function takes on the shape of the reactor curve which depends
principally on I*. We can generalize to the extent that if the break fre-
quency caused by the time delay r is low compared with the highest
Figure 4-5 indicates that at low frequencies the phase shift approaches
zero for reactors with negative temperature coefficients, instead of — 90°
53
30
26
0.001
0.01
100
0.1 1
FIG. 4-4. Amplitude response of combination reactor and temperature coefficient feed-
150
-100
0.001
0.01
100
0.1 1
FIG. 4-5. Phase shift response of combination reactor and temperature coefficient feed-
again, the phase shift of the combination approaches the phase shift of
the reactor. Interestingly enough, some phase lead may result from the
combination, and the position and magnitude of this lead depend upon T.
It will be recalled that the reactor gain is a function of level and con-
sequently the decibel scale for the amplitude in Fig. 4-4 is relative to some
Generated on 2011-10-24 11:41 GMT / Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google
CONTROL OF NUCLEAR REACTORS AND POWER PLANTS
urally stable at high power levels is more difficult to control at low levels.
For this reason one might limit the range of automatic control to 1 or 2
decades of power level. For automatic control, then, the reactor should
not be shut down below 1 percent of full power. For analytical purposes
pendent of level.
From the example of Figs. 4-4 and 4-5, it is evident that this simple
stable. Other methods can be used for determining the stability of more
refining the above analysis and adding more terms to the feedback transfer
common one.
poisoning feedback loop can then be tied around this combined system.
Figure 4-6 shows the block diagram of the network that will be analyzed.
COMBINED
REACTOR AND
TEMPERATURE COEFFICIENT
TRANSFER FUNCTION
poisoning feedback.
dX'
dt
dt
= \,r +
= -\,r
- \xxr
(4-8)
(4-9)
54
divide Eqs. (4-8) and (4-9) by Z/ since the value of 2/ is not the same for
AUTOMATIC REACTOR CONTROL 55
become
function (Sec. 3-4), let us divide the variables into steady-state parts and
X = X0 + SX <t> = 0o + 50 7 = 70 + 57 (4-12)
«/(.) = (4-21)
o -p A;
Generated on 2011-10-24 11:42 GMT / Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google
i \.
function with its amplitude and phase shift very dependent on A~o which,
in turn, depends on 0o. The symbols 0 and n are being used inter-
engineer uses n.
X, = 2.9 X 10~6
\x = 2.1 X 10-6
y, = 0.056
yx = 0.003
in generalized form
v (7x -
r—-j- — .-
AI T 0x90
Generated on 2011-10-24 11:42 GMT / Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google
tion of neutron flux level. It can be seen that the poisoning rises linearly
flux levels the xenon concentration rises more slowly until at flux levels
concentration.
From a transfer function point of view there are two flux levels which
should be examined carefully. From Eq. (4-24) it can be seen that when
phase of the transfer function shifts its ultimate end point with frequency.
That is, at high frequencies when 0o < 3 X 1011, there is a total phase
shift in the xenon feedback path of —90°. When 0o > 3 X 1011, there
lasts until the point where \iyi/(yx — <rxX0) + X/ = 0. The transfer
function gain at zero frequency when yx — <rxXo — 0 can easily be deter-
function of 0o-
AUTOMATIC REACTOR CONTROL 57
.-r + X, - 0
as here the sign in the parentheses in the numerator of Eq. (4-24) would
7/ + 7x
(4-26)
But from Eq. (4-24) this cannot happen without infinite flux. Therefore,
at high flux levels above <j>0 = 3 X 1011, \ifi/(yx — <rxX0) + X; will always
1014
101
10s
108
10T
101'
1015 1016
Figures 4-7 and 4-8 show the relative response of the xenon concen-
tration to small oscillations in flux normalized in gain about the <p0 = 1014
case. The absolute level for this curve is +19.1 db. It will be noted
Generated on 2011-10-24 11:42 GMT / Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google
from the phase curves that for <£0 below 3 X 1011, the total phase lag at
high frequencies is 90° whereas at higher flux levels the total phase lag is
0o = 1010. At higher levels the gain steadily decreases with flux. This
fact could have been presumed from the saturation effect of Fig. 4-7a.
CONTROL OF NUCLEAR REACTORS AND POWER PLANTS
-40
10'
-50
-100
-150
-200
-250
-300
FIG. 4-8. Phase shift of transfer function of xenon feedback factor 5X(s) S<p(s).
58
Stability with Xenon Feedback. It has been shown that any elementary
this statement. Let us tie a xenon feedback loop around a reactor con-
Generated on 2011-10-24 11:42 GMT / Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google
AUTOMATIC REACTOR CONTROL 59
that of Fig. 4-9a. From Fig. 4-8 the xenon transfer function shows
180° phase lag in the neighborhood of 10~6 cycle/sec, and it is this fre-
(b)
(a)
KTC
(C)
FREQUENCY IN CPS
FIG. 4-9. Elementary block diagram of xenon feedback loop stability considerations,
(a) Individual transfer functions, (b) Combined transfer functions, (c) Approximation
shift of 0°. This condition comes about because the time constant of the
4-5). This situation is shown in Fig. 4-96 and c, and for this analysis we
shall assume that none of the frequencies above 10~3 cycles is important.
The transfer functions for the xenon concentration shown in Figs. 4-7
factor c)F the reactor is first caused to deviate from unity. The rods are
system: The power output level of the reactor must be the demanded
power level within a given error, and the multiplication factor must
ultimately be 1.
A given reactor may contain many control rods. These control rods
rods are moved quite slowly and usually function in control work as a
called shim rods and may be worth several percent in reactivity. Because
the purpose of shim rods is usually to take care of slow changes such as
"common usage.
and to handle intermittent transients which may occur, one rod may be
safety, the maximum worth of a regulator rod is usually kept below 0.006
in reactivity. In this manner the regulator rod can never inject as much
as prompt critical reactivity into the reactor. In the event that more
reactivity than 0.006 is needed for a particular operation, the shim rods
after the regulator rod has run out of reactivity, the shim rods can be
called upon to move in the same direction as the regulator rod and thus
We have seen from Sec. 2-11 that manual control of a reactor is apt to
systems can be designed to hold reactor power level to within 0.5 percent.
hflWfWPr, nftJSft is limially not a aoriniia prnhlpm^ nnrl fnr mn.ny piirprvaog
Figure 4-11 shows a broad block diagram of the type of control system
POWER
DEMAND
COMPARATOR
ERROR
SIGNAL
Generated on 2011-10-24 11:42 GMT / Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google
AMPLIFIER
ACTUATOR
NEUTRON
DETECTOR
REACTOR
CONTROL
RODS
FIG. 4-11. Block diagram of the essential elements of a reactor automatic-control loop.
tion chamber. This reactor output is then compared with the desired
power demand in the comparator, and any error between the actual out-
put and the demanded output is noted and amplified in the error-signal
rods the proper amount and in the proper direction to eliminate the
62
63
devices have all been used, and some will be described in Chap. 5. A
brief description now follows of each of the components in the loop from
trol loop is its nonlinearity. As mentioned in Sec. 3-4, the reactor gain
provides that the error signal is essentially the subtraction between the
should be error/level.
7o = Eb/K. Then
parator circuit.
form of com-
Ve = KVn -
- - Et
V F (n
V e = .C/b I
n0
Sn
n0
(4-27)
(4-28)
(4-29)
(4-30)
That is, if the reactor power level changes, for complete cancellation of
the actual level existing at any moment in the reactor. The difference
ve =
R + DP
Vn - Eb
(4-31)
Vn = V0 V< = 0 (4-32)
Eb =
R + DP °
Fo = Eb l + —
(4-33)
becomes
V, =
but
R + DP
Eb
n ~ Fo)
R + DP Fo
(4-34)
(4-35)
(4-36)
64
Here R' and R", in parallel, have the same resistance as the original R
of Fig. 4-13. The bias voltage of the amplifier also becomes the equiv-
1'HH JL
FEEDBACK
MAGNETIC
AMPLIFIER
one control signal tends to saturate the core while the other tends to
be of the form IL = A(n — n0) where A is the effective gain of the mag-
netic amplifier and n and no are input currents proportional to the neutron
Generated on 2011-10-24 11:42 GMT / Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google
66 CONTROL OF NUCLEAR REACTORS AND POWER PLANTS
customary positive feedback, and the familiar expression for the gain of
_ A(n- no) . .
with 1,
/. = -° (4-38)
= n-n0
n0
motor directly, such as would be done if the control motor were a two-
phase a-c servomotor, a few kilowatts might even be required from the
problem in comparison with that of the rest of the system. The amplifier
Generated on 2011-10-24 11:42 GMT / Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google
any physical form of actuator and mechanism may be used. The actua-
range from a few cycles per minute to a few cycles per second. The
actuator usually contains a power amplifier, and the output from the
VOLTAGE
of an ionization chamber.
Chap. 5.
the question: What do we expect from this loop? The first requirement
permitted. The second requirement is that the loop shall respond sat-
are generally of two types: a change in the power demand level or a change
tems involving the reactor. Both proportional and on-off type systems
now examine the over-all loop response as a function of some of the char-
CONTROL OF NUCLEAR REACTORS AND POWER PLANTS
transfer functions.
system.
0.001
2000
0.01 0.1 1 10
FIG. 4-19. Normalized transfer functions of external reactor control circuit for three values
GB(s) =
(4-40)
68
would be the form of the frequency variant portion of the transfer func-
/« = wn/2?r. For simplification let us assume that the damping coeffi-
cient f is fixed at f = 0.5 and that we are using a reactor having the trans-
AUTOMATIC REACTOR CONTROL
69
-1,0
fer function given for an I* of 10~4 sec. Our simple block diagram
KBGB(s) is shown in Fig. 4-19 for three values of /n. These values of
and the Nyquist plots14 giving the loci of the open-loop gain points may
performance.
The common types of network that have been used in reactor control
systems are given in Fig. 4-21. It must be emphasized again that this
analysis was for a reactor with zero temperature coefficient. The method
coefficient.
fn=1.0 CYCLES
PER SECOND
fn=0.1 CYCLES
PER SECOND
system is the simple on-off type of control. Here, when the difference
and caused to run at constant velocity until the error signal is reduced to
dead zone in its control and is inherently incapable of the same accuracy
CONTROL OF NUCLEAR REACTORS AND POWER PLANTS
that mechanical brakes, gearing, and other forms of friction may be used
to keep a control rod in a given position within the dead zone. In the
(c)
FIG. 4-21. Types of compensation networks commonly used in reactor automatic control
loops, (a) Tachometer feedback, (b) Position feedback, (c) Derivative network.
70
developed by the control-drive motor and thus the complete system power
type system is that it is less susceptible to noise because of its dead zone.
experimental reactors, and the control system for the CP-3 reactor has
Generated on 2011-10-24 11:43 GMT / Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google
AUTOMATIC REACTOR CONTROL
71
presence of the dead zone in the on-off system may well cause the control
CP-3, are reactors that in general merely maintain a fixed power level.
These reactors usually have large values of I*, and reactivity is changed
only at very slow rates. The design of all auxiliary systems about the
no fast transients can occur. Therefore, in the study of the design of the
control system for such reactors, the important factor becomes stability
Generated on 2011-10-24 11:43 GMT / Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google
PHASE
~j COMPENSATION [
ERROR
SIGNAL
AMPLIFIER
CONTACTOR
AMPLIFIER
REACTOR
CONTROL ROD
tion. The reactor of this block diagram may be the type of reactor
described in Sec. 4-3. The reactor and its temperature coefficient will
vide phase shift compensation. The analysis will first be shown without
this compensating network, and the network later added to increase the
72 CONTROL OF NUCLEAR REACTORS AND POWER PLANTS
close a set of contacts when the error-signal level reaches a fixed amount,
then open these contacts when the error-signal level drops below another
similar manner when the sign of the error signal reverses. These con-
stopped when the signal level is too small to keep either set of contacts
ity after an initial time lag in getting started. A control rod is assumed
feature in this reactor control loop, causes the loop gain to be a function
back factor which equals — 1 in this reactor control loop. The over-all
Output = KAG KA
Expressing the over-all transfer function in the form given in Eq. (4-41)
is from a polar plot of G~l and — KA. The value of G~l can be plotted
for all values of frequency, and the value of — K A can be plotted for all
values of amplitude. If the two loci intersect, that is, if G~l = — KA, the
section on the G~1 locus, and the amplitude can be found from the — KA
locus.
In the analysis of the block diagram of Fig. 4-22, the loop gain of the
function G is the product of the transfer functions of the reactor, the error-
signal amplifier, and the drive motor. The function A comes from the
contactor amplifier alone and describes the effects of the relays. The
-b „ -a
it
1 no~n
ERROR SIGNAL
W.
a.
Figure 4-23 illustrates the relay cycle and the terminology that will
be used. The error signal into the relays is proportional to (n0 — n)/n0.
As the error signal increases, it reaches the point b, which closes the relay
Generated on 2011-10-24 11:43 GMT / Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google
starting the drive motor and creating the reactivity rate demand signal
V. Once the control rods are started moving in the direction to reduce
the error signal, the hysteresis of the relay causes the contacts in the relay
to remain closed until the point a is reached and the drive motor is turned
tive signals, with — b being the negative start limit and —a being the
negative stop limit. Let us now assume that the output of the error-
signal amplifier is x = \x\ sin otf. Figure 4-24 indicates the phase rela-
of the output signal. The terms which are of consequence in the analysis
are the ratios b/a, V/b, and \x\/b. From this figure it can be seen that
greater than 1, phase shift occurs between x and v, with the output phase
lagging the input phase. The worst case is when a = 0, and at this point
CONTROL OF NUCLEAR REACTORS AND POWER PLANTS
a 45° phase shift occurs for \x\/b = 1. When \x\/b is very large, it can
also be seen that the phase shift between output and input approaches zero.
wave output is also a function of b/a and x\/b. It is obvious that the
T-CONTACTOR
\ OUTPUT
V-FUNDAMENTAL COMPONENT
\ OF CONTACTOR OUTPUT, V
INPUT
b=Q
b>a
*D
74
gain of the contactor, that is, its output/input ratio, falls off as |a;| is
increased. The ratio V/b appears directly as a gain factor in the con-
the fixed gain multiplier V/b is plotted in Fig. 4-25 for various values of
AUTOMATIC REACTOR CONTROL
75
b/a. It can be seen from this curve that the locus terminates abruptly at
\x\/b = 1. Obviously, for all amplitudes of x less than the start limit b,
the system does not respond. Figure 4-25 also indicates clearly that the
may replot the results of Fig. 4-25 in polar coordinates as Fig. 4-26. In
this figure we have also reversed the sign to obtain —Ab/a(\x\/b'). The
120'
150°
180°-4
1.1 10
1*1=1.4
abrupt stop at x\/b = 1 and the maximum phase shift of 45° are again
emphasized. These curves are presented again for various values of b/a
specific loop. The transfer function form of the contactor has just been
derived. The actual reactivity rate into the reactor is the contactor out-
put v delayed by the lag in the drive motor, which has the form 1/(1 + rs).
CONTROL OF NUCLEAR REACTORS AND POWER PLANTS
reactor and motor. For this example we may choose a specific value for
T = 0.5 sec. This is a reasonable value which many actual motors can
exceed.
For the transfer function of the reactor we shall select the combined
4-14 and 4-15 and carry through the analysis for different values of KTc.
It will be noted that our input to the reactor is now not a change in reac-
=x=lxl sin
Sn/n
sSk
FIG. 4-27. Servo block diagram of discontinuous reactor control loop indicating transfer
input rate. That is, the transfer function is of the form (Sn/n)/sSk.
The 1/s term will modify the transfer function representation of Fig. 4-4
tion of frequency.
transfer function G(s), invert this function, and plot it on a polar diagram
as shown in Fig. 4-28. It can be seen from this that the larger the tem-
Generated on 2011-10-24 11:43 GMT / Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google
smaller enclosed area drastically limits the range of V/b which can be
permitted for a stable system. The shape of these curves will, of course,
76
Let us now go back to the comparison of the amplitude function and the
frequency function by combining the curves of Fig. 4-26 and Fig. 4-28 to
the same scale. This combination is shown in Fig. 4-29. Under these
conditions it will be recalled that V/b = 1 and it is obvious that the sys-
tem will oscillate for all but the very largest of the temperature coeffi-
AUTOMATIC REACTOR CONTROL
77
cient gain time constants. Even for the case where KTc = 0.019, the
10~3 to 10~6 5fc/sec. The start limit b depends upon the accuracy to
100°
90°
110°
120'
130
140'
170'
180
20
-20
GAIN, db
V/b exists whereby, if its value were larger than a given amount or smaller
than another amount, the system could oscillate. For reactors having
does not exist. The only problem is to maintain V/b below a given
amount. Gain margins and phase margins here are determined by expe-
rience. In linear servo practice a phase margin of 35° and a gain margin
of the contactor.
may provide a control system that does not oscillate but nevertheless
does not have suitable phase margin for a desired transient response.
100°
110°
120
130
140'
150'
170
180
160'
20
-20 -40
GAIN, db
-60
FIG. 4-29. Comparison at unity gain of amplitude function — A»/a(|x|/b) with frequency
function G~1(s).
and KTC = 0.0047. This situation is shown in Fig. 4-30. It can be seen
Generated on 2011-10-24 11:43 GMT / Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google
that although adequate gain margin exists, the phase margin is only 22°,
which may be small when one considers that this elementary analysis did
not take into account any of the second-order effects which must exist
and a G~l curve modified, as shown by the dotted line in Fig. 4-30. Here
79
sient response, but these qualifications depend upon the reactor usage
damaging to most reactors providing it does not last too long. However,
100°
110°
120'
130
140f
170'
180'
150°
160°
20
-60
0 -20 -40
GAIN, db
FIG. 4-30. Stability of discontinuous control-loop example with and without phase com-
pensation.
power demand change rate can be limited externally to any desired value,
Given a specific control loop, how does the reactor output power level
from any cause? Controlwise, the problem resolves into: Given the
reactivity?
simplify the block diagram of Fig. 4-11 into the servo block diagram of
Fig. 4-31. Here again the reactor transfer function has been indicated
mechanism have been lumped together into the common transfer func-
between F0 and Fn, and the neutron detector is indicated as having only
FIG. 4-31. Servo block diagram of reactor control loop used for transient analysis.
5fce. It will be noted from this block diagram that no phase compensation
has been added and, for simplicity, the temperature coefficient of reactiv-
ity has been ignored. From the block diagram of Fig. 4-31 we can solve
for the Laplace transform Vn(s) in terms of the transfer functions of the
system components and the reactivity. We can then take the inverse
transform to obtain finally Vn(t). From Fig. 4-31 we can see the follow-
Vn(s) = K0n(s)
Generated on 2011-10-24 11:44 GMT / Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google
(4-42)
(4-43)
(4-44)
(4-45)
80
the reactor given in Sec. 3-4 that only oscillations about a steady-state
K0KBGR(s)Ske(s)
K0KBKBGR(s)GB(s)
(4-52)
I •".||*».If-'Itv«v>"v«v«V i A ro\
h 1 + KoKBKBGR(s)GB(s) l
We can assume that during the reactivity change, the power-level demand
are going to attempt to control our reactor with a fast control system of
such a speed that we do not have to consider delayed neutrons but rather
that the response of the control system is so fast that we can ignore the
(4-54)
be given by
We are now ready to substitute the expressions for the transfer functions
*2
k^ (4-56)
82 CONTROL OF NUCLEAR REACTORS AND POWER PLANTS
«*.(«) = —
(4-57)
where Afc represents the magnitude of the step change. Equation (4-56)
where
S2 + diS + OO
T s4 + t
flo = <*>2
ai = 2fo>
..
(4-58)
the constants used. For illustration, Eq. (4-59) can be solved for the case
Ul
s^-6k=0.02 STEP
6k=0.015 STEP-
8k=0.003 STEP
r6k=0.001 STEP
0.2
0.4
Generated on 2011-10-24 11:44 GMT / Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google
0.12
0.14
0.16
TIME, SECONDS
FIG. 4-32. Transient response of an example of proportional-type control loop for various
KoK. = 120F0, and K0K,KB = 1.62. Figure 4-32 shows the response of
the transient response are a direct function of the input Sk step amplitude
and that the time for the transient to return to its steady point is approx-
the response, as determined in this manner, will be correct only for rela-
DEMAND
<8>
Sec. 4-10.
required from an over-all reactor control system. It has been seen, from
mized one against another, provided that a set of standards are laid down.
tive temperature coefficients. The magnitude and the effective time con-
4. The restoring time of the system—the time for the transient to die
out essentially.
problem.
Fig. 4-34. In this block diagram we have a reactor that, for the sake of
COMPARATOR
SYSTEM
example, has a mean neutron lifetime of 10~4 sec and, consequently, the
transfer function given in Figs. 3-15 and 3-16. The reactor also has a
The heat-transfer coefficient from the fuel of the reactor to the mod-
study its effect upon the control system from ranges of KTC = 0 to
84
(4-60)
AUTOMATIC REACTOR CONTROL 85
HQ
The choice of this form of the transfer function of the control-rod system
some of the practical servo systems now in use in reactor control systems.
block diagram of Fig. 4-34 that an error can exist between n and n0 and
yet the reactivity requirements that k = 1 for steady state of the reactor
steady-state error, but for many control systems this portion of the trans-
It can now be seen that to study this given reactor control system
the methods outlined in Secs. 4-7, 4-8, and 4-9. To determine the stabil-
ity and to solve for the transient responses for a range of values of all
the various parameters within reason for a system of this sort by means
of these methods is a long and tedious procedure. With this large num-
that a specific simulator has been designed for the example of Fig. 4-34
used will be to set up the simulated plant, inject given types of disturb-
86 CONTROL OF NUCLEAR REACTORS AND POWER PLANTS
ances, and study the effects of changing the parameters one at a time.
we can select for this study are that the reactor is operating at full power,
disturbance will then be inserted into the reactor, and the form of the
type of input are of interest because the analytical calculations for this
gives the worst transient response in most physical systems, and thus an
power level.
one in the over-all reactor control system. In fact, for many types of
will be defined by the use of KTc — ATC, as indicated in Sec. 4-3.
rod servo system. It will be found in this type of problem that the damp-
ing factor, the gain, and the integrating break frequency can all be set
These optimum constants can then be used while the natural frequency
87
ure 4-35 shows the effect of varying -f from 0.5 to 1.5, when a 0.0035A;
grating break frequency of 0.032 cycle/sec. It can be seen that the over-
damped servo provides the best of the three damping coefficients con-
sidered. This figure also shows that a higher gain is permissible for the
state following the transient differs little between the cases shown in
170
100
34567
TIME, SECONDS
FIG. 4-35. Control-system transient response to 0.0035(t step function as damping coeffi-
cient is varied.
Fig. 4-35 in that the damping ratio affects principally the extent of the
Gain. As Fig. 4-35 indicates that f = 1.5 gives the best system
Generated on 2011-10-24 11:44 GMT / Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google
response, only data taken from f = 1.5 need be considered from here on.
The next parameter to optimize is the loop gain K. The effects of vary-
ing K are indicated in Fig. 4-36. This figure illustrates that the loop
transient shape. The primary effect of higher gain is to reduce the ampli-
tude of the tail of the transient without seriously affecting its slope. If
100J—
10
11
4567
TIME. SECONDS
FIG. 4-36. Control-system transient response to 0.0035k step function as loop gain is
varied.
4567
TIME, SECONDS
10 1)
FIG. 4-37. Control-system transient response to 0.0035fc step function as ratio of inte-
Sk
\n — nQ)/no
might be permitted.
88
Integrating Break Frequency. We have just seen that the gain and the
control servo alone. The measure of the servo performance of the tem-
peak power values thus obtained are projected horizontally to the right,
the intercepts with the zero temperature coefficient curve give the equiv-
300
0.1 1.0
FIG. 4-41. Reactor peak power level reached as a function of natural frequency and
of limiting the peak power output value which a reactor can attain under
the condition of step input transients. This answer can also be obtained
moment, except to state that it is a strong one. For the case of a step
Generated on 2011-10-24 11:45 GMT / Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google
dn
Tt
Sk —
n+
..=i
(4-61)
(4-62)
92
~dt
AUTOMATIC REACTOR CONTROL 93
If we now inject a step Sk into the reactor, we can isolate the steady-
n = n0 + Sn (4-64)
Then
d(Sn) _
ttL \ V J V
Sk 8 - Sk
T* o 7*
fj — O/C
and the maximum initial value that Sn can obtain for positive steps in
Sk
n0 (4-67)
mn ,
given in Eq. (4-67). The time constants are the important factors here.
then the peak in Sn can never rise above the value given in Eq. (4-67).
I* of 10~4 sec. The time constant for the initial rise l*/(fi — Sk) becomes
(see Sec. 3-2), reactors having fast temperature coefficient time constants
Figure 4-42 plots Eq. (4-67) and indicates the peak powers that can be
ing the peak power to the value indicated for a 0.0035A; step in Fig. 4-42.
The curves of Fig. 4-39, however, indicate that the temperature coeffi-
lished four criteria in Sec. 4-10 for comparing reactor control systems,
22
I"
£
1.4
1.2
1.0.
BREflKDOWN
LEVEL
0.004
0.002
Generated on 2011-10-24 11:45 GMT / Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google
8k-STEP
SAFETY FACTOR
must be considered.
The first problem concerns itself with the question: What are the lim-
itations of the reactor that the control system must protect? Basically,
the form of the permissible amount of power that may be taken from any
Q
AUTOMATIC REACTOR CONTROL 95
above its normal full power rating. If one now considers operation for
short times, that is, times of the order of the time constant of the thermal
heating of the generator, a slightly greater output from the machine may
be obtained. If the generator will not reach its normal internal tempera-
ture ratings for 5 min at full power, then conceivably one might take more
than full power from the generator for a time shorter than 5 min. This
might be drawn from the machine for 5 min, one could conceivably take
twice full power from the machine for 2)^ min or four times full power for
little heat capacity. This means that the time scale on Fig. 4-43 might
be shifted so that the continuous rating plus the safety factor must be
carried down into times comparable to 0.1 sec. If this is the case, for
times shorter than 0.1 sec it is indeed possible to run power reactors at
higher power level than their rated level up to the point where something
be translated into power level for use with our four basic criteria. For
response curves in Figs. 4-38, 4-39, and 4-40. Two sets of information
are at once apparent. The first is that under some conditions no exter-
with modest input transients and no control system would not appear to
be too severe for many types of reactor plants. The second obvious
information that may be derived from these curves is that if one does use
an external control system, the higher the natural frequency the better;
or the faster the control system acts, the smaller will be the over-all
transient disturbance.
The question then arises: Why not build extremely fast control sys-
the components which are used in these fast systems are not at the
96 CONTROL OF NUCLEAR REACTORS AND POWER PLANTS
The faster the capabilities of the system, the more severe the transient.
Therefore, the argument may be used that if nothing about the reactor
can occur. From a safety point of view one should then design any
needed control system to be as slow as possible and still satisfy the over-
control. That is, means must be provided to extract control rods for
reactor.
•or the ranges that the temperature coefficient might have during the
reactor lifetime.
as a function of time.
transfer functions.
tion techniques or others, and evaluate by means of the four criteria given
6. Choose the slowest, safest system that fits all the above conditions.
REFERENCES
5. Harrer, J. M., and J. A. Deshong, Jr.: Discontinuous Servo for Control of Power
6. Grace, J. N.: Synthesis of Control Systems for Nuclear Power Plants, "Conven-
tion Record of the IRE 1954 National Convention," pt. 9, "Medical and Nuclear
ing Contactor Servomechanisms, Trans. AIEE, vol. 69, pt. 1, pp. 270-284, 1950.
10. Bode, H. W.: "Network Analysis and Feedback Amplifier Design," D. Van
12. Moore, R. V.: The Control of a Thermal Neutron Reactor, Proc. Inst. Elec. Engrs.
Proc. Inst. Elec. Engrs. (London), vol. 100, pt. 1, p. 102, 1953.
14. Nyquist, H.: Regeneration Theory, Bell System Tech. J., vol. 11, p. 126, January,
1932.
15. Weinberg, A. M., and W. K. Ergen: Some Aspects of Non-Linear Kinetics, Proc.
16. Lipkin, H. J.: A Study of the Non-Linear Kinetics of the Chatillon Reactor, J.
Generated on 2011-10-24 11:45 GMT / Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google
In Chap. 4 reactor control loops were examined and it was shown that
the output of the control system was usually a control-rod motion which
actuate these control rods. Of course, moving control rods is not the
types of control rods. That is, we are not going to consider forms of
reflector control or other schemes that change the leakage of the reactor,
but the assumption will be made in all cases that moving a control rod into
the reactor structure reduces the reactivity and tends to turn the reactor
off.
there have been reactors built. The only requirement for the absorptive
type of rod is that the rod be black or at least "dark gray" to thermal
neutrons. Many shapes of control rods have also been used. Some
have been round, some square, some hollow, and some solid, all depending
upon the design of the individual reactors. For thermal reactors, boron
and cadmium are the most commonly used absorbing materials, and
these materials may be alloyed into some other metal or may be clad by
ical requirements of the mechanisms that move these rods in and out of
the reactor. Brief descriptions will be given of the features of the simpler
typical systems.
tion: safety rods, shim rods, or regulator rods. Each function may call
this power level. The control aspects and speed requirements of the
the regulator rod was usually automatic, but manual operation could be
permitted.
As has been previously indicated, shim rods are used either for auto-
matic follow-up of the regulator rod or for the slow movement of larger
Safety rods, on the other hand, are for scramming or rapid shutdown
action, and their motion inward is quite fast. Details of the speed
exist for each of these three functions. In some reactors these functions
will be separated, and in others they may be combined. That is, during
startup a shim rod may be operated at its normal slow speed, but as soon
reactors is that regardless of the normal function of the control rod, it will
also have provision for some form of fast insertion into the reactor.
For many purposes it is also desirable to move more than one rod at
room pressure ambient. Thus they can havfe their mechanisms out in
the open where they are easily accessible and where lubricants can be
The AEG has announced that the first commercial central station
conceivable power reactors are also of the pressurized type. The coolant
100
inside a pressure shell or some complex means used to control the reactor
completely inside a large pressure vessel type of tank or, as is more com-
mon, some form of thimble may be used in which the mechanism can
Fig. 5-16 an external thimble setup. In Fig. 5-la the entire mechanism
-CONTROL ROD
DRIVE
MECHANISM
EACTOR
ESSURE
ESSEL"^,
CONTROL ROD
/////////////
\l
r/////// '//////
\\
INTERNAL ^
THIMBLE X
11
Generated on 2011-10-24 11:45 GMT / Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google
1 REACTOR
'////////////,
/////
/////////////
EXTERNAL
THIMBLE
REACTOR
- CONTROL
ROD DRIVE
MECHANISM
,REACTOR
PRESSURE
VESSEL
CONTROL
ROD
(a) (b)
FIG. 5-1. Pressurized reactor thimbles, (a) Internal thimble, (b) External thimble.
and control rod are actually outside the reactor pressure vessel, but the
structure. In the external type of thimble the control rod alone extends
into the reactor but the thimble contains a part of the drive mechanism.
Some form of coupling then must be arranged through the thimble wall
to the basic control-rod drive. It can also be seen that the portion of the
and assemble the pressure vessel and reactor structure for the internal
thimble system when there are many rods used in the reactor.
control-rod mechanism, there are other factors which also must be con-
always exists is that of providing some form of energy storage either in the
rods completely into the reactor. Many forms of energy storage, such
later.
expected life of the reactor and the mechanism. Because of the difficulty
must be designed for perfect operation for a long time. This problem
Means for softening these forces are usually provided in most mechanism
designs.
design limitation.
control mechanism might have to operate under external shock and vibra-
ment. For example, take the case of scramming the control rod. From
so-called hair-trigger type, with which just a very small motion will
cause the rod to scram, but this type of mechanism could also cause a
such that it did not unlatch easily under shock for its normal operation, it
might take too long in releasing the rod for a scramming operation.
Systems. The motors and mechanisms that may be used for control-rod
CONTROL OF NUCLEAR REACTORS AND POWER PLANTS
drives can be arbitrarily divided into two groups: those mechanisms used
Electric motors are generally used in the basic power arrangement for
the control-rod drive. The simplest cases are those whereby only a con-
stant speed is required of the motor. The motor is then geared down
through the rod-driving mechanism to set the rod speed at the desired
AC LINE
-L 90° PHASE
SHIFT
FIG. 5-2. Control loop for two-phase a-c servomotor drive system.
phase in a three-phase motor causes the motor to lose its ability to change
the direction of travel of the control rod. A-c induction motors are
preferred over d-c motors in that no accident can cause them to operate
Two-phase induction motors are easily reversed and have the advan-
tage that certain designs may be used directly as servo control motors.
Figure 5-2 indicates a control loop such as might have been used in Chap.
diagram it will be noted that the d-c output of the neutron detector must
102
REACTOR CONTROL MECHANISMS
103
drives and screw-type drives are quite common. Figure 5-3 indicates the
simple rack and pinion, whereby a drive motor of any type drives a pin-
means of slip at the ends of the rod travel and also provides one means of
= CLUTCH
PINION
-RACK
-CONTROL ROD
REACTOR
direct gravity scram by releasing the control rod from the gearbox. This
Fig. 5-46 indicates the movable screw arrangement. Figure 5-46 can
Unless the shape of the control rod is round, some form of keyway is
also been suggested. These drives are similar to the conventional auto-
Generated on 2011-10-24 11:46 GMT / Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google
Generated on 2011-10-24 11:46 GMT / Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google
REACTOR CONTROL MECHANISMS
105
cable arrangement, but any of the other means may be used. The neces-
have been developed.1 One such type is the so-called canned synchro-
ELECTROMAGNET
NON MAGNETIC
PRESSURE SHELL
at very slow speeds, and the amount of gearing that must be used can
advantage.
Here three-phase windings are shown, and we can first consider the case
where the windings are supplied from a d-c bus. The armature can be
Generated on 2011-10-24 11:46 GMT / Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google
CONTROL OF NUCLEAR REACTORS AND POWER PLANTS
pole bar in Fig. 5-6. If the windings are fed as shown with winding A
and winding B placed across the d-c line and winding C being left open,
a fixed field configuration will be set up inside the motor and the bar will
may be tolerated.
By feeding full current through one phase winding and half current
through the other phases, 60° steps may be obtained. Figure 5-7 shows
106
magnitude are feasible for many applications. It will be noted that the
Generated on 2011-10-24 11:46 GMT / Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google
the motor air gap there is a 0.02-in. Inconel can which protects the
stator windings and punchings from the water and also acts as the pres-
sure wall. The can is, of course, backed up to take the actual pressure
by the stator core. The leads of the stator windings are brought out
from the top of the frame through high-pressure Kovar glass terminal
pressure backup again for the stator and prevents release of any contam-
inated water to the outside atmosphere in case a leak develops in the can.
The rotor and the shaft are made of a magnetic stainless steel. The
To illustrate the performance of the motor of Fig. 5-8, Table 5-1 presents
TABLE 5-1
Design characteristic
State
Rating
Speed, rpm
Normal
High-speed operation
Maximum
Holding (continuous)
0-16
53
Generated on 2011-10-24 11:46 GMT / Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google
1,800
Torque, Ib-ft
Running (intermittent)
Normal
0 15
Minimum
Normal
0.109
120
Maximum
Normal
130
Maximum (intermittent)
150
250
2,500
Pressure, psi
developed which creates linear motion directly without the need of going
through rotating motion first.2-3's Linear motors have been used for
some time for other applications such as catapult drives4 and X-ray
obtain a crude feeling for the operation of the linear motor consider first
were figuratively to be sliced open at one axial place, flattened out, and
similar to the cross section of Fig. 5-9 would be obtained. Here the field
coils are nothing but circular doughnuts around a long tube. The arma-
ture consists of a series of ringlike poles on a long bar. The winding slots
108
Generated on 2011-10-24 11:46 GMT / Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google
CONTROL OF NUCLEAR REACTORS AND POWER PLANTS
POSITION
INDICATOR
DEMAND
NEUTRON
SIGNAL
CONTROL
ROD
'////////////- //,
HYDR
P\P€.
4ULIC
LINES
/A y////////////
HYDRAULIC
MOTOR
* GEARING
, REACTOR
/ PRESSURE
/ VESSEL
REA
^-CONTROL
Generated on 2011-10-24 11:47 GMT / Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google
ROD
:TOR
//////////////////////////////////,
110
FIG. 5-11. Pressurized system using internal hydraulic motor rod drive.
REACTOR CONTROL MECHANISMS
111
or hydraulic systems will depend upon the load, the accessibility of the
DEMAND
^CONTROL
ROD
REACTOR
Y/////////////////////////////////'/
Figure 5-11 shows one arrangement for using a hydraulic motor com-
ularly advantageous if the hydraulic fluid can be the same fluid as the
coolant or moderator so that any leakage from the hydraulic motor does
not contaminate the system. The hydraulic motor may be of any con-
leakage is permitted, radioactive coolant may escape and soon make the
pressure seal could conceivably be as severe as "should not leak one drop
in a month."
Figure 5-13 indicates one of the forms that a rotating shaft seal might
The packing glands serve also to reduce the pressure from the inside to
the outside in gradual steps until ultimately the last gland is at atmos-
pheric pressure. Some leakage will still result through these glands, and
tap-offs are provided at various points between glands so that this leakage
may be bled off into a storage tank. Figure 5-13 indicates the final tap-
off at atmospheric pressure. From the storage tank another line and
pump are provided to reinsert any coolant leakage back into the primary
system. Similar seals for direct transmission of linear motion are also
available.
nisms that can be used to provide the normal slow drives for control rods
have some form of scramming mechanism built in, either directly at the
rod or indirectly in the control-drive feed. There are two distinct types
used depends upon whether the rod is inserted into the reactor vertically
always used to assist in the scramming process. When the rod is inserted
rod to release by gravity. Figure 5-5 shows such a system for the pres-
surized type drive, and simpler systems can be devised for nonpressurized
reactors.
the fact that some designs require a large amount of magnet current
mechanical latch and accomplish the same effect, usually with more
speed and with large decreases in driving power. One such scheme is
indicated in Fig. 5-14a. Here a large arm is held against a detent in the
When the current is disconnected from this coil, a spring forces the large
tage is taken of the fact that the reluctance motor type of motor drive
ate effectively at any line frequency supplied to it. Under normal con-
ditions the supply frequency might be a few cycles per second. However,
inward motion only the 60-cycle line will be connected to the reluctance
accelerated rates.
Figure 5-156 indicates a system using two motors. One motor is the
fa)
DIFFERENTIAL
GEARING
(b)
FIG. 5-15. Powered scram drives, (a) Two-frequency reluctance motor operation, (fa)
Two-motor system.
direction only. Again proper switching throws one motor or the other
5-5. Energy Storage Devices. All reactor control systems must pro-
vide for some form of energy storage in order to move the rods to their
simplest and most direct energy storage means is gravity, when the
mechanism design permits its use. For reactors having vertical control
rods, gravity dropping of the rods in the event of power failure is very
practical. For reactors with horizontal rods, gravity may also be used,
CONTROL OF NUCLEAR REACTORS AND POWER PLANTS
but the system is more complicated. Any type of energy storage may be
The type of storage used when gravity storage is not directly applicable
batteries used with d-c motors and appropriate switching provide a good
WEIGHT
PISTON
ACCUMULATOR
OPERATING
""" VALVE
HYDRAULIC
PUMP
HYDRAULIC
CONTROL
VALVES
DUMP
VALVE
(b)
FIG. 5-16. Hydraulic tank energy storage systems, (a) Electrical output, (b) Hydraulic
output.
same.
116
117
normal motor driving system, and the other is the scramming drive,
This system is most suitable for use with small light rods.
0,.^..
SCRAM
TWO WAY
NORMAL
SPRING
SLOW
SPEED
MOTOR
DRIVE
SYSTEM
FLY WHEEL
ELECTRIC
MOTOR
ELECTRIC
GENERATOR
CONTROL
Generated on 2011-10-24 11:47 GMT / Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google
SWITCHING
ROD
DRIVE
ELECTRIC
MOTOR
(a)
FLY WHEEL
ELECTRIC
MOTOR
HYDRAULIC
PUMP
(b)
CONTROL
VALVES
ROD
DRIVE
HYDRAULIC
MOTOR
FIG. 5-18.
output.
Flyw
heel
in'er
used to supply the power for a rod drive motor or scramming system.
118 CONTROL OF NUCLEAR REACTORS AND POWER PLANTS
If the charging motor power fails, the inertia of the flywheel will keep the
Figure 5-186 shows the hydraulic analogy of the same system. Both
of these inertia systems are such that there is a finite time after power
either the rods are inserted within 5 min after power failure or the fly-
failure are used. But if manual valving is employed, the gravity storage
rods, or each rod may have its own storage system. The gravity or
whereas the flywheel system can be placed very simply on each rod drive.
a failure in the system could affect not one rod but all of them. For a
given amount of energy storage, it can usually be shown that the flywheel
system occupies less space and has less weight than the accumulator.
The gravity system may be somewhat more reliable basically in that the
trouble. The flywheel drive also will take more power from the supply
lines, but this is generally of small consequence. In any event, the choice
of one energy storage form over another will depend upon space and
ply is desired.
For a vertical rod drive system, simple or accelerated gravity drop upon
scramming drives means that control rods are inserted into a reactor with
considerable force. After the rod has reached the end of its useful travel,
reactor nor its structure will be damaged. Spjrings and dashpot arrange-
where each rod is at all times in order to anticipate the state of criticality
119
indication.
ROD RELEASE
MAGNET
SPRING AND
DASHPOT
CONTROL ROD
motor. Figure 5-8 indicates a drive motor having the rotor of a position
pole pickup coil is used with this motor on the outside of the structure.
of the position of the control rod. In many cases the indication through
Generated on 2011-10-24 11:47 GMT / Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google
Generated on 2011-10-24 11:47 GMT / Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google
REACTOR CONTROL MECHANISMS
121
types indicated in Chap. 4 and that the reactor has no negative tempera-
ture coefficient. The curve of Fig. 5-21a indicates the shape of the tran-
sient response we would expect from our control system, with P being the
peak power excursion reached by the transient for an input reactivity step.
to assume that the reactivity change created by the rod would be of the
shape indicated in Fig. 5-216. If reactivity and rod position are linearly
POWER
LEVEL
REACTIVITY
CHANGE
OR
ROD
MOTION
(a)
TIME
- 1 CYCLE
related, the rod would start to move slowly, gather speed, and then slow
down as the desired end point was approached. For simplicity let us
Generated on 2011-10-24 11:48 GMT / Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google
assume that this required motion is sinusoidal about the halfway point
and that the sine wave has a peak half amplitude of A. The time taken
by the transient to die out effectively may be considered one cycle of the
rod motion sine waves.7 This method of analysis is not strictly correct.
sine wave derived in this manner and /n are of the same order of magni-
such that this sinusoidal motion of a given frequency / and a given ampli-
(5-1)
co = 2rr/
122 CONTROL OF NUCLEAR REACTORS AND POWER PLANTS
rod displacement
rod velocity
v = Aw cos t (5-3)
hp = Fv (5-5)
where the force on the rod F = ma, and if the rod mass = m, in poundals,
then
hp =
550
sin cot cos wt = Yt, sin 2tat, the peak horsepower required from the driv-
_ , , ,_ _.
Peak hp = - (5-7)
try a numerical example. Assume the transient of Fig. 4-32, and arbi-
trarily state that a complete sine wave cycle would take 0.15 sec or
= 418 hp
the constants of our example are impractical or we are asking for too much
by way of transient response. It can be seen from Eq. (5-7) that the
a far more manageable number. It will be recalled from Sec. 4-12 that
the slowest possible control system should be chosen for safety reasons.
The above argument has been presented on the basis of the reactor
the temperature coefficient was principally upon the front end of the tran-
sient, and the time for the disturbance to be returned to zero depended
REFERENCES
Pa.
4. Jones, M. F.: Launching Aircraft Electrically, Aviation, vol. 45, pp. 62-65, October,
1946.
9. Harrer, J. M.: Control Rod Mechanisms, Nucleonics, vol. 13, no. 6, p. 48, June,
1955.
Generated on 2011-10-24 11:48 GMT / Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google
CHAPTER 6
the neutrons from the reactor are being used for experimental purposes
it has been presumed that although the reactor is giving off heat, no use
is being made of this heat. This situation corresponds in our d-c gen-
up a reactor and attempt to obtain power from it, we shall also affect its
and control of power reactors from which useful heat power is being
extracted.
Fig. 6-1. Here the reactor is considered merely as a source of heat energy.
and the output shaft of the turbine can be made to drive many types of
this discussion, and this coolant may be gas, water, or a liquid metal.
Two external paths exist for affecting the reactor. One is a local path
Generated on 2011-10-24 11:48 GMT / Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google
path is the direct connection to the plant via the coolant system. Any
or flow will affect the reactivity of the reactor. Variations in the coolant
the moderator. It is the effect upon the moderator and fuel that usually
causes a reactivity change, and any external influence that changes mod-
124
NUCLEAR POWER PLANT CONTROL
125
tant safety and control element. The temperature of the moderator and
the fuel can be affected by the internal heat of the reactor, by external
plants where the coolant or moderator is either a gas or water, the primary
reactivity.
FIG. 6-1. Elementary block diagram of nuclear power plant showing essential elements.
These are the temperature coefficient as the |d5fc/dT|con8tp and the pres-
order to get a better feel for the large number of items that may affect
than the elementary one given in Fig. 6-1. Figure 6-2 is a block diagram
auxiliary systems may be used, and each of these can conceivably affect
Generated on 2011-10-24 11:48 GMT / Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google
to indicate the reactor power level and the reactor period at any instant.
CONTROL OF NUCLEAR REACTORS AND POWER PLANTS
This system can affect the reactor principally by giving erroneous read-
wrong power demand error signal and thus causes control-rod motion.
output and all the characteristics of the plant are determined_by tem-
ever of the coolant usually being permitted. Because this system fur-
I CONTROLS I
FIG. 6-2. Block diagram of a nuclear power plant indicating some of the important
auxiliary systems.
However, from a practical point of view it is obvious that the plant must
not be shut offevery time some minor control in a secondary loop maP
Generated on 2011-10-24 11:48 GMT / Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google
126
NUCLEAR POWER PLANT CONTROL 127
minimum the number of controls which can cause a scram. For most
reactor power level at a slower rate than scram rate or merely to ring an
shutdown.
directly upon the rate of coolant flow. Consequently, this flow is also
involved in reactions upon the reactor. The coolant flow may be fixed,
ule. As the power input to a pump varies with the cube of its output, it
can be seen that to operate the plant at low power outputs without cut-^f\'" .
ting back coolant flow usually results in a loss of efficiency. Pumping \_/
power can become greater than 5 percent of the plant power output if
precautions to change the flow are not taken. The pump controls there-
fore may become quite complex in that multiple units may be switched
on and off the line and variable flow schedules can be used in addition.
its elementary form might be a tank partially filled with coolant which,
This tank is connected to a high point in the primary coolant system via
a pipe. It can be seen that the coolant in the tank and in the pipe can
represent an inertia. The gas pressure above the coolant in the pres-
the connecting pipe, the system conditions are proper for a pressure oscil-
lation to occur between the pressurizer and thg_primary loop' This pres-
fully, the time delays may be such that the initial pressure oscillation
are used in modern steam practice will be used for the nuclear power
plant. The control problem for the nuclear plant becomes very similar
to the control problem of the steam plant. A great deal must be known
about the characteristics of the output load, however, before the reactions
back upon the reactor can be explored. In the case of an aircraft power
output.
CONTROL OF NUCLEAR REACTORS AND POWER PLANTS
output is called the program. Figure 6-3 indicates again a basic plant,
PUMP
FIG. 6-3. Elementary block diagram of nuclear power plant indicating the symbols used
F, = steam flow
T, = steam temperature
of the primary and secondary sides of the plant. The difference between
the average temperature of the primary coolant in the boiler and the~
P=
- T.)
(6-1)
slightly with coolant flow rate, thermal power, and boiler level but may
tables when the steam pressure for 100 percent power is specified. As
verges to zero and the value at which the temperatures converge may be
128
NUCLEAR POWER PLANT CONTROL
129
once the constants of the primary side of the plant are known, the second-
ary constants are automatically specified and vice versa. For example,
Eq. (6-1) states that if one desires a plant in which the average tempera-
ture of the primary coolant is constant with power level, then the tem-
sume a constant K.
external control.
POWER OUTPUT
Let us examine the simple plant shown in Fig. 6-3, making the follow-
ing assumptions. The steady-state program desired is the one of Fig. 6-4
the load. This greater loading causes more heat to be extracted from
the heat exchanger, and for a shortjjeriod of time, the heat capacity of the
heat exchanger and coolant can usually supply the additional load.
However, this extra energy extracted from the system requires that The
temperature of the coolant into the reactor must drop. If the reactor
tor will possess more reacfivTtyl If the reactor was initially in a critical
of the coolant rises, and more energy is then available from the reactor.
Finally, in the steady state the reactor returns to its critical condition
will be noted that without any control mechanism whatever, the reactor
CONTROL OF NUCLEAR REACTORS AND POWER PLANTS
system has stabilized itself about a given average temperature and auto-
power demand.
POWER OUTPUT
maximum level.
more load is required, the inlet temperature to the reactor would drop.
More reactivity would be inserted into trie reactor because of the negative
program.
130
measures th~e" temperature ^and then varies the cooTant pump speed in
131
it will be noted that the average coolant temperature must drop with
ture. It can be seen then that any program other than the constant-
principal one is the large change in steam pressure over the power range,
Fig. 6-7. The constant-T.v program is the one preferred by the reactor.
POWER OUTPUT
coolant flow.
CONTROL OF NUCLEAR REACTORS AND POWER PLANTS
132
because of the T.v change, and in some cases serious primary coolant
needed.
level to another, the plant must fight the tendency of the negative tem-
520
500
4RO
^^
la)
ATOV=
4 psi
±±
< 540
U.
CO
Generated on 2011-10-24 11:48 GMT / Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google
t! 520
ct
' 500
fr
H»
4801-
ATOV=46.3«F
Ap=0
I- 520
soo
20 40 60 80 100
8k CHANGE
TOTAL ATOV
there are two logical extremes to the programming, and on the surface
primary loop; the other by the secondary loop. In most practical plants
program which is part way between these two extremes. That is, a
O.
133
pressure would both be taken and no attempt made to hold either con-
stant. This sort of approach is shown in Fig. 6-8, using typical numbers
FIG. 6-10. Block diagram of plant and universal control system capable of setting up any
The method that is used at a given power level and at a given flow is to
ATav. A curve of this sort is shown in Fig. 6-9, and at any AT.v the
pressure variations.
view, however, we can quickly see that the proper speed of response to
any conceivable useful type of control program. To tie back to our pre-
134 CONTROL OF NUCLEAR REACTORS AND POWER PLANTS
vious case, let us assume that this system has been set up to demand the
tion that the demand signal no out of our universal control device into
p, T,, Tc, Th, or T.v, the TIO power demand signal cannot change from,
Now let us follow through some of the typical demands that might be
made on this system. Suppose we are operating at a low power level and
minutes causes the demand signal no to rise slowly and the nuclear loop
tor instantaneously from the demand loop because of the velocity limit
is reflected through the boiler into the primary loop as a change in aver-
age temperature of the coolant in the reactor, and the negative tempera-
appears into the comparator and the entire system slowly changes over
to a constant-pressure program.
and poisoning, the n loop signal, which is likely to be also velocity limited
by virtue of the fact that the control rods are not permitted to change
reactivity very fast, will handle this change slowly and safely. If a fast
takes over and the system behaves as though no external controls were
present.
isslowT EffecTfvely, then, one can have the advantages of both constant-
design.
is carried by the primary coolant from the reactor to the boiler must be
135
simple relationship
P = F(Th -
(6-2)
CONSTANT
FLOW
VARIABLE
FLOW
POWER OUTPUT
(O)
There may be reasons why Th must never exceed a certain value. Prob-
lems of stress, corrosion, and wear may all be involved. Also, as pre-
average-temperature system. In
plant and have come to the conclusion that the basic plant, consisting of
POWER OUTPUT
plant, but the approach to the determination of the stability for a prac-
tical plant will be indicated. The method we shall use will be to develop
transport delays in the loop, the transfer of energy in the boiler, and the
use of this energy in the steam system. Once the basic form of the equa-
as a basic control loop. Figure 6-12 indicates the loop and the symbols
we shall be examining.
assumption that the coolant and the moderator of our reactor are one
BASIC
CONTROL
L00P
PUMP
BOILER FEED
WATER PUMP
and the same. In this way we shall eliminate double heat transfer
system with either gas or water as the moderator. For our thermo-
This total energy manifests itself in a rise in the temperature of the fuel.
ture difference between the fuel and the surrounding coolant causes heat
136
(6-4)
when we consider that we shall ultimately feed back into our reactor a
the heat flowing out of the fuel and being carried away by the coolant.
term of Eq. (6-3), and it is carried off into the flow of coolant as
at TO
Fig. 6-13
core
rri \ rp
r«=-^y- (6-6)
Introducing this factor, we can transform Eq. (6-5) into the form of
Generated on 2011-10-24 11:49 GMT / Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google
+ -2 + T = ft + Tc (6-7)
TO / at TO
Transport Delays. With Eqs. (6-4) and (6-7) we have now related
the reactor heat output to Tc, Th, and T^. We now must transport this
heat to the boiler via a pipe, as indicated in Fig. 6-12. We shall assume
that all our heat transfer occurs at a point in the reactor and ignore the
time delays in transporting the average heat from one position in the
reactor to another.
of the heat from the reactor to the boiler. The temperatures can be
given as
the reactor outlet temperature, only it attains a given value after a fixed
transport delay r3. We are assuming no heat loss in the pipe transporta-
imate the time delay by a differential equation form. We shall use both
For slow changes in temperature we can ignore all terms beyond the
equation to indicate the time delay between reactor outlet and boiler
inlet temperatures.
J/TT
dT
or by Tc + r4 ^f = Tbo (6-13)
Generated on 2011-10-24 11:49 GMT / Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google
where T^ is the outlet temperature of the boiler and r4 the transport delay
tt
dT
Boiler Equations. We can now use the same type of logic and approx-
imations in order to write the heat balance equations for the boiler.
140 CONTROL OF NUCLEAR REACTORS AND POWER PLANTS
On the steam side of the boiler we may express the fact that the rate
of heat storage in the boiler metal, steam, and water is equal to the differ-
ence between the rate at which heat is transferred across the boiler tubes
flT
Ka = a throttle constant
(6-17) and (6-18). First, the heat transfer between the primary coolant
side and the steam side of the boiler is represented by means of a single
of the boiler metal is lumped with that of the steam and water. Actually
Generated on 2011-10-24 11:49 GMT / Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google
there is a time delay associated with a transfer of heat between metal and
steam. Neglecting this time delay will change the shape of the initial
loop. The power output has been taken to be proportional only to the
flow of steam through the throttle, assuming that the enthalpy of the
saturated steam is constant. The flow of steam is then taken as the prod-
ity constant. The further assumption has been made that the time spent
with the time spent in the pipes and that the heat exchange in the boiler
the errors involved in these assumptions are usually less than the amount
141
develop the form of the transfer functions for each of the components in
the basic loop. We may first start with the reactor. It will be recalled
from Sec. 4-3 that the elementary reactor with a negative temperature
back path was fed a variation in neutron power, and from its output came
simple T.v feedback loop, there exists another loop containing both Tc
8k IN
(OUT OF PLANT)
Th
To
and TK. The situation is even more complicated because of the fact that
more sophisticated, but still approximate, transfer function for the reactor
thermal system. The equations of heat flow from the reactor to the
r, + 2is
Generated on 2011-10-24 11:49 GMT / Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google
TO
Tc
(6-19)
(6-20)
(6-21)
2 \T2 / 1 J
where
Factoring
and
7=
(6-22)
(6-23)
(6-24)
«)
(6-26)
Fig. 6-15. Here the over-all reactor thermal transfer function artificially
Q(s)
T,(s)
Th(s)
FIG. 6-15. Block diagram servo representation of a simple reactor thermal system.
appears to be in two parts, an inlet part and an outlet part, with the
reactor heat feeding in between them. The generation of Tav is, of course,
a separate external problem and its transfer function will later be obtained
farther along in the coolant flow path. Now a transfer equation of iden-
tical form to Eq. (6-26) can be derived for each individual section of the
reactor. The assumption is again made that the average heat transfer
heat is then transported from section to section, and the final outlet tem-
142
143
Qiji = 6272, and (?i2(s) represents the transport delay between the first
exceeds the additional utility of the more accurate results when too many
n(s)
Q,(s)
02(s)
GM(s)
G,2 (s)
Gc2(s)
Th(s)
FIG. 6-16. Block diagram servo representation of a two-section reactor thermal system.
in the primary loop of Fig. 6-12, and both of these have transfer functions
from the reactor to the boiler, and second, after the heat exchange process
has occurred, the coolant must be transferred back from the boiler to the
reactor.
Eq. (6-8)
Generated on 2011-10-24 11:49 GMT / Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google
= Th(t - T.)
rff\ rp (t _ \
rw(s) = Th(s)e~"'
J. cv«Jy J. oo \« /C
(6-27)
(6-28)
The transfer functions then become the output divided by the input or
(6-30)
Tc(s)
lost no heat in getting from the reactor outlet to the boiler inlet and fro.n
the boiler outlet to the reactor inlet. The phase lag increases linearly
are Eqs. (6-11) and (6-13), and the approximate transfer functions are at
once recognizable as
(6-31)
Th(s) I + r3s
and -ff. 7 r -~ ^ :
The heat-flow equations for the boiler were Eqs. (6-16), (6-18), and
dT
„ TM , ,
1 + rtos
+i+
where KB = - ( bb - - - (6-37)
K.AB \_
KbAb V
TU = F —— =i (o-3o)
Kb b [ ~K^ + (2Re/KbAt) -
145
Tbo =
B=
Msc, + Mmcm
KaAB
dTssat
(2Rc/KbAb)
at corresponding value of Ts
(6-39)
(6-40)
1.0
O.fl
. 0.6
J0.4
0.2
Generated on 2011-10-24 11:49 GMT / Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google
0.1
K0AB
100
is presented for a given boiler with heat transfer characteristics such that
2Rc/KbAb = 4. It can easily be seen from this illustration that the boiler
does not provide much fixed attenuation even for the heaviest loads.
(2Rc/KbAb) - 1
(2Rc/KbAb)
(6-41)
Basic Plant. The general dynamic requirements of the plant are con-
cerned with the stability of the complete system and the rate and mag-
if power level can be changed only slowly, whereas a plant used to propel
bandwidth of the control system and by the thermal capacity of the pri-
mary coolant. The coolant acts as a reservoir of energy which can absorb
tively fast changes in output demand. Thus reactor power may be per-
mitted to lag behind the power delivered to the load, thereby relieving
over until the temperatures have been reset to the values specified by the
ble for studying the natural stability of the plant and for synthesis of a
value. The plant variables of interest are those which are fed back
through the inherent feedback paths of the plant and those which are
is valid only for amplitudes of power excursions that are small compared
with the average value. The analysis may be repeated for a sequence of
range of interest.
rate of loss, which in turn depends on the geometry of the reactor and on
reactor power, the plant contains inherent feedback, as shown in Fig. 6-18.
response calculations.
147
REACTIVITY
KTC.KP AND KX
DENOTE
COEFFICIENTS
OF REACTIVITY
CONCENTRATION
TEMP. COEF.
REACTIVITY
FEEDBACK
I LOOP
/__SISJE^_J I
v/
\ TEMPERATURE
FEEDBACK LOOP
FIG. 6-19. Basic plant control loop showing temperature coefficient reactivity feedback
block diagram of the basic plant control loop that will be analyzed and
Our basic control loop consists essentially of two loops: first, the tem-
loop is the new portion of our study and is most important to stability
at Th is fed around through the boiler and back into the reactor as an
is at the boiler, and at low power levels, as just shown, there is only a very
reactor thermal system and in the boiler, however, contain some attenua-
the plant. Figure 6-20 shows the temperature feedback loop with all
148
have derived and have been using throughout this text have been in Laplace-transform
NUCLEAR POWER PLANT CONTROL U9
n(s) 1 - KLGL(s)
(6-42)
The minus sign indicates that the feedback is regenerative. For example,
^>
n(a) l-KLGL(s)
The term 1/[1 — KLGL(s)] appears very prominently in all stability
compute 1/[1 — KLGL(s')] for a fictitious reactor plant, using simple num-
bers. Referring to Fig. 6-20, the following constants will be chosen for
£ - 4 g - 3
For the outlet portion of the reactor thermal transfer function let
where
KB \ j"â„¢8 = 0.794 \
1 + TboS I
and KB is chosen as
0.794 = -2db
We therefore should be substituting for s, jZ-n-f or ju. Actually, the symbol s has no
meaning where we are performing nonlinear operations and is used in these following
sections only to simplify notation. For all the operations involving frequency ju
is implied.
CONTROL OF NUCLEAR REACTORS AND POWER PLANTS
150
12
a4
uf
§
UJ-4
!r-8
-12
-16
Hi-//
^U \ , . \....
COMBINED -
AMPLITUDE
- FUNCTION -
0.001
1OO
0.01 0.1 1 10
FIG. 6-21. Relative amplitude response of transfer function of temperature feedback loop
100
50
UJ
t -50
^-100
a.
-150
-200
SHIFT WITHOUT
TRANSPORT DELAYS
0.001
100
0.01 0.1 1 10
CO
FIG. 6-22. Phase shift of transfer function of temperature feedback loop Kr.Gds) showing
individual phase shifts but omitting linear phase shift of transport delay.
NUCLEAR POWER PLANT CONTROL 151
Using these constants, Fig. 6-21 shows the form of the relative ampli-
down on the plot. It will be noted from this figure that as the frequency
expected for most physical systems. In this case the amplitude levels
0.001
0.01
0.1 1
100
Figure 6-22 indicates the phase shift in the system without the trans-
port delays of the pipe. It can be seen that the phase of this remainder
of the system settles out at 180° phase shift at high frequencies. The
transport delay phase shift that occurs is linear with frequency and will
Figure 6-23 shows a plot of the amplitude and phase of 1/[1 — KLGL(S)]
for our example. The cyclic nature of the loop transfer function is due
to the linear time delays causing the phase angle of KI,GL(S) to increase
Generated on 2011-10-24 11:50 GMT / Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google
steadily with frequency. The peaks and valleys in the plot of the ampli-
tude of 1/[1 — KL,GL(S)] occur at frequency intervals which are roughly
multiples of l/2rt, where n is the total time delay taken by the coolant in
going around the primary loop. Physically speaking, the peaks and
around the loop either reinforces or opposes the temperature signals intro-
Generated on 2011-10-24 11:50 GMT / Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google
NUCLEAR POWER PLANT CONTROL
153
TEMPERATURE T0 •
VOLUME
of the two pipe delays than the pure time delay formula we have been
some mixing at the inlet to the reactor and the boiler and most likely
components.
tion.
or
F(?'0 + A7Y) = F At
AT,
At
"V*
i + (V - F At)T0
F_
T0
dT0 , F „. F
In Laplace notation
TVs) _ F/V
(6-47)
(6-48)
(6-49)
(6-50)
(6-51)
As V has the dimensions of cubic feet and F is in cubic feet per second,
CONTROL OF NUCLEAR REACTORS AND POWER PLANTS
FIG. 6-26. (a) Symbolic representation of mixing in inlet and outlet of reactor and boiler,
(b) Block diagram of temperature feedback loop indicating placement of mixing transfer
functions.
'bo
BOILER
MIXING
FIG. 6-27. Transfer function representation of a temperature feedback loop with two
mixing terms.
r.(«) _ i
IXs) 1 + rs
(6-52)
\S4
point of view we can now insert this type of mixing transfer function into
the temperature feedback loop. Figure 6-26 shows one way this might
Generated on 2011-10-24 11:50 GMT / Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google
NUCLEAR POWER PLANT CONTROL
-4
-^^n.
x\^
1 1 1 1 1 III
. -i — r T-[ i in
i .,.,, nn
- """^
\\
\\
g -12
~100 at
cc
_J
Generated on 2011-10-24 11:50 GMT / Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google
-150 §
LJ
1-20
i \-r- PH
«E SHIFT
WITHOUT
PORT DELAYS
-200 1.
bj
AMPLITUDE—
\ \ TRANS
\\
-24
\\
-28
i1ii11
\\
-300
..iiiiii
! 1 1 LM
-35O
0.001
100
0.01 0.1 1 10
FIG. 6-28. Transfer function example of temperature feedback open-loop gain KtG/,(s)
with mixing.
5n
— R^
1 — 1 1 1 1 1 11
1 — r-mnr
0 10
i5
^V
Ke= 0.794 /
\/
A/W-
3 °
CONTROL OF NUCLEAR REACTORS AND POWER PLANTS
We can now return to our previous example, using the loop block
diagram of Fig. 6-27, with the constants assigned from Eqs. (6-44), (6-45),
and (6-46), and denote simple values to r6 = T6 = (1/ir) sec (break fre-
quency 0.5 cycle/sec). We can again carry through the analysis for
1/[1 — KLGL(S)], first obtaining the amplitude and phase response of
KL.GL(S) shown in Fig. 6-28. Now the open-loop gain decreases satis-
factorily toward zero with frequency. Figure 6-29 shows the amplitude
and phase response of 1/[1 — KI.GL(S)] for our simple example, and now
the periodic oscillation has been damped out to where the function is
plausible.
10
^' '""
^\
iii11111
1 ,11
Iiiiii!t
iii.
\/
A AAAn,
POWER LEVEL P
Generated on 2011-10-24 11:50 GMT / Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google
I ~5
2-io
*J
VZ
\JV
FLOW
I 15
i 10
15
-5
-^^
^\
V_X v
/\A
FLOW F/2
-10
0.(
1'!11111
tiiti11
FIG. 6-30. Amplitude of 1/[1 — Ki,Gz,(s)] as the flow and power level are changed.
ing the case where the flow of coolant has been fixed at a given value and
the power output of the plant has also been fixed. Let us assume that
ing the coolant flow. For illustrative purposes let us cut the flow in half,
the magnitude and phase of 1/[1 — KLGL(S}] of our simple example, it
can be seen that if the frequency scale values of both the magnitude and
phase were divided by 2, the resulting curves would very closely approx-
156
given power and a given flow as against one-half of that power and one-
half of the initial flow. It can be seen that the curves are similar, as it is
NUCLEAR POWER PLANT CONTROL 157
apparent that all of the linear time delays in the primary coolant loop are
the loop is roughly equal to that of the boiler gain KB, which from Fig.
6-18 does not change very much for a power output change of 2 to 1.
amount changes for the most part the frequency scale only of the transfer
function. The general form of 1/[1 — KLGL(s)] remains the same.
It should be emphasized again that this analysis is proper only for fixed
boiling reactors, can be handled with the same general type of approach,
the form of the reactor thermal transfer function when the reactor was
the present has been for a simple one-section reactor. Before we proceed
with the study of the reactivity feedback loop, it is of interest to point out
the method of analysis that would be used for a more complicated reactor
thermal system, the reactor is split up into sections. Figure 6-31 shows
broken up into four series sections. The other elements in the loop are
the same as indicated in Fig. 6-27. The neutron flux from the reactor
may be assumed to produce equal heat in each of the four sections of the
manner. Let us assume that Qi is finite and Q2, Qa, and Qt = 0. The
loop T(s) is simply the product of all the transfer functions between
n(s) and T(s) in the direction of signal flow. As indicated previously, the
feedback loop transfer function KLGL(s) is the product of all the series
n(s)
a loop, Q,.Q,,Q,=0
Q!,Qj.Qi=0
T(s)
n(s)
[T(s)/n(s)l
(6-53)
(6-54)
assuming we are dealing with linear transfer functions, we may apply the
6
O -I
158
Generated on 2011-10-24 11:50 GMT / Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google
NUCLEAR POWER PLANT CONTROL
\S9
n(s)
T(s)
n(s)
Q-.Qt.Q4-*0
T(s)
n(s)
T(s)
T(s) 1 [
T(s)
T(s)
n(s)
T(s)
T(s)
open loop<
'. Ol.O,.Ot-oJ
(6-55)
derived, we can return our attention to the remainder of the basic con-
trol loop (see Fig. 6-19). This is the portion of the loop whereby the
Sk = (rC)(Tar - To)
(6-56)
and To
is a reference temperature.
s again is to represent j'co and the entire operation is for small signal varia-
TVs)
n(s)
Gh(s)
and
and
Therefore 2
or
L - KLGL(s)
T.(s) Th(s)KLGL(s)
n(s)
- KLGL(s)
n(s)
'
'
Gc(s)
(6-57)
(6-58)
(6-59)
(6-60)
(6-61)
(6-62)
ITVs)
2 n(s)
CONTROL OF NUCLEAR REACTORS AND POWER PLANTS
Sec. 6-10.
10
— J^J , ,.
^\
X. / 1 /I AA A (.
II!..,
iii.
I.,!.
1IiI1
•o 0
l\T!
\^ j M
.AMPLITUDE
UJ -10
50
Generated on 2011-10-24 11:51 GMT / Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google
|-20
<
/\
/\
/An
/ at
-30
O.C
^\^_
'^
X-- PHASE
v, ,
-100
-150
)0
1O
iiii11M
I1I1111
1 J i 1 1 11 1
1 1 t .1 1 i i 1
"S^^\
^o
-50
Ct
'. -10
a?
g -2°
vS /
/In
NUCLEAR POWER PLANT CONTROL
161
then be obtained from Eq. (6-58) and is shown in Fig. 6-33. Here the
We might now examine the over-all situation, again restating the pur-
10
^xT
1iil
n,
JU
Generated on 2011-10-24 11:51 GMT / Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google
/I/W-
ii.111
i i in
JD
"° 10
\/
\r
^XsX' AMPLIT
JDE
CT^S
\y
uj -20
i-
_j
s 30
<
-40
^\
^x
I\
~~i
tr
I «
fln
1 mn.
yll K
5,^ PHASE
"X^\
50
-50
162 CONTROL OF NUCLEAR REACTORS AND POWER PLANTS
There the signal may again reinforce or oppose the temperature signal of
the initial point, depending upon the phase shift acquired in passing from
one point to another. The linear time delays in the primary coolant
6-32, 6-33, and 6-34 can provide valuable insight into the behavior of the
nuclear power plant. In the first place, they may be helpful in obtaining
a "feel" for the transient response of the nuclear power plant to external
of this data provide a measure of the degree to which the bare reactor is
the transfer functions provide basic data upon which to design a satis-
factory control scheme. Finally, they may help to point out adjustments
are based.
of frequency which have been presented in this analysis have been based
throttle opening was kept constant, and second, the primary coolant flow
rate was constant. The results obtained are confined solely to relating
The response to changes in throttle setting or coolant flow rate have not
tions specified.
pressure is negative and the net effect is to make the reactor self-regu-
tem. The actual form of the negative feedback is not derivable from the
NUCLEAR POWER PLANT CONTROL
143
of the physical plant associated with the reactor. That is, an increase in
plant, is a control system which is already built into the plant and which
COOLANT
FLOW RATE
CONTROL
ROD
8k
?\ 8k
"
^'^
PRIM
Generated on 2011-10-24 11:51 GMT / Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google
'
BOILER
LOOP
<1
ruRBINE
TEMP. ( KTGT(S) ,
COEF. \ J
T,
TEMPERATURE
COEFFICIENT
Tov
FIG. 6-35. Block diagram of nuclear power plant indicating plant temperature coefficient
feedback loop.
never out of order. Any external control system that is desired must be
before designing an external reactor control system, one must first exam-
ine the relative stability of the reactor plant system before any controls
are added.
ties back around the reactor through the plant is given in Fig. 6-35.
From this figure we see that in order to determine the stability of the
complete reactor plant system about a given nuclear power level no with
transfer function
n(s) 3P.v(«)
= (TO
(6-63)
Sk(s) n(s)
164 CONTROL OF NUCLEAR REACTORS AND POWER PLANTS
where n(s)/8k(s) is the bare reactor transfer function as has been devel-
added if desired. It will be recalled from Sec. 3-4 that the reactor itself
n0, the bare reactor was represented by the transfer function of Eq. (3-41)
(6-64)
Plots of the magnitude and phase of the bare reactor transfer function
were presented as Figs. 3-15 and 3-16. The open-loop transfer function
of Eq. (6-63) is thus the product of the temperature coefficient (TC), the
transfer function for T^(s)/n(s), and the bare reactor transfer function.
using the curves of Figs. 3-15 and 3-16, I* = 10~4 sec for the bare reactor
transfer function and thus compute XrGr(s) for the example. Figures
6-36 and 6-37 indicate the magnitude and phase of KTGT(s)/(TC) thus
obtained. From these curves it is easily seen that the system of our
practical reactors might range between zero and 10~3Sfc/°F. The ampli-
tude curve of Fig. 6-36 is at +15 db when the phase shift reaches —180°
curve of Fig. 6-36 in order to obtain the total gain of the over-all tem-
Generated on 2011-10-24 11:51 GMT / Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google
cient were smaller, this margin would be still larger and the stability of
The equation for the reactor feedback loop gain, Eq. (6-63), can be
rewritten as
Our analysis has been at a constant power level whereby KB = 0.8 and
eters are changed. Assume that the power level is reduced. The reactor
power level is dropped, the gain portion of the transfer function depending
upon the reactor will be reduced in a linear manner. The gain of KLG^s)
NUCLEAR POWER PUNT CONTROL
165
100
50
-50
_,- -100
-150
-200
-250
0.001
0.01
10
100
0.1 1
FIG. 6-37. Open-loop phase shift of transfer function of temperature coefficient reactivity
with Eq. (6-37), so that as the power level is dropped, the gain of KLGL(s)
goes up from this cause. This effect is very noticeable because the pri-
described. The term 1/[1 — KLGL(s)} is very dependent upon KB. For
KaAB/Kt,Ab = 0.5. That is, the boiler gain is 0.8 at zero frequency at
this power level. Therefore the value of 1/[1 — KI£TL(S)] = 5 at zero
cent and 1/[1 — KLGL(s)\ = 50. In other words, a portion of the over-
all loop transfer function gain is reduced with power level whereas another
The above reasoning was based upon constant coolant flow. If the
coolant flow is changed at the same time that the power level is changed,
there are two more effects which can result. The gain constant 7, which
becomes larger inversely with flow. KB also changes slightly with flow,
power level rather than at full power output if the flow is reduced at low
power level. If the system can then be shown to be stable at its minimum
usable output, it usually can be assumed, for simple systems, that this is
a worse case and the system is stable a priori at full output. For com-
plex systems in which intricate flow patterns are called for in the pro-
Generated on 2011-10-24 11:51 GMT / Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google
granted.
higher, but now the transient response should be poorer. What we have
ture coefficient controls the feedback factor. With a very small tempera-
tion is concerned, but a transient injected into the system would create
becomes larger, the system will still possess good stability, but now tran-
transients, and the maximum value must not approach the sustained
oscillation conditions.
oscillation is caused when the gain of the system is greater than unity at
ISO^phase shut. Now, in the case of a bare reactor or for a reactor with
the loop of Eq. (6-65) stable. Its performance, with respect to transient
response, would still be very poor in that large excursions, but not infinite
The transient analysis of even a simple system such as has been used
reactor and on how many mixing terms were used. The obvious solution
in the example we have used have quite simple analogues with the excep-
tion of the linear time delays. These delays must be represented quite
is tcTexamine Lhe response of the Teactor power output, the thermal plant
these two types of input disturbance. As the computer set up for such a
plant system with all its auxiliary components, we shall merely outline
CONTROL OF NUCLEAR REACTORS AND POWER PLANTS
Consider the plant of Fig. 6-38, which will be of the same type that we
energy stored in the heat capacity of the boiler and coolant. The
THROTTLE
COOLANT
CIRCULATOR
BOILER FEED
WATER PUMP
temperature, since more heat is extracted from the coolant. The change
at once, and the reactor power starts to rise almost immediately. The
view, and its general temperature pattern is similar to the pattern of the
Generated on 2011-10-24 11:51 GMT / Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google
has passed, the average coolant temperature comes back to its initial
value, but now the power delivered to the turbine is the greater required
output, and this overshoot serves the purpose of recharging the stored
168
Becaxise of this fact, it would appear possible to take a plant of this type
NUCLEAR POWER PLANT CONTROL
169
control system.
shown in Sec. 3-2 that when a control rod is suddenly withdrawn, the
ill
TIME
opening.
TIME
dn Sk
Thus, in a very short amount of time a bare reactor would quickly reach
a very high power level. However, what actually happens in the case
power level gets very high, the heat flow through the reactor fuel to the
coolant raises the coolant temperature enough to stop the rapid rise.
Figure 6-40 shows the effect of a step change in reactivity upon the sys-
tem. The rising coolant temperature extracts reactivity and brings the
reactor power level down. A small secondary peak may exist because
CONTROL Of NUCLEAR REACTORS AND POWER PLANTS
power. The reactor fuel acts as a filter between the reactor power and
the reactivity change. A transient heat spike can be seen in the fuel
therefore, the thermal power follow the transient much more slowly.
After the transient has passed, the reactor coolant temperature must
caused by the rod motion. This condition causes all the temperatures
of the plant to rise, and as a consequence, the thermal output power and
the reactor power will rise somewhat above their original values. A com-
pletely similar, but opposite, set of curves results when negative throttle
FIG. 6-41. Block diagram indicating the feedback loops which affect reactor stability.
examine a combined reactor and plant to see how the reactor responds
was handled as a simple local feedback loop around the reactor and the
feedback gain was combined with the reactor gain by means of conven-
tional servo analysis for two elements in parallel. Then when a com-
Generated on 2011-10-24 11:51 GMT / Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google
external control loop could be tied around the combined system. Figure
6-41 indicates how a plant can be set up to cause an external reactor con-
170
171
tem indicated by the dotted box of Fig. 6-41. The reactor and its feed-
back loops again may be linearized for small amplitude sinusoidal varia-
tions at an average power level no. These transfer functions are indicated
in the diagram of Fig. 6-42a. The system may then be reduced to the
equivalent form of Fig. 6-426, in which the reactor and its temperature
REACTOR
^rodl+A
^6K^
n(s)
6k (s)
„ "(=).
1 +.
1
Generated on 2011-10-24 11:51 GMT / Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google
1_
bK P
TEMP.
COEF
EXTERN
LAIN REACTOR
EMP. FEEDBACK
, ToV(s)
TC n(s)
AL CONTROL
FEEDBACK
SYSTEM
KBGB(s)
(a)
FIG. 6-42. Equivalent block diagrams of the reactor with temperature coefficient and
n(s)/<5fc(s)
(TC)[T\v(S)/n(s)] 1 +
(6-67)
(6-68)
the more simple problem of reactor control as analyzed in Secs. 4-7 and
4-8. The bare reactor had a given transfer function, and it was shown
system or with an on-off step type control system. When a simple reac-
flow, was examined, it was shown that all that was needed was to modify
similar manner. Then a new combined reactor function with all these
plant we have now shown that the bare reactor transfer function need
SYSTEM
COMBINATION
CONTROL
6k (s)
n(s)
la)
5k(s)
KRGR(s)
n(s)
ARE REACTOR
KTC GTC
LOCAL
TEMP. COEF.
FEEDBACK
(b)
6k (S)
n(s)
BARE REACTOR
Generated on 2011-10-24 11:51 GMT / Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google
PLANT TEMP.
FEEDBACK
(c)
REACTOR
CONTROL SYSTEM
FIG. 6-43. Combinations of reactor and various feedback loop transfer functions, (a)
Elementary reactor, (b) Local temperature coefficient feedback, (c) Plant temperature
coefficient feedback.
Figure 6-43a shows a bare reactor and then an external reactor control
system transfer function of any type KB(?B(S). Figure 6-436 shows the
present. The poison feedback loop again is not shown. It can easily
be seen that the analysis for any external reactor control system can
172
173
For illustrative purposes we can plot Eq. (6-68) for the example we have
6-44 indicates this plot of the combined reactor and plant temperature coef-
tion is of the same general form as the combined reactor, local temperature
coefficient feedback system of Figs. 4-4 and 4-6. In particular, the trans-
phase shift approaches zero at zero frequency and —90° at infinite fre-
quency. Thus we see again that the example of the reactor plant nega-
59
7- 50
—
? J 45
I" 40
LU
S 35
5 30
i 25
25
C1-
In -50
Generated on 2011-10-24 11:52 GMT / Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google
I -75
-100
^APPROACHES
0.001
APPROACHES 90°-:
0.01
10
100
0.1 1
FIG. 6-44. Magnitude and phase of over-all reactor transfer function with temperature
from the plant. Reactor control then, at most, becomes a minor auxiliary
We have determined that the reactor plant and reactor control could be
loop upon the reactor and the temperature coefficient feedback loop.
Two different situations can arise. These will depend on whether the
plant by itself has a large amount of natural stability or has poor transient
must control the reactor power level according to the output demanded,
ments suggest a control system of the type shown in Fig. 6-46. Reactor
of plant variables and must be a measure of the load on the plant and the
gram. Usually the steam flow is a good indication of the load require-
Generated on 2011-10-24 11:52 GMT / Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google
to the steam flow. Throttle position also may be used. TEe particular
174
NUCLEAR POWER PLANT CONTROL
175
variable used to indicate deviation from the set program might be any
been decided upon, the stability of the complete system may be studied
earity. Consequently, the results are strictly valid only for small devia-
4»
MEASURED ERRO
REACTOR SIGN/!
POWER
SIGNAL
L REACTIVITY g|<
MECHANISMS ^
REACTOR
KEACTOR POWER ..
PLANT
j
Generated on 2011-10-24 11:52 GMT / Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google
n0
INHERENT
POWER
DEMAND
FEEDBACK
GENERATING CIRCUITS
MEASURED
CONTROL
VARIABLES
FIG. 6-46. Block diagram of control system for a plant containing inadequate natural
stability.
If the analysis of the basic plant, such as has been carried out in the
by the temperature change then alters reactor power, thus resetting the
many cases automatic control of the power level for improved transient
sary. Thus the role of the automatic system is reduced to that of main-
gest a control system of the type shown in Fig. 6-47. Here the tempera-
level until the net reactivity is again reduced to zero. Thus by moving
the control rods a given distance, the temperatures shift to a new con-
ple, the over-all system stability must again be studied and optimized by
PROGRAM
REFERENCE
FIG. 6-47. Block diagram of control system for a plant containing adequate natural
stability.
from a plant. Obviously, each program can call for a completely differ-
ent type of control, but there are recognizable patterns of control systems
which can fit frequently occurring situations. The most obvious type
in Sec. 6-3, this is the natural program of the reactor. In addition, there
We shall now examine a few types of control system setups for the con-
176
Figure 6-48 indicates two types of control systems. Figure 6-48a rep-
=K
(6-69)
power, assumed proportional to the neutron flux level, and the resultant
error signal actuates the control rods. The integral of !F.v error is neces-
COMPARATOR
?&[>— 0*--
t Tn
-T~L Tovref
l>
II
ROD
Lhj '
rVc
LOAD
REACTOR |*
BOILER
<
Generated on 2011-10-24 11:52 GMT / Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google
(o)
COMPARATOR
DRIVE
f~
IS
T-
BOILER
LOAD
n0=K1Fs+K2[(ToVre(-Tov) + i (TOVref-Tov)dt]
(t»
more power from the primary coolant and lowering the average coolant
ant. The sooner reactor power equals the steam demand, the shorter
the duration of the transient. The control system of Fig. 6-48a requires
that an error in Tav exist for a length of time following a throttle change,
in order to furnish the necessary change in demand signal n0) since the
the steady state. In other words, this system behaves pretty much in the
CONTROL OF NUCLEAR REACTORS AND POWER PLANTS
same manner as would the natural plant stability by itself. The Tm,
signal ensures that the long-term depletion and poisoning changes are
wiped out.
Figure 6-486 shows a control scheme that does not have the limitations
just described. In this system we again assume that the enthalpy of the
Since power at any level is then the product of steam flow and enthalpy,
(Tmr.rtf -
dt
(6-70)
Here the demand signal is nearly directly proportional to steam flow and
COMPARATOR
COMPARATOR
n0=K,Fs+K[(pref-p)+l^pref-p)df]
(b)
FIG. 6-49. Two types of control systems for constant-steam-pressure program, (a)
178
is adjusted much more rapidly than in the system of Fig. 6-48a, where
179
very close to unity; that is, Ki = 1 percent reactor power demand per
1 percent steam flow. The scheme of Fig. 6-486 may be called the direct
steam pressure program, and these are indicated in Figs. 6-49a and 6-496.
Figure 6-49a employs the pressure error to generate the power demand
was the Tav error. This system has the same inherent limitations as
COMPARATOR COMPUTER
DRIVE
ROD
In > 1
Pref IF,
^L
^* 1 — 1
LOAD
<
1
Generated on 2011-10-24 11:52 GMT / Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google
(3
REACTOR
=J--J
BOILER
demand, plus a function of the pressure error to reset the power demand
signal.
same as that of Fig. 6-496, except now the pressure reference is not con-
is desired from this system, the pressure reference can be tapered to drop
universal control scheme of this type, any program can be set up by meas-
has been determined that the natural stability of a plant by itself will be
The first system that comes to mind is the very elementary one of
manual control. Figure 6-51 shows two possible manual control schemes,
one for holding the average temperature constant and the other for hold-
ing steam pressure constant. The human link in the system is specified
reading. The plant takes care of itself through its own natural stability,
ROD
DRIVE
Q- O
ROD
MANUAL
CONTROL
REACTOR
'INDICATOR
BOILER
(a)
LOAD
ROD
DRIVE
-o
Generated on 2011-10-24 11:52 GMT / Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google
(/*]
-Pref
MANUAL PRESSURE
CONTROL ERROR
INDICATOR
those of Figs. 6-52 and 6-53 may be used. Here a temperature or pres-
sure error is made to actuate control rods directly and thus take care of
that is present in the system is used only for such information as giving
alarms and is not tied back into the control. Both systems of Figs. 6-52
and 6-53 are slow shimming operations and may be used either with
180
NUCLEAR POWER PLANT CONTROL
181
TEMPERATURE
ERROR
Th REFERENCE
FIG. 6-52. Automatic control system for a plant with sufficient inherent natural stability
TAPERED PRESSURE
REFERENCE
SOME FUNCTION
OF F,
Fs MEASURED
STEAM FLOW
(b)
FIG. 6-53. Automatic control system for a plant with sufficient inherent natural stability,
(a) Primary control variable, steam pressure, (b) Primary control variable, steam flow.
6-16. Stability Analysis for Demand Loop. The analysis of any of the
plant control systems just presented should again be in two parts, that is,
problem.
As before, two variations of the problem exist in that there are two
at this point the problem becomes one of synthesizing the demand loop.
Frequency Analysis for Plant with Reactor Control Loop. Let us first
FIG. 6-54. Power demand control loop for a plant with insufficient natural stability.
reactor control loop is specified. The power demand loop for this con-
dition is shown in general form in Fig. 6-54. The stability analysis for
this loop is outlined in the following manner. Let us break the loop at
some convenient place such as between the power demand signal generat-
ing circuit and the comparator at the point marked X on Fig. 6-54. We
can now examine the power demand open-loop gain and phase shift by
The over-all loop gain is then examined for phase margin and gain mar-
Open-loop gain =
n(s)
Generated on 2011-10-24 11:52 GMT / Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google
n(s)
(6-71)
182
transfer function and n0(s)/n(s) is the plant portion of the power demand
NUCLEAR POWER PLANT CONTROL 183
tor control loop around the combined reactor plant temperature coeffi-
n0(s) l+KnGn(s)
type, may have its gain of the form (V/b)Ab/a(\x\/b), where the output
linearities may be handled has been pointed out in Chap. 4, and KnGn(s)
That is, the power demand signal first consists of some term generally
term which will be used to set up the plant program around some variable
can be found as
For a more specific example let us assume that the plant operating
The following equation might then apply for a given control system:
Generated on 2011-10-24 11:52 GMT / Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google
As all the other terms of this equation are essentially fixed by the prin-
cipal components of the plant, the synthesis of the demand loop would
Two details must be mentioned at this point. The first is that, al-
The method used to determine Fe(s)/n(s) is taken directly from the dif-
1 84 CONTROL OF NUCLEAR REACTORS AND POWER PLANTS
- T.) - KaAp
(Mmcm + M,c.) ^
F, = (BKaA)T,
(6-77)
(6-78)
where the symbols are as defined in Sec. 6-4 and the constants are greatly
gain is the fact that the information on plant conditions is not available
TC
COMPENSATION
IF NEEDED
DEMAND \
R P LOOP .
CONTROL
AMPLIFIER
LINEAR OR
NON LINEAR
REACTIVITY
ACTUATOR
Generated on 2011-10-24 11:52 GMT / Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google
MECHANISMS
REACTOR
COMBINED WITH
PLANT TC
FEEDBACK
DEMAND SIC
GENERAâ„¢
CIRCUITS A
COM PA RAT
NAL uc.
G T0 OR ME
- X s,6
ASUREMENT
NAL DELAYS
MEASURED
CONTROL
VARIABLES
PLANT
Tref
OR Pref
FIG. 6-55. Demand control loop for a plant with sufficient natural stability.
This time constant can have a very serious effect on the stability of the
demand loop.
this second type of control system we may examine the block diagram of
Fig. 6-55. Here the separate reactor control loop is eliminated and a
pressure reference (linear or nonlinear with power output) and the actual
NUCLEAR POWER PLANT CONTROL 185
(ps) n(s)
T = Tref - To (6-81)
leading to transfer functions for the plant portion of the loop of the form
PM KlF_M + K^ (6.83)
It must be pointed out again that all analyses such as have been indi-
most of the parameters being used are a function of level or flow or both.
tion being placed on the reactor system for stability. This limitation
It can be shown that below this speed the system will be stable. It is
rod reactivity change rates from a reactor safety point of view. Some
of these requirements may call for the control rods moving at a faster
system design for power range operation is going to depend upon finding
186 CONTROL OF NUCLEAR REACTORS AND POWER PLANTS
tional limitations.
REFERENCES
2. Schultz, M. A.: Temperature Programs and Control Systems for a Nuclear Power
5. Grace, J. N.: Synthesis of Control Systems for Nuclear Power Plants, "Conven-
tion Record of the IRE 1954 National Convention," pt. 9, "Medical and Nuclear
of radiation detectors exist, but very few are suitable for control purposes.
the flow of coolant might be used. This method has two disadvantages:
First, the information is slow; second, over much of the reactor operat-
involved. Also, for reactors without a plant attached, this method may
be unfeasible.
poses, it is evident that the total power emanating from the reactor con-
sists of the sum of all the energies involved in the nuclear disintegrations
the direct fission process and less than 10 percent by indirect beta and
beta and gamma power does not directly follow the fission power time-
187
188 CONTROL OF NUCLEAR REACTORS AND POWER PLANTS
arise where the relationship between neutrons and power is much more
complex, but for most purposes the simple assumption of direct propor-
time and position in the core. In order to read power output directly,
sible throughout the core and an average taken of their output readings.
Second, a detector might be placed so far from the core that the core can
to be at all applicable, with the second approach being the one most gen-
erally used.
ments are usually placed at some distance from the reactor imposes
normal full power output might be 1,000 kw, if the output were turned
down to 1 kw, one would consider that the plant was essentially shut
the medium at any power level 1 kw still represents a great many neu-
trons. If these neutrons are not monitored continuously and the pos-
and at all levels. In the case of a power reactor this requirement may
of the entire operating range from source strength to full power output.
or a very sensitive instrument may be used and its position changed with
trates the movable instrument system. The diagram of Fig. 7-2 shows
189
constant neutron flux. In this way the position of the instrument with
The advantage of the movable instrument system is that only one type
10"
UJ
UJ
io-
io-
10"
io-
UJKO
O.UJO;
Q UJ
Oo
0.0:
z uj
uj o
t>
Generated on 2011-10-24 11:53 GMT / Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google
28
ro
Ul
S*uj
g8
FULL
POWER
. CRITICAL
SOURCE
LEVEL
FIG. 7-1. Typical instrument-range coverage for a multiple fixed instrument system.
reactor is usually very nonlinear and may contain sharp slope changes at
calibrated out of the instrument, they are apt to change with time.
Figure 7-1 also indicates the conventionally used terms to denote the
range as such. These ranges are the source, or startup, range, the period
The source range is the range of neutron level between the points
where the reactor is shut down as completely as possible and where the
reactor is critical. At the complete shutdown level the detector sees only
a reactor goes critical depends upon how fast reactivity is removed from
the reactor. This level is usually between 20 and 1,000 times the source
REACTOR
("
/ Up ^
MOVABLE FIXED
1|
I ' isl i 1
= . SERVO
g MOTOR *
NEUTRON
CIRCUIT
~\
Generated on 2011-10-24 11:53 GMT / Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google
REFERENCE
LEVEL
FIG. 7-2. Block diagram of movable instrument system in which the detector is auto-
a given rate or period. This interval in level is called the period range.
The period used for specific reactors depends upon their operational
requirements and may vary from 3 sec to a few minutes. In this range
circuits providing both level and period information are usually con-
190
might be considered for reactor control can be divided into two types,
600
Q 400
o
Generated on 2011-10-24 11:53 GMT / Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google
ro
2200
2600
2400
VOLTAGE
3000
1000
100
10
FIG. 7-5. Integral bias curve for typical BF3 counter showing counting rate as a function of
192
to the BFa counter. Its advantages over the BF3 counter are that it
REACTOR CONTROL RADIATION DETECTORS
193
does not contain a fine wire electrode and that it usually operates at a
is that for equal volumes it is usually less sensitive than the BF3 counter
and has a shorter voltage plateau. This shorter plateau usually means
that the tube will perform more poorly in the presence of large gamma
fields.
ELECTRODE NO. 1
ELECTRODE
NO. 2
MATERIAL
choice of this coating material and by the use of neutron shields placed
Two types of fission counters are available, a spiral type1 and a cylin-
drical type.6 Figure 7-6 indicates a drawing cross section of the spiral
type of fission counter. Two separate foils, usually coated with uranium,
are wound concentrically and electrically insulated from each other with
a very small spacing of about 0.02 in. The spirals are mounted into a
Generated on 2011-10-24 11:53 GMT / Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google
REACTOR CONTROL RADIATION DETECTORS
195
100
10
-NEUTRONS:
-ALPHAS
ure 7-9 shows an integral bias curve for a fission counter which uses a
particle of fixed energy, the tube circuit must be biased to select pulses
collected as current.1
ate. The ratio of neutron flux to gamma flux at the normal operating
usually such that the signal caused by the neutron flux is about 1,000
times that caused by the gamma rays. Therefore, when the reactor is in
gamma-ray intensity does not vary directly with reactor power as does
the neutron intensity. Thus, a lower limit is set for neutron detection
at the level where the ionization caused by neutrons equals that caused
by gamma rays.
drical type. That is, the internal electrodes may consist of a series of
Fig. 4-16.
is, by balancing out the component of the signal caused by the gamma
Generated on 2011-10-24 11:53 GMT / Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google
REACTOR CONTROL RADIATION DETECTORS 197
The instrument is a small one and in the past has taken two forms, one a
cylinder roughly the size of a fountain pen, the other a disk about the
size of a half dollar. These instruments are rugged and reliable and do
One disadvantage of the instrument is that its design usually must make
seconds result and the instruments have not been used in the past for
that they operate in quite high neutron fluxes and ultimately suffer from
boron depletion.
neutrons are also very sensitive to gamma rays. Also, the multiplier
portion of the detector is very susceptible to heat in that the noise level
Generated on 2011-10-24 11:53 GMT / Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google
literature.11~14
sities around a power reactor must be taken for granted, and provisions
made to live with them. The gamma radiation might originate in the
capture gammas in the shield or the instrument itself. For most purposes
f Sections 7-4 to 7-7 were originally presented by J. C. Connor and the author in
1000
2000
1000
20 40 60
FIG. 7-12. Gamma intensity versus distance, during reactor operation and shutdown.
with distance from the reactor. Gamma-level problems then are sub-
100
10
Generated on 2011-10-24 11:54 GMT / Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google
0 20 40
reactor shield.
198
199
integral bias curve for a BF3 counter and amplifier operating with and
10,0 00 r |-I
1000 r
; 100 r
(ARBITRARY UNITS)
FIG. 7-14. Integral bias curve for BF3 counter with and without gamma radiation back-
ground.
of the curve at high discriminator settings begins to fall off and ultimately
Generated on 2011-10-24 11:54 GMT / Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google
reaches a point where no usable plateau range exists. This point may
106 r/hr some form of semipermanent damage does occur. After even
short exposures to these very high levels the counters do not operate
properly, and several weeks may be required for the instrument to return
CONTROL OF NUCLEAR REACTORS AND POWER PLANTS
not used at such high gamma levels, but the damage may occur even if
position of the BF3 gas and the amount of gas amplification which is
possible.
NEUTRONS+a ;
.+NOISE
7+NOISE,
GAMMA LEVEL
=1.7x105 R/HR -
(ARBITRARY UNITS)
background.
This secondary process must compete directly with the ionization caused
200
REACTOR CONTROL RADIATION DETECTORS
201
plus gamma rays and the other sensitive to gamma rays alone. A sub-
o 10"
CD
10"
1CT'
•UNCOMPENSATED
-COMPENSATED
AFTER LONG-
VX^-TIME REACTOR
\V IRRADIATION
ORIGINAL —
COMPENSATION
0 5 10 15 20 25
Generated on 2011-10-24 11:54 GMT / Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google
FIG. 7-16. Ionization chamber current versus time after reactor shutdown for various
degrees of compensation.
if the reactor has control rods of any type, the depletion may be compen-
sated for by rod motion. A rod movement can cause a change in the
able to move control rods for any one or a combination of the following
reasons:
5. To change the flux distribution within the core in such a manner that
Figures 7-17a and 7-17& are two views of a thermal reactor containing
four absorber type control rods which can move parallel to the Z axis.
The curves to the right of Fig. 7-17a show what might happen to the neu-
tron flux distribution along the Z axis in the core if all four control rods
since the peak in the flux distribution shifts up and away from the instru-
since the flux peak moves toward the axis of the instrument.
The curves plotted to the right of Fig. 7-176 show what might happen
to the neutron flux distribution along the Y axis for a nonuniform rod
203
if all rods are at a uniform height. Curve D shows that the peak in the
and 4 remain at the same height. Other rod withdrawal programs might
exists for a reactor of the design shown which is not subject to some
DETECTOR
BL
t—X
NEUTRON
FLUX
®DETECTOR
fa)
SIDE VIEW OF
®DETECTOR
(b)
TOP VIEW OF
Generated on 2011-10-24 11:54 GMT / Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google
Fig. 7-18, containing four control rods equally spaced radially from the
the reactor subcritical and all control rods inserted. Figure 7-18 is a
plot of the counting rate data that might be obtained by withdrawing and
measured with respect to a line between the detector and the center of
the core. Note that the maximum effect occurs for those rods closest
to the detector. The dashed curve through the four rod position points
indicates the counting rates which might be obtained if the reactor had
Rod shadow effects of this nature are greater the closer the instrument
is to the core. At points in the shield far from the reactor, the neutron
detectors, however, may be placed farthest away from the core, thus
/ii^V
i— 0DETECTOR
EACTOR
l^
-90°
TOP VIEW
OF REACTOR
ROD z^x'
s'
'XNx
xv.v
^
Generated on 2011-10-24 11:54 GMT / Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google
ROD 1 ^"'
^J3
^^ ROD 4
no=
ei
ALL RODS IN
PECT TO DETECTOR M
IS
-180°
-90°
+ 90°
+180°
FIG. 7-18. Magnitude of instrument reading change as control rods are pulled one at a
time.
is apparent that the number of fast neutrons reaching the detector will
trons and hence the neutron-detector calibrations depend upon the mag-
204
REACTOR CONTROL RADIATION DETECTORS
205
unless great care is taken in the instrument design and usage, counting
100
2200
2800
2400 2600
VOLTAGE
still most useful, particularly once they have been directly calibrated in
terms of power level. Three basic types of power calibration are usually
tion between any of these methods and the neutron detectors can be
accomplished.
Experiments of this type are most easily conducted at low power when
radiation levels around the core are low enough to permit physical access
to all parts of the plant. Data obtained at low levels can then be extrap-
ibrations at low power are important to assure that such reactor design
in the core for a known length of time and measuring the induced activ-
ity.2 The power P developed within the core is given by the product of
the number of fissions occurring per second, and Q, the energy release
P = nvZfQ (7-1)
centimeters per second, and 2/ is the macroscopic cross section for fis-
mine average core flux if the core structure is at all complicated. Access
to the core for removal and insertion of foils must also be possible.
Generated on 2011-10-24 11:54 GMT / Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google
neutron source of known strength, and plot the resultant rise in counting
be shown that the rate of change of neutrons with time after an initial
transient, and consequently the rate of power increase, is linear and pro-
n + ) \iLi + b (7-2)
dt I*
t=i
where the symbols have their usual meanings and S is the neutron source
strength.
zero. The boundary conditions are that when t = 0, n = n0, (7, = Cl0,
having a single decay constant X « 0.1 sec~1. For this situation we find
that
n=
"' "
_ _ . __
[X + (ft/I*)]* X + (ft/I*)
when t "2> 10 sec. In this type of solution for large values of t and any
reasonable value of I*, the negative exponent of the transient term in the
solution becomes quite large, and after an initial waiting time, dn/dt
dn _ X „ ,_ ,
In other words, after an initial transient, the reactor neutron level rises
If the kinetic equations are solved using all six delayed-emitter groups,
the expression for n would contain six transient terms, each of which
decays after several half lives to a negligible value. Then, after a given
dn/dt is given by
Generated on 2011-10-24 11:54 GMT / Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google
i i 0i i 0i . i i 06 \ dn „ . .
K%-S (7-7)
lâ„¢ (7-8)
n = no + 10r*St (7-9)
transient is past and the reactor is rising linearly with time. For our
n0 = 10-2SD (7-10)
208 CONTROL OF NUCLEAR REACTORS AND POWER PLANTS
thermal neutrons in the core if S is the total source strength in the core
defined as
I* = - (7-11)
where ft is the mean free path and v is the velocity of a thermal neutron
in the core. The probability that a neutron will be absorbed is simply the
tion, then
= absorption
production
envSf
P = ^ (7-15)
depends upon the validity of the assumptions made in the derivation, the
power level at which the measurements are made, and the strength and
209
measuring coolant flow and temperature rise across the core, power being
instruments.
may vary as a function of time and operation and to see also the variations
THERMAL
POWER
CORRECTION
COMPARATOR
NEUT
POWE
~1
!0
RON I
* "" I AMPLIFIER
1 GAIN
| CHANGER
Generated on 2011-10-24 11:54 GMT / Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google
n<
BTU
CALCULATOR
,i
AMPLIFIER
_., 1
NEUTRON
DETECTOR
II
""1
REACTOR
ii1
HEAT
IXCHANGER
STEAM
PLANT
^ PHI
"\
IP]
FIG. 7-20. Block diagram of automatic calibration system for nuclear instrumentation.
tor computes the product of the coolant flow and the temperature differ-
tor. The signal is amplified and then compared with the output signal
comparator then adjusts the gain of the amplifier in such a manner that
the neutron power reads the same as the calculated thermal power.
210 CONTROL OF NUCLEAR REACTORS AND POWER PLANTS
Let us examine now what the Btu calculator actually reads. It cer-
tainly reads all of the thermal input to the coolant loop. This includes
pump power, incidental beta-gamma power from the reactor, and fission
power from the reactor. The neutron detector reads essentially fission
tioned. Let us assume that the amplifier gain changing mechanism sets
Let us now suppose that a power demand change is made on the system
after this change is met by the plant, the beta-gamma power given off by
the reactor has not changed appreciably from its high power level and
might now represent a full 60 percent of the actual thermal power output.
full power plus a 16 percent correction factor and would indicate that
Btu calculator time constants, which should be quite long in order not to
Generated on 2011-10-24 11:55 GMT / Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google
disturb the control system, will catch up with this situation and all meters
would eventually decay out and ultimately become 6 percent of the new
circuits which follow the detecting elements just described are a complex
usage.16'17 These circuits fall into two categories: pulse counting and
view, is that the output of these circuits is always a d-c signal which is
The circuits which are peculiar to reactor control are the period cir-
provide an inverse period signal. The heart of this operation is the log-
REACTOR CONTROL RADIATION DETECTORS 211
arithmic amplifier which also is used to compress the wide range required.
amplifiers.1s'19'23'24-2s
REFERENCES
3. Korff, S. A.: "Electron and Nuclear Counters: Theory and Use," D. Van Nos-
5. Baer, W., and R. T. Bayard: A High Sensitivity Fission Counter, Rev. Sci. Instr.,
York, 1954.
with Electrically Adjusted Gamma Compensation, Rev. Sci. Instr., vol. 25, no. 2,
9. Barbaras, G., et al.: The Design and Construction of Boron Coated Thermopiles
10. Barbaras, G., et al.: The Design and Construction of Boron Coated Thermopiles
tor, Rev. Sci. Instr., vol. 19, pp. 744-770, November, 1948.
12. Gilette, R. H.: Calcium and Cadmium Tungstate as Scintillation Counter Crystals
for Gamma-ray Detection, Rev. Sci. Instr., vol. 21, pp. 294-301, April, 1950
13. Morton, G. A.: Photomultipliers for Scintillation Counting, RCA Rev. vol.. 10,
14. Hofstadter, R., et al.: Detection of Slow Neutrons, Phys. Rev., vol. 82, p. 749,
June 1, 1951.
15. Schultz, M. A., and J. C. Connor: Reactor Power Calibration, Nucleonics, vol.
18. Meagher, R. E., and E. P. Bently: Vacuum Tube Circuit to Measure the Loga-
rithm of a Direct Current, Rev. Sci. Instr., vol. 10, p. 336, 1939.
19. Wade, E. J.: Instruments Used with Experimental Reactors, "Convention Record
of the IRE 1954 National Convention," pt. 9, " Medical and Nuclear Electronics,"
21. Dunlap, G. W.: Detection and Measurement of Nuclear Radiation, Elec. Eng.,
August, 1948.
22. Cochran, D., and C. A. Hansen, Jr.: Instrumentation for a Nuclear Reactor,
Meter for Reactor Control, Nucleonics, vol. 11, no. 11, November, 1953.
24. James, W. G.: Logarithms in Instrumentation, ORNL 413, Oak Ridge National
ing Rate Meter Covering a Wide Range, Proc. Inst. Elec. Engrs. (London), vol.
26. Jaques, T. A. J., H. A. Ballinger, and F. Wade: Neutron Detectors for Reactor
Instrumentation, Proc. Inst. Elec. Engrs. (London), vol. 100, pt. 1, p. 110, 1953.
27. Carmichael, H.: Design of the Chalk River Ion Chambers, Natl. Research Council
28. Jordan, W. H., H. B. Frey, and G. Kelley: An Instrument for Measuring the
Logarithms of Neutron Level and the Period of a Pile, Oak Ridge Natl. Lab. Rept.
range from zero to full power is a range covering nuclear fissions resulting
in a few neutrons per second to many billion neutrons per second. Unless
accident is most likely to occur. The primary reason that startup range
is the most dangerous one is because here the measuring instruments are
sense, but at low neutron levels the statistics of the instrument circuits
the neutron level in a reactor starts to rise rapidly in the startup range,
The basic startup problem is first getting the initial neutron level up
high enough that the detecting instruments can measure their fraction
of the neutrons with good statistics. Many factors enter into setting
this initial startup level, and these will be examined in this chapter.
We are not concerned here with the startup of a nuclear power plant,
complete series of details depending upon the design of the plant and
rapidly for fear that dangerous stress forces might develop. Conse-
necessary.
There are essentially two types of reactor startups that will be de-
scribed: first, the initial cold startup of the reactor and, second, subse-
neutrons that set instrument levels, and then we shall examine these
213
214 CONTROL OF NUCLEAR REACTORS AND POWER PLANTS
observation at all times, there must be available neutrons at all times for
sion will create a few neutrons in a reactor, but their number is usually
tion of the neutrons from this source which is usually observed by the
there are usually two other types of created sources of neutrons. These
particle. (a,n) reactions are very prominent, and the Be(a,n)C reaction
is the one generally used. The alpha particle can be obtained from the
upon the number of curies of activity in the initial Po and upon the
Generated on 2011-10-24 11:55 GMT / Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google
might be such that one curie of Po-Be can be made to give off in the
Actually it is not this source strength which is important, nor even the
It is quite obvious, however, that the larger the source rating in curies,
the more counts per second will be indicated by the detecting instru-
ments. The only limit to the size of a radioactive source is the practical
one of how large a source can be safely handled in the fabricating and
sion a small factor may be gained by splitting the source into a number
of discrete pieces, each piece being more easily handled than the over-all
source itself. Physically the source can be made quite small. Com-
215
roughly J4 in- m diameter by a few inches long, or they are also avail-
Because of the fact that Po210 is radioactive with a half life of only
138 days, the neutron emission from the Po-Be source drops off rapidly
because of this source strength dropoff. Figure 8-1 indicates the rela-
time from the initial insertion of the source to the end of the life of the
1.0
0.5
0.2
0.1
0.05
0.02
TIME, DAYS
reactor is 2 years, then the source will have fallen off by a factor of approx-
examine induced source levels in a reactor that has operated for some time.
from Chap. 7, an inevitable large gamma flux exists around power reac-
this method the gamma energy must be at least equal to the binding
8 Mev for most nuclei except for some of the light ones. Deuterium and
and for this reason they are feasible for use in induced neutron sources.
CONTROL OF NUCLEAR REACTORS AND POWER PLANTS
getic and does not have the fission neutron energy spectrum.
down of the reactor after any appreciable power operation a large gamma
flux exists in and around the core. Now if the reactor contains either
beryllium or water in its structure in any form, new neutrons will appear
OPERATION AT FIXED
POWER LEVEL
AFTER 10 HOURS
OPERATION AT SAME
OPERATION AT SAME
OPERATION AT SAME
20 40
FIG. 8-2. Relative core flux after shutdown caused by photoneutron production in D2
many deuterium nuclei would be available for the (7,71) reaction. The
cross section for the (7,71) reaction with deuterium is sufficiently high
shutdown. It will be recalled from Fig. 7-12 that the gamma radiation
the photoneutrons would also drop off. Figure 8-2 indicates the relative
has been shut down. This series of curves shows that the number of
216
217
10"
and the longer the reactor has been operated at power levels prior to
neutrons at shutdown. The curves of Fig. 8-2 also indicate that when-
ever possible after a reactor has been shut down, it should be started up
delayed neutrons are emitted from the fission products. These delayed
neutrons are given off at various times, but after a sufficient time interval
tor must be turned off from its power-level operation, all efforts should
be made to start it up again in less than 20 min in order to use the fission-
10-
icr
«r
24
4 8 12 16 20
emission.
8-2. Initial Reactor Startup. After a reactor has been assembled, the
Many questions are in the minds of the designers. The first question is:
Will the reactor go critical at the point that the calculations have indi-
cated? The reactor at this stage is presumably cold and at room tem-
perature. Any plant that might be used with the reactor is not in opera-
tion. The instruments are set at their most sensitive positions and
in that higher sensitivities are more easily obtained. The use of the
scalers means that the operation will take a fairly long time. Low count-
ing rates will initially be used, and large time constants will be necessary
ways depending upon its design.1 Some reactors, such as the water
amount is present that the reactor becomes critical. Other reactors such
as the NRX reactor add moderator in the form of heavy water to such a
level that the system becomes critical. Still other heterogeneous reac-
tors have fixed fuel and moderator and remove control rods so that the
is apt to be a very slow and careful one, the performance of the reactor is
to 1/(1 — fc), or the counting rate from an instrument channel will be
Generated on 2011-10-24 11:55 GMT / Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google
the multiplying factor. In this way, when 1/c.r. approaches zero, the
the large heterogeneous reactor the abscissa of the plot might be rod travel
in inches. Figure 8-4 shows such a plot for bringing a water boiler to
shape of the curve as the reactor approaches criticality will depend on the
very close to the reactor, it will effectively see the source for a long period
manner, the reactor will take longer and longer to settle out at a fixed
OPERATIONAL CONTROL PROBLEMS: STARTUP
219
neutron level or counting rate. At criticality the reactor level will con-
tinue to rise indefinitely because of the additive neutrons from the source.
Actually, criticality will be reached at such a high level that the source
described methods and then the source is quickly removed from the reac-
tor. If the reactor were exactly at critical, the counting rate would stay
TO SOURCE
INSTRUMENT
FARTHER AWAY
FROM SOURCE
THAN NECESSARY
KILOGRAMS OF URANIUM
placement.
-CRITICAL
-SUBCRITICAL
Generated on 2011-10-24 11:55 GMT / Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google
TIME
critical.
rate would drop off slowly, and if the reactor were supercritical, the level
supercritical case, before any rapid rises are permitted to occur, some
would usually have provisions for removing the source. Large power
operation has been reached and the reactor then shut down, there are
in Sec. 8-1. The second major effect will be the change in subcritical
Let us assume that we have a reactor whose design is such that the nor-
10~45fc/°F and that the normal hot operating temperature of the reactor
into the reactor in changing its temperature from cold to hot. The
over-all hot k would now be 0.8, and the subcritical multiplication factor
a high power level for a while, the so-called peak xenon effect will build
would insert would depend upon the reactor design, its sepecific power,
the length of operation at any given power level, and the time after shut-
down that the reactor is started up. It is well within conceivable prob-
ability to have the poisoning effect also change the counting rate up to
another factor of 3 or 4.
After long-term operation at power level some of the fuel has been
burned up. This uranium depletion also changes the subcritical multi-
very much upon the time after shutdown the reactor is again started up.
perature, poisoning, and depletion all may change the readings of the
or minus from the initial startup cold level and may be measured in thou-
OPERATIONAL CONTROL PROBLEMS: STARTUP
221
greatly appreciated.
at a given time, and with some knowledge and confidence of the reactor
and the plant, the first problem is: How fast should the reactor be brougth
be laid down.
-4 -2
reactivity.
P = d(Sk)/dt (8"2)
That is, the period changes linearly with the reactivity as long as the
critical at a finite rate the curves start to bend over, as is seen in Fig. 8-7.
CONTROL OF NUCLEAR REACTORS AND POWER PLANTS
L=
-_
L=
[d(6k)/dt]P
(8-3)
(8-4)
level depending upon how fast reactivity is inserted into the reactor.
-12
-14
-4 -6 -8 -10
FIG. 8-7. Period versus reactivity of a reactor for given rates of change of reactivity.
Equation (8-4) obviously also does not hold when the level gets too close
per second can therefore be swapped off directly against time. For
222
Obviously, then, the reactor can also go critical at any desired period.
Figure 8-8 indicates the type of curve that can be obtained showing
223
tor. This curve changes slightly as a function of I*, but the general shape
Rod Effectiveness. Until now we have been talking about linear rates
view.
mal reactor reduce the reactivity in two ways. They absorb thermal neu-
point in question before the insertion of the rod, and they also increase
the thermal leakage from the reactor. Physically the presence of the rod
decreases the number of fissions in its vicinity. Thus for a given total
power output the flux must be higher in some other portion of the reactor.
of rods is only slightly less than the sum of the individual values for each
rod used alone. This is the condition whereby the rods are not too close
together, since the shadowing of one rod by another will depend upon the
distance between rods. For a small reactor the effectiveness of the rods
0.16
0.14
0.12
0.10
0.08
j 0.06
0.04
0.02
0 4 8 12 16 20
pletely withdrawn from the reactor its effectiveness again is quite reduced.
can be used where x is the distance the rod is extracted out of the reactor.
From Fig. 8-6 it has been shown that at greatly subcritical operation
almost any reasonable reactivity rate produces the same results but close
to criticality the level and period are sharply dependent upon reactivity
TIME
FIG. 8-9. Power level versus time for a reactor being started up on either linear rate of
considered as that value of reactivity rate which the rods possess as the
With all of the variability in startup level that exists and additional var-
soning, etc., reactor startup systems usually must be designed, for safety
rods are at a lower effectiveness most of the time. Hence, it takes longer
Generated on 2011-10-24 11:56 GMT / Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google
for the reactor to attain criticality. Figure 8-9 illustrates this effect by
then linear rod motion rates on a rod setup whereby the rods have an
224
OPERATIONAL CONTROL PROBLEMS: STARTUP
223
equal to the previous linear reactivity rate. The length of time to reach
critical for this particular startup from —13 percent in reactivity is nearly
doubled.
as pulling a rod for a given length of time and then stopping and then
pulling again, etc., is sometimes used. For analysis purposes the rates
of change of reactivity and consequent levels and periods for this type of
given power level is on a period basis; that is, the reactor will be brought
140
60
40
20
TIME
FIG. 8-10. Period versus time for a reactor starting from shutdown and going to power
level.
reached before the reactor goes critical at given rates of reactivity with-
manipulates the control rods to maintain this period until the desired
There are several requirements upon which the startup time or the
such a reactor is most undesirable, and shutting the reactor down to fixed
a few percent of the plant full-power output rating. It then would take
the reactor only a minute or so to get up to a useful output level from this
stand-by type of state. This step presupposes that the steam plant and
auxiliaries are always ready to run. Starting from source level is rarely
tion cycles that are turned on every morning and turned off every night.
Startup times should be minimized so that the day is not wasted in getting
sible, and the operational startup requirements for each reactor must be
In order to get a feel for the numbers involved let us again examine an
this operation. Therefore 10.8 min are available to bring the reactor to
—13 percent, such as has been used in some of the previous examples, the
average reactivity rate required for this startup would be 2 X 10~45fc sec~1.
insertion rates for a reactor plant provided the safety and poisoning con-
which safety startup considerations are based is that the speed at which
reactivity may be inserted into a reactor can be that speed at which the
faster reactivity is inserted into a reactor, the shorter will be the period
at a given power level and the greater the possibility that the reactor
much more severe than the others. This is the so-called startup accident,
inserted into a reactor at a given rate and nothing is done to stop it. In
driving the rods becomes stuck. The motor keeps turning, and the rods
are extracted from the reactor until they are all the way out. In this way
means of stopping rod motion are not available until the reactor power
The problem then becomes: How bad is this accident, and what sort
of periods are involved? If the level of reactor power rises too rapidly,
an overshoot of the top protection level signal may result and the reactor
Two methods of analysis of reactor power level and period are available.
The first method is to attempt to solve the reactor kinetic equations for
This method is tedious and restrictive in that the entire range of probable
Generated on 2011-10-24 11:56 GMT / Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google
periods and levels involved cannot be seen without many numerical cal-
culations. Another method that might be used is to solve for the bound-
worst case.
tively there will be two such bounds, the first an upper bound on how short
the period can get as the reactor becomes greatly supercritical and the
minimum period involved which will depend upon the rate of insertion of
reactivity.
t The development of the period upper bound follows that of H. W. Newson in the
classified literature. The lower bound analysis presented was first accomplished by
F. Engel, Jr.
228 CONTROL OF NUCLEAR REACTORS AND POWER PLANTS
dn Sk - /3 . V % r< i a to a\
n + ) \id + S (8-5)
dt I
If we note that the last two terms on the right-hand side of the equation
dn ^ Sk - /3 ,0 „.
_>__n (8_6)
P = -T^TT, (8-7)
dn/dt
it follows that
7*
P<
for Sk > /?. For Sk < ft the inequality of Eq. (8-8) loses its significance
with regard to fixing the magnitude of the period. However, this inequal-
ity establishes a real upper bound for the period when the reactor is above
prompt critical.
can possess by saying that within a given upper criticality range the
A lower bound to the period in the subcritical range may also be estab-
g£
P = ,,,.. ,,. that this lower bound will depend on the rate of change of
Sk sec~',
P = =^ (8-9)
For reactors subcritical by a large amount it has been shown that this
does not hold. The formula is inaccurate because it assumes that all
emission times. It is felt intuitively from a control point of view that the
effect of the delayed neutrons is to slow down any level changes, thus
tive lower bound. A rigorous analysis will now be presented which sub-
For simplicity, consider the pile kinetic equations for the lumped-
^=«*_IL2B + XC + S (8-10)
f = ^-AC (8-11)
For the subcritical condition the reactor must ultimately reach some
Thus at equilibrium
= Jn + XC + (S = 0 (8-12)
CLv t
dn Sk dC . „ Sk . „ _
di=pn-lTt+s = i*-n + s = 0
n=
(*-^)
, = ^ (8-18)
Generated on 2011-10-24 11:56 GMT / Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google
equilibrium value of the neutron flux level must be greater than the level
possible value for the time rate of change of the neutron flux level then
230 CONTROL OF NUCLEAR REACTORS AND POWER PLANTS
would be
A. Sl*Mk y (8_19)
5fc0(5fco +
Sl*y
Sko(Sko -
(8-20)
n _ SI* (Skr _ _ Sk
Eq. (8-15), but the above process appears to have greater validity.
startup type accident have thus been established. Figure 8-11 illustrates
these bounds for several conditions. It will be noted that the upper
1.2 X 10~45fc sec~1 into a reactor having an I* of 10~4 sec. At large sub-
criticalities the reactor follows the lower period bound. Above prompt
It will be noted, of course, that the upper bound is really in the wrong
period follows the upper bound as developed so closely that this bound
may also be used as a practical value of the minimum period. Using this
limit type of approach, the severity and time involved in a startup acci-
tically changed.
developed depends upon how far subcritical the reactor is at the start
231
any level above the startup level is a function of from how far below that
level the initial reactivity insertion started. Let us assume, for example,
100
50
20
10
in
0.5
0.2
0.1
0.05
0.02
0.01
\II
V«—7=10~58k/SEC
PERIOD—A
LOWER \
Generated on 2011-10-24 11:56 GMT / Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google
BOUND \
-b. 6k V-
7=10-46k/SEC
i =1.2x10-4 6k/SEC-
l*=10-4 SEC
=10-38k/SEC
-4
-2
PERIOD
BOUND
r^ I*
r'
SEC
SEC
X10
,-s
0246
REACTIVITY, 8k
FIG. 8-11. Period as a function of reactivity indicating upper and lower bounds for a
startup accident.
rod motion. A good example might be that when the reactor power level
The problem then exists of how much time is available from when the
CONTROL OF NUCLEAR REACTORS AND POWER PLANTS
moving in. This available time depends upon the period that the reactor
upon how far below this protection level the startup accident began.
tion level as a function of the decades of rise in reactor power level before
the protection level is reached. The curves again are for the example
that the reactor was originally —13 percent subcritical. It can be seen
from Fig. 8-12 that if only a few decades of power-level change are
Db0PPr
INSE
-APPRO
\
Generated on 2011-10-24 11:56 GMT / Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google
EAcnvr
RTION
< 4xicr4
rv
WE---
Sk/SEC
REACTIVITY
--INSERTION RATE -
>x
^^^ -
18 !
46 8 10 12
DECADES OF RISE
FIG. 8-12. Period attained by a reactor during a startup accident at a given fixed pro-
tection level versus the number of decades below the protection level reactivity insertion
involved between the startup point and the protection level, the periods
accident occurs when the reactor is started possibly 10 decades below the
protection level, very fast periods can occur, particularly at high reactiv-
232
power level that a reactor would attain above the protection trip level as
a function of the time delay in doing something about the rising power
level. In other words, while waiting for the protection signal to act
OPERATIONAL CONTROL PROBLEMS: STARTUP
233
important, as the power level could conceivably become very high during
this delayed interval. Conversely, Fig. 8-13 indicates that if a 200 per-
tion signal is provided at 100 percent full power, the protection system
must reverse the reactivity and cause the power level to start coming
down with a delay of less than 0.7 sec, provided that the reactor is on a
Chap. 10.
1000
500
FIG. 8-13. Relative power level attained by a reactor on a given period as a function
until now have been based upon the reactor having no temperature coeffi-
cient. We have seen in the case of automatic control that the addition
stopping the control rods or reversing their motion during a startup acci-
NO PROTECTION
SMALL NEGATIVE
TEMPERATURE
COEFFICIENT
LARGER NEGATIVE
TEMPERATURE
COEFFICIENT
POWER
RANGE
TEMPERATURE COEFFICIENT
SCRAM
TIME
FIG. 8-14. Reactor power level as a function of time for several types of protection
systems.
Let us crudely examine the process that exists during the startup acci-
dent for several types of protection systems. Figure 8-14 illustrates some
Generated on 2011-10-24 11:56 GMT / Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google
of these situations. Let us assume that the reactor comes into the power
range on a given period. Curves similar to those in Fig. 8-12 can be used
reactor power level would continue to rise on a fast exponential until the
temperature of the reactor will rise to such a point that sufficient negative
reactivity will be inserted into the reactor to overcome the positive reac-
234
OPERATIONAL CONTROL PROBLEMS: STARTUP 235
tivity put in by the startup accident. When this occurs, the power level
must reverse itself and the reactor will tend to reach an equilibrium posi-
tion between the startup accident reactivity and the negative reactivity
rod pulling continues, the average temperature of the reactor will con-
tinue to rise to match this rod pulling and the power level will gradually
follow, assuming some simple coolant flow pattern. This second rise
following the first sharp peak is usually a long time effect taking in the
startup accident becomes smaller and smaller until with a very large
method of startup and operation. That is, the rods are pulled all the
way out deliberately and the power level rises to such a point whereby
the negative reactivity set by the average temperature comes into balance
power level. However, for most reactors having large excess reactivities
dangerous.
external means for protection. One simple means is to stop the rods
Generated on 2011-10-24 11:56 GMT / Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google
scheme is to reverse the rod motion at the same speed at which the rods
violently as possible. It will be noted in Fig. 8-14 that for internal pro-
tection the power level remains high at some point where a balance in
reactor is always protected or safe and above which some damage may be
course, are related. For convenience, the safe condition is usually defined
of a reactor may be such that below 200 percent of its rated power level
the reactor is always safe. If the power level is permitted to rise above
200 percent full power, something will occur that will damage the system.
Generated on 2011-10-24 11:56 GMT / Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google
OPERATIONAL CONTROL PROBLEMS: STARTUP 237
The peak poisoning after shutdown (as well as the equilibrium poisoning)
ing as the ratio of the thermal neutrons absorbed by the poison to the
TABLE 8-1
Equilibrium poison
1012
1013
10"
2 X 1014
0.0054
0.028
0.045
0.046
0.0055
0.03
0.20
Generated on 2011-10-24 11:56 GMT / Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google
0.38
Edlund.7
Once the poisoning has reached its peak, it starts to decay back to its
insertion of control rods or other poisons, or else the power level will rise
been discussing how fast the control rods might be extracted from a
reactor, but this present consideration concerns how fast they must be
A special extreme case exists, which may or may not be taken into
power operation. The curve of Fig. 2-9 is for the case of poison build-up
with the reactor at zero power. The second case is the condition whereby
the reactor is turned on at the time peak xenon is reached and then
immediately run at full power. Here the xenon poisoning is burned out
very rapidly. Strictly speaking, this is not a startup problem and will
To determine the speed of control rods for any of these poison condi-
ably get into a situation whereby it is necessary to move the rods to com-
made in the control system for maintaining at least sufficient rod effec-
tiveness which can be inserted at a given rate to take care of this poison-
course, undesirable.
10 20 30
40
FIG. 8-16. Poisoned startup example in which the reactor is not capable of overriding
peak xenon.
Although the problem of safety does not exist, if one attempts to over-
ride the initial portion of the xenon build-up curve of Fig. 2-9 between
front edge of this curve has a slope approximately 10 times that of the
rear end, and if this poison effect must be overridden, a rate 10 times faster
238
Many reactors do not provide for complete xenon override. That is,
they do not have sufficient excess reactivity to overcome the poison even
when their rods are all the way out. For these reactors, in the event
they are shut down after an extended operational period, a startup must
shut down. It is quite evident that this sort of operation requires that
speeds during startup having been examined, we can now briefly look at
startup control systems. The first and most obvious control system
plication factor of a reactor to such a state that the reactor power level
8-10 indicates a period versus time curve for such an operation. A man-
ual control on reactor rod motion would simply be first to extract the
rods at a given rate until the reactor reached the specified period, in the
case of Fig. 8-10, in 20 sec, and then rod motion would be stopped or
would be maintained at the fixed period until the specified power level was
control rods may be moved in or out provided the reactor period is not
shorter than a fixed amount. Once shorter periods are achieved, the
and is gradually brought from long periods toward a 20-sec period. For
periods longer than 25 sec the control rods may be moved either in or out
rods from being withdrawn any farther and rod motion is stopped. In
the event that a 15-sec period is reached, interlocks take over and permit
Figure 8-18 shows such an automatic control system for period control.
CONTROL OF NUCLEAR REACTORS AND POWER PLANTS
This loop is very similar to the rod control loops of Chap. 4. The prin-
cipal difference is that instead of the level demand being placed into the
fixed voltage. The reactor, through the neutron detector and period cir-
50
40
vi
£20
Q.
10
LEVEL
TIME
FIG. 8-17. Permissive type startup control. Period controls direction of rod motion.
CONTROL
RODS
REACTOR
NEUTRON
Generated on 2011-10-24 11:57 GMT / Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google
DETECTOR
ACTUATOR
PERIOD
CIRCUITS
PERIOD
SIGNAL
ERROR
SIGNAL
AMPLIFIER
COMPARATOR
PERIOD
DEMAND
240
cuits, would create a matching voltage, and any error signal would be
determine the constants and compensations needed for this type of con-
trol loop. The error-signal amplifier may be of the on-off type or pro-
portional type.
OPERATIONAL CONTROL PROBLEMS: STARTUP
241
whereby both period and level are used to control the motion of the rods.
For the purpose of automatic control it is essential only that the reactor
after scram is involved. Consider the block diagram of Fig. 8-19a. This
control loop is similar to the previously described loop of Fig. 8-18, with
the exception that a level and a period detector and a level and a period
demand signal have been inserted to feed information into the compara-
PERIOD
DEMAND
0-10 SEC
0-10 V
' CONSTANT
10 SECOND
PERIOD
TIME
(b)
(a)
FIG. 8-19. Combination period and level control, "magic-number" system, (a) Block
Generated on 2011-10-24 11:57 GMT / Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google
this loop. If the reactor is initially at a very low level, we can assume
might also arbitrarily assign numbers on the basis of the voltages that
one might expect in the comparator. That is, for example, a variation
from 0 to 10 volts. The level detector may be expected to give out infor-
start at some arbitrary low level, for example 8 decades below full power,
rods will be extracted from the reactor by the actuator mechanism which
has been given an error signal. This signal corresponds to the fact that
10-volt input to the comparator. It will be recalled from Figs. 8-9 and
8-10 that even though the rods are pulled at some maximum permitted
speed, for a long time the reactor will not be able to attain a 10-sec period.
actuator mechanism, and the rods are removed from the reactor at max-
out a 10-volt signal, the error is reduced to zero and the rod motion is
loop has a fast time constant compared with a 10-sec period. The period
of the reactor has now leveled off at 10 sec and is maintained by the servo
plus what is coming out of the period detector, must now match up
against the 10 volts. This situation automatically calls for the output
of the period detector to be reduced, and consequently, the pile must now
operate on a longer period. As the level increases still farther, the servo
system gets less and less information from the period channel and more
and more information from the level detector. This action is as shown
in Fig. 8-196. Here the reactor comes up through the startup range and
period range on a 10-sec period, and this period slowly levels off into
Generated on 2011-10-24 11:57 GMT / Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google
one whereby period information is the controlling factor until the power
level is reached and then level information only takes over and period
information drops out. The system that the designer selects is one which
well as the desired operating program. For example, if the level informa-
a fast servo system for level control and expect to be able to use period
same speed servo system obviously may be used. If the two time con-
OPERATIONAL CONTROL PROBLEMS: STARTUP
243
system to set up a different response time in the period range from that
Another type of interesting comparator circuit for use when the time
constants are not widely different is shown in Fig. 8-20. Here power and
POWER
INFORMATION
(IOOV = IOO%
POWER)
PERIOD
INFORMATION
(+IOV=5 SECOND
PERIOD)
POSITIVE
REFERENCE
VOLTAGE
NEGATIVE
Generated on 2011-10-24 11:57 GMT / Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google
REFERENCE
VOLTAGE
FIG. 8-20. Simplified schematic diagram of a comparator using both period and level
information.
is varied by changing the gain of the information received from the power-
measuring circuit. The diodes separating the demand circuits are used
in such a manner that only that voltage which is most positive will assume
held by the period control which prevents the reactor from ever changing
REFERENCES
York, 1954.
2. Bernstein, S., et al.: Yield of Photoneutrons from U236 Fission Products in Heavy
4. Bernstein, S., et al.: Yield of Photoneutrons from UMi Fission Products in Be,
5. Ergen, W. K.: Hard Gamma Emitters among Fission Fragments, ANP-59, May
3, 1951.
6. Moore, R. V.: The Control of a Thermal Neutron Reactor, Proc. Inst. Elec. Engrs.
POWER OPERATION
its rated power level. It will be recalled that the power range was denned
as the top 2 decades of output below the full power rating of the reactor.
It is the problems that occur within this range which will be discussed.
process. Once the plant achieves this operating range, the temperature
coefficient plus the control system maintains any desired power level.
causes of these motions plus some of the physical means for creating the
six items that might cause a reactor system to require a change in reactiv-
tional point of view and from a design point of view, in that each individ-
may require any of the types of temperature and pressure variations called
Generated on 2011-10-24 11:57 GMT / Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google
for in Chap. 6. The question to be answered is: What are the limitations
that should be imposed upon a reactor plant in changing from one power
level to another? In a large plant one usually looks outside the reactor
for the limitations. It has been shown that the reactor can change its
level quite fast. The boiler may have a temperature change rate limita-
tion, or the turbine may not be able to handle power as fast as the pri-
mary system can deliver it. In any event, it is reasonable to assume that
245
CONTROL OF NUCLEAR REACTORS AND POWER PLANTS
such as we had in the startup range. For illustrative purposes let us say
that the reactor plant should not be capable of having its power level
changed from 1 percent of rated output to full output in a time faster than
illustrates the ultimate condition of making this entire level change at the
fastest possible period. From the figure this period can be seen to be
first 13 sec the power level would increase from 1 to 2.7 percent of rated
100
50
20
10
50
60
20 30 40
TIME, SECONDS
operation.
POWER LEVEL
power level.
Generated on 2011-10-24 11:57 GMT / Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google
power and in the next 13 sec from 2.7 to 7.4 percent, and so on. In
other words, the absolute power level rises very slowly at first and then
13 sec of the operation the power level would go from 37 to 100 percent.
nothing more.
246
the reactor is the limiting component of the system in the speed of chang-
ing power level, then a combination of linear level and period type of
startup accident that the danger which exists in going from one power
247
fore if only a comparatively small level change is called for, this change can
that permits a given safe period operating area for each power level. At
any given operating power level, period signals longer than a prescribed
power level, only slow rates of change are safely permitted. This system
time.
changes of, say, 1 percent of full power per second or a similar type of
FIG. 9-3. Rod position used as a thermometer for constant-Ti, constant-coolant flow
program.
The maximum reactivity change rate required from the control rods
to change level depends directly upon the type of program. For the
cient in the reactor, the level change can be made without any rod motion
effect. We can see that the control-rod position can actually be used as a
a control-rod worth over a small range can be considered linear and worth
248 CONTROL OF NUCLEAR REACTORS AND POWER PLANTS
10~65fc per inch of motion, then this control rod as a thermometer has the
There are many causes for a pressure change in a reactor primary sys-
Variations in the pressurizer tank level caused by the local control are
set up.
sure changes.
state basis is the build-up of poisons. It will be recalled from Secs. 2-14
and 4-4 that these poisons originate chiefly in a thermal reactor from
xenon 135 and the concentration varies both at a steady power level and
cannot isolate the depletion from the equilibrium poison build-up, and
the solid line curve of Fig. 9-4 is the one usually measured experimentally.
Figure 9-5 shows again the familiar xenon build-up and decay curve
after shutdown. In the curve shown here the peak xenon builds up to
that this peak xenon could be burned out very rapidly at full power and
the rate of change of reactivity. Figure 9-5 indicates this rate of burnout
for full power operation and a corresponding burnout rate for a partial
OPERATIONAL CONTROL PROBLEMS: POWER OPERATION
249
0 2 4 6 8 10 12 14
FIG. 9-4. Equilibrium xenon and depletion build-up versus time for a continued fixed
power-level operation.
-PEAK XENON
NORMAL DECAY
10
30
40
20
TIME, HOURS
FIG. 9-5. Peak xenon build-up and burnout for different power operations.
for the illustration given, the reactivity change drops below that caused
The equilibrium xenon depends upon the neutron flux level as indicated
in Eq. (2-26).
Generated on 2011-10-24 11:57 GMT / Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google
CONTROL OF NUCLEAR REACTORS AND POWER PLANTS
It can be seen from Fig. 9-5 that the xenon reactivity is constantly
changing and calling for rod motion. A good example of how this level
changes on a daily cyclical basis can be seen from Fig. 9-6. Here a
program is assumed such that the reactor is on for 8 hr a day and turned
off for 16 hr. It is assumed that the on time is at full power level and the
off time is at 10 percent power level, possibly needed to supply the auxil-
100
Q.
10
EXTENDED
OPERATION
AT THIS LEVEL
XENON
HOURS
16
HOURS
HOURS
16
HOURS
Generated on 2011-10-24 11:57 GMT / Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google
FIG. 9-6. Basic xenon build-up and decay for square wave on-off operation.
output from a reactor is the same as that of any other large piece of power-
ure 9-7 indicates some of the flux distributions that can be obtained with
a fictitious three-rod slab type reactor. Figure 9-7a indicates the three
rods inserted all the way into the reactor, and the flux distribution is
examined at a cross section going through the middle of the slab. For
this type of rod configuration the flux is apt to be reasonably flat. Fig-
ure 9-76 shows the shape of the flux distribution for all the rods out. The
250
OPERATIONAL CONTROL PROBLEMS: POWER OPERATION
251
Figure 9-7c illustrates a situation whereby the two outer rods are in and
the center rod is part way out. The flux distribution here indicates two
peaks in the center of the reactor. Figure 9-7d indicates a tipped flux
operation, and whenever this sort of situation appears, control rods are
usually moved to reflatten the flux. The reactor, of course, can be oper-
CONTROL
ROD
r\
REACTOR
r\
I-NEUTRON
FLUX
(a)
r\
(c) (d)
ment for changing the rods to compensate for flux distribution is involved.
Generated on 2011-10-24 11:57 GMT / Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google
An alert operator will occasionally trim rod positions to keep his reactor
9-2. Automatic Control. The required reactivity rates for power level
may seem possible to have all these items add up at the same time and
require very large reactivity rates of change. Actually, the time scales
In any event, a summation of the above items calls for a maximum rate
reactivity cannot be changed fast enough during the xenon burnout, the
loop. If this gain is too high, the system can conceivably oscillate.
Therefore the designer is faced with the problem of first summing up all
ance with the natural stability of his reactor and the required program
and comes up with a permitted maximum rod motion rate for stability.
He hopes that these two numbers overlap in the sense that his maximum
permitted reactivity for stability will provide him with plenty of reactiv-
ity change rate for his operational and safety requirements. If he has a
wide overlap, particularly one that is wide enough to take care of all his
able overlap does not exist, he has several choices. First, he may rede-
sign the control system or provide proper compensation for this system
operational restrictions on the way the reactor will be run. For example,
xenon, but rather a slower burnup rate must be used. If these opera-
Here, in the event that a combination of circumstances calls for it, rod
trol rod may be designated as a regulator rod. This rod is then tied to
inserted quickly into the reactor. A practical number for the reactivity
of a regulator rod such as 0.0065A; may be selected. That is, the max-
253
again because of the local control-loop gain. Figure 9-8 illustrates a good
regulator rod effectiveness setup in that here the regulator rod has a
restricted travel to limit its reactivity and its effectiveness does not vary
very much over its operating range. Figure 9-9 illustrates a poor regula-
tor rod setup. Here the regulator rod is forced to travel above and below
SHIM
SAFE
RO
TY^
fy
REGUl
ROD T
.A
RA
••-REGU.I
VEL ,
Al
OR RC
REACTOR
REGULATOR ROD
EFFECTIVENESS
FIG. 9-8. Regulator-rod effectiveness for shim rods all the way in.
SHI
SAFE!
OR
RODS
'
\_
Rl
RC
:GULATO
REGULATOR ROD /
EFFECTIVENESS /
V/
'
D TRAVE
REACTOR
\
CONTROL OF NUCLEAR REACTORS AND POWER PLANTS
rods may be used to follow up the regulator rod. Figure 9-10 crudely
illustrates this process. When the regulator rod approaches the end of
its permitted travel, switches are closed which operate a control system
to cause motion of the shim follow-up rods in the same direction as the
regulator rod. In this way the regulator rod will ultimately be returned
to its center position with the bulk of the reactivity being supplied by
the shim rods. For this type of operation the regulator rod has by far
the fastest motion in reactivity change per second. For stability pur-
a rate between 10 and 100 times slower than that of the regulator rod.
REGULATOR
ROD L^
\J
SHIM ROD
CONTROL
\
Generated on 2011-10-24 11:58 GMT / Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google
»
SHIM
(ODS '',
Under special design conditions a rate of only three to four times slower
255
motors are fed from a common power supply which has poor regulation.
In this way, as more and more rods are thrown onto the line, the motor
speeds become slower and the reactivity rate slows down, with the max-
imum reactivity rate obtainable being the maximum rate that permits
MAXIMUM
PERMISSIBLE RATE
14
16
0 2 4 6 8 10 12
REFERENCES
p. 58, 1950.
4. Cole, T. E.: Design of a Control System for a Low-cost Research Reactor, Nucle-
6. Moore, R. V.: The Control of a Thermal Neutron Reactor, Proc. Inst. Elec. Engrs.
once it has operated at any power level. Shutdown does not mean
level. There are two types of shutdown that are of interest. One is
the normal reactor shutdown; the second is a shutdown under duress, the
tor merely inserts the control rods at their usual rate, and the neutron-
power level falls off reasonably fast at first and then ultimately settles
down to falling off at the delayed neutron period of 80 sec. The problem
ined. Once these ground rules are laid down, there will be a brief discus-
sion of possible accidents and finally some of the elementary circuit con-
10-1. Shutdown Philosophy. There are two basic approaches that are
Because of this fact, the designers of all reactors to date have been
directly from this all-pervading need to protect the plant, coupled with
the additional factor that in early nuclear power plants many of the
256
OPERATIONAL CONTROL PROBLEMS: SHUTDOWN 257
on the need for providing continuous service. As with the case of a radio
broadcasting station, the nuclear power plant must be kept "on the air"
at all times. If the nuclear plant is being used to supply electric power
every time he plugs into it. The excuse that the reactor has just been
ophy says that the reactor must be saved in all failure events. Any plant
failure, any significant auxiliary component failure, or any item that could
always shut off as quickly as possible. The second philosophy is the one
whereby each component in the plant system must be able to take care
of itself. If the turbine fails, it must protect itself; if a pump fails, means
the nuclear industry progresses, the tendency will be away from the first
complex plant. Three basic situations are involved, all of which can
be related. The first type of incident that may require the shutting
during the course of the operation of the vessel, naval action might rup-
ture the plant system. In cases of this sort, where the primary loop may
be violently torn apart, there really is not very much that can be done
reflect back into the reactor, and any conceivable accident that creates
involving the reactor and its structure, but not the neutron level. A good
in a nuclear power plant the coolant through the reactor is suddenly lost
for some reason. The reactor may be shut down neutronwise, either by
the loss of the coolant or by scramming the rods, but the reactor might still
long-time power level in watts and PI is the beta-gamma power t sec after
conditions, and thus the reactor might destroy itself even though the
neutron level had been shut off most satisfactorily by a scramming system.
that is, the one in which the accident might be prevented by cutting down
no harm whatever to shut down the neutron level quickly in the event of
any real accident. We shall therefore consider in some detail this situa-
preventing accidents: (1) fast insertion of control rods, (2) slow movement
Fast Rod Insertion. Operating on the basis that a fast rod insertion
Generated on 2011-10-24 11:58 GMT / Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google
trol rods and found that there was usually a direct basis for setting this
speed. This basis was that each particular operating condition called
case of safety rods, however, the approach historically has been a reverse
system? the question is: How fast can one actually insert rods of a given
type into a reactor? This rate being the best obtainable has then become
the actual requirement. The time interval involved in starting the move-
ment of the rods and the rate of motion of the control rods are severely
limits in decades are likely to fit into the sort of pattern given below. To
assure getting rods fully inserted in 10~2 sec would probably require
explosives. To assure getting the rods into the reactor in lO~1 sec would
require a very complicated mechanical design, but to get the rods into
the reactor in 1 sec is relatively simple and probably would not require
OPERATIONAL CONTROL PROBLEMS: SHUTDOWN 259
more than a gravity drop. It can be seen, therefore, that most practical
tion time between 0.1 and 1 sec. Fast rod insertion requirements will
tal plaque over his desk that reads: "If nothing is permitted to move
made as slow as possible. In Chap. 8 it was pointed out that in the case
guard a plant without the use of a scram. The simple corollary is that
any reactivity rate of change should never be more than the protection
There are two reasons why the ideal safe-slow condition does not
point of view a plant that takes a long time to start up is not quite so safe
take longer than an hour or two, in which the operator sees effectively
possibly the plant is far more subcritical than it actually is. If such a
are possible. For this reason, unless the plant is fully automatic, the
control designer sadly shakes his head at the realization that his plant
The second reason for faster control, particularly in the negative reac-
tivity direction, is that despite the fact that everything about the plant
may be designed to move slowly, there are just some unknown accidents
which conceivably could create a fast condition. The fear of the conse-
protection consists of always watching the neutron level with fast instru-
that is carefully watched. Any fast rate of change involves a fast rate
This system suffers from the same difficulty as the previous ones; that
is, the failure of a fast corrective circuit is apt in itself to create a fast
accident.
of fast safety rods will now be examined. In an actual reactor plant some
rods. Another system which might be used is that all rods might be
safety rods. They are then withdrawn completely at startup, and the
ture coefficient. Still another more versatile setup exists whereby any
rod may be either a safety rod or a control rod, depending upon its elec-
type. Assume that a reactor has 16 rods and each rod is independently
not be independent in that the position of any one rod would affect the
worth of the others. The amount a reactor is shut off when all the rods
are inserted is usually not known precisely because of the variations pos-
sible from temperature and poison. Assume, for our example, that the
Generated on 2011-10-24 11:58 GMT / Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google
Let us see what proper rod manipulations might be. If it were known at
all times exactly what the shutdown reactivity was, the problem would be
down. Here, if we desired to start up our reactor, the proper rod move-
ment sequence might be that four rods would first be withdrawn and
reconnected as a safety rod bank. The rule to adopt is that the safety
bank should always contain rods worth more in reactivity than those
which are capable of being moved in the reactor at any one time. To
in the safety bank. Practically, we might not wish to move three rods
critical before three rods moved very far. However, the electrical or
mechanical interconnections should be such that with four rods out of the
conservative design dictates that possibly only two rods should be per-
mitted to move, because the hypothesis may be made that the mechanism
than 18 percent, a safer situation exists in that nine rods may be pulled
how many other rods are moved in the reactor, there is always more
reactivity in the safety bank than can be taken out of the reactor by
poisoning usually exist. Proper instrumentation can give a feel for where
the reactor is, but as has been shown in Chap. 7, there are so many var-
iables which can affect the instruments that, at the present state of the
art, only a very rough idea is available in most reactors as to the exact
shutdown status. The only obvious safe type of startup operation then,
is withdrawn.
that it is always known that a given number of rods are capable of over-
riding the loading. Under these conditions, after a reactor has been shut
the loading. Where this step is possible, safety rods are then cocked and
ready for any startup mishap. Startup accidents are likely to occur in
getting ready for startup in the process of cocking safety rods, partic-
ularly if the safety rods are capable of moving fast. It is rather intriguing
Generated on 2011-10-24 11:58 GMT / Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google
to realize that a reactor is probably safer with some of its rods out than
The question has not been discussed as to what constitutes a safe shut-
system that should be set up for a given reactor plant, the first fundamen-
is, regardless of the condition of the plant or the status of its operating
cycle, an operator seeing anything peculiar always has the choice of turn-
ing the reactor off. Large scram buttons are placed convenient to the
are placed around the control room, particularly close to exit doors.
It is assumed that for some reason the reactor gets on a fast period, the
curve of Fig. 10-1 being for a 0.1-sec period, and the power level of the
CONTROL OF NUCLEAR REACTORS AND POWER PLANTS
reactor quickly goes through the power range up through rated full power
120 percent of full power, but this level may be set at any convenient
point. When the power level hits the trip level, an electrical signal is
given out which usually has to pass through such devices as a detecting
and rod motion is started. There is a time delay between when the power
level hits the trip level and when the rods actually start to move. Fig-
ure 10-1 indicates that even when there is no time delay of this sort,
there probably is some overshoot above the trip level because the power
0.1
0.3
O.4
0.2
TIME, SECONDS
FIG. 10-1. Overpower scramming operation. Reactor on 0.1-sec period, and trip level
be inserted by the rods before the slope of the power curve is turned
around. In other words, even with no delay time, the peak power level
reached will overshoot the trip level and a direct proportionality is usually
mechanism, and the rod worths. It will also be noted from Fig. 10-1
power that is destructive but rather the energy that is involved. The
energy in this case is the area under the curve in watt seconds. When
the energy gets above a given amount, the plant is damaged. The
262
OPERATIONAL CONTROL PROBLEMS: SHUTDOWN
263
E = Kip**, (10-2)
for a given reactor on a given period. Because energy and peak power
are thus related, through Eqs. (10-1) and (10-2), it is usually simpler to
tant, as can be seen in Fig. 10-1, to cut down on the time delay involved in
rate is also important. Figure 10-2 roughly indicates the effect of scram-
ming speed upon peak power attained for a level trip operation similar to
that of Fig. 10-1. Here a fixed small time delay is presupposed, and it
TRIP LEVEL
FIG. 10-2. Peak power attained versus rod speed when scramming. Level trip with fixed
can be seen that without any scramming speed whatever, the peak power
considerations alone. This figure does not indicate whether or not this
Above a given rod speed the peak power reached always is roughly the
possible out of the delay time and starting the rods moving as quickly as
studies of this sort are needed for many types of accidents if one is to
Period Scram. We have seen that one of the difficulties "with level
scrams is that if they are fixed at a comparatively high level, they permit
the power level to attain momentum and overshoot the trip point. A
period scram, particularly in the source and period ranges, is more useful.
If the reactor starts to change its power level at a too-fast rate, the reac-
tor may be scrammed at a comparatively low level and the cause of the
difficulty located. Period scrams are not too useful in the power range,
as has been pointed out in Chap. 9. In this range the reactor is quite
The setting of the period scram trip point is not too important and
depends upon the type of reactor. Period scram settings between 1- and
periods, some reactor designers set their circuits to trip also on fast neg-
ative periods. The reasoning behind this setup is that if the reactor
nate them.
matically set up a given amount above this actual power level. This
amount above the actual power level can be on a linear percentage or log-
1,000,000 r
100,000 ?
10,000 r-
1000 =,
100 -
level.
264
OPERATIONAL CONTROL PROBLEMS: SHUTDOWN 265
arithmic basis. Then if the level of the reactor were to move slowly, the
variable level scram would move with it. That is, let the power level be at
100 units and the variable level scram initially set to trip at 200 units. If
the power level were to move slowly to 1,000 units, the variable level scram
would follow and automatically set itself up at 2,000 units. On the other
hand, if the power level were to change quickly, the variable level scram
and consequently the power level would run into the tripping level for a
fast operation.
indicated in Fig. 10-4. Let us assume that the power signal, either linear
cathode circuit, and in the event that power signal changes are very slow,
the capacitor involved charges up to the signal level plus the battery volt-
age. The grid and cathode voltages then effectively move together for
slow changes. In the event a fast voltage input change is involved, the
grid circuit changes directly but, because of the time delay, the cathode
circuit lags. Consequently, a trip will occur when the grid voltage and
It can be seen that circuits of this type are not dependent upon the
power level but rather are dependent upon the rate of change of power
fact, when most continuously variable level trip circuits are analyzed, it
turns out that they are really period circuits masquerading under another
name.
cuit. It will be recalled that the difficulty with fixed level scramming is
that the reactor has the possibilities of attaining a short period by per-
mitting its power level to rise unchecked over several decades. The
multiple fixed level scram system is indicated in Fig. 10-5. Here fixed
10,000,000 E
1,000,000
100,000
10,000
It
LJ
§ 1000
Q.
100
10
the reactor power level is brought up and fixed scram level 1 approached,
the operator checks to see that everything is under control and operating
process is gone through, and so on. In the event that the reactor gets
266
difficulty in the plant shuts off the reactor automatically, other types of
signal in the primary loop, can usually be related to power level, but there
may be a long time constant involved in the process. Some of the most
6. Pump failure
There are many other devices of this sort, all of which may be an indi-
of devices that can shut off a reactor becomes too long, it is found that
ing manner. Let us assume that we have a reactor situation in which the
safety rods are fully withdrawn and their normal scram insertion method
is by gravity drop. When the rods are fully withdrawn, we can further
assume the situation given in Fig. 10-6. Here a reactor system is pos-
tulated in which the normal travel of the safety rods is 100 in. Fig-
ure 10-6a indicates rod position versus time for a simple gravity drop
10-66. Again the first several inches are useless as far as inserting much
negative reactivity into the reactor is concerned. And when one com-
bines these two effects of the initial slow rod movement rate and the low
initial rod effectiveness, one obtains the curve of Fig. 10-6c. Here it is
quite evident that the first 200 msec in time are practically worthless in
bringing negative reactivity into the reactor. This type of rod ineffec-
circuit, and in situations such as that of Fig. 10-1, for example, a delay of
There are two methods that may be used to combat this initial ineffec-
pull the rods all the way out but to leave them partially inserted. In
the situation of Fig. 10-6, if the rods were pulled out to within 20 in. of
the top, a loss of only 10 percent of the total reactivity would be involved.
to
£ 80
~. 60
_j
LU
3 40
ac
^^"
---^
-""
^X
o 20
or
°C
Generated on 2011-10-24 11:59 GMT / Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google
_— ==
>0
^—-•-
TIME, SECONDS
100
fek 80
|1«°
£0
S^ 40
0 _J
<r <
SS 20
__ — -
>
°<
) 20 40 60 80 100
(b)
PERCENTAGE OF
ro .& o> co c
0000 O C
x^"
x^
^x^
0.1
0.2
0.5
O.6
O.7
0.3 0.4
TIME, SECONDS
(o
FIG. 10-6. Rod effectiveness in a gravity drop scramming situation, (a) Rod position
versus time, (b) Rod worth versus position, (c) Rod worth versus time.
OPERATIONAL CONTROL PROBLEMS: SHUTDOWN
269
ner. It is not important that the reactivity insertion rate slows down
from the initial rate because once the peak power has been contained,
the slope of the power versus time curve in most accidents will remain
negative.
100
5 80
ir
<n
u.
O 40
ui
5 20
or
Generated on 2011-10-24 11:59 GMT / Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google
id
0.2
0.6
0.4
TIME, SECONDS
FIG. 10-7. Rod effectiveness versus time for initial motion accelerated.
10-3. Accidents. It has been pointed out previously that there are
really three categories of accidents which can exist about a reactor plant.
was deemed advisable in this case merely because it could do no harm and
tron levels were involved and scramming would or would not be desira-
This third category breaks up into two classes: first, those classes of
limit and, second, those classes of accidents which a large negative tem-
perature coefficient would assist. The example we have been using for
the startup accident or any similar type of accident whereby the neutron
coefficient helps to limit this accident, because as the neutron level rises,
270 CONTROL OF NUCLEAR REACTORS AND POWER PLANTS
The larger the temperature coefficient for this type of accident the
better. The larger the temperature coefficient, the smaller will be the
peak power level reached by the reactor, and conversely, for a given per-
mitted peak power level, shorter periods may be used at any level or
The new case that we wish to discuss is the one whereby the larger the
when for some reason flow connections are made that would suddenly
AVERAGE COOLANT
TEMPERATURE
DROP IOO°F
TIME
switch new coolant into the system from an external storage tank. If
entrance of the new coolant into the reactor would cause the reactor
perature would now insert positive reactivity into the reactor, and the
Generated on 2011-10-24 11:59 GMT / Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google
neutron level would rapidly rise. It is conceivable that this level would
rise so fast and so far that even normal rod motions, plus the later action
change of this temperature drop are both involved in the severity of the
the relationship between power level and time for given average coolant
accident for a given reactor plant. An optimum value very clearly exists.
ative temperature coefficient would create instability and cause the sys-
oscillations.
FIG. 10-10. Optimum range of temperature coefficient to prevent startup accident and
cold-coolant accident.
limit or ward off accidents by means other than scramming. This gen-
272
2. Stop all rod motion. It has been shown that in many cases the
form a given function, once this safety condition is switched in, only neg-
the rods are driven inward either by the normal control motor with dif-
ing. Rod insertion rates of between two and ten times normal extraction
Any given reactor system can use some or all of these devices, and they
may be used in two ways. The first method is a parallel sort of operation
correction in the system is always used first and then the corrective action
rate
the systems presently employed are given in the literature.2 The dif-
ditch backup safety devices which operate in the event that all of the
ceivable, for example, that a control-rod passage might be blocked and the
chemicals with high-absorption cross sections and the use of boron shot
injected into control-rod passages are feasible. For reactors of the water
boiler type, it is also possible to dump out the fuel into an auxiliary tank
having a safe geometry. For reactors that have separate moderator and
coolant, the moderator might also be dumped. The Oak Ridge graphite
reactor. The Canadian NRX reactor can have its moderator removed.
is not fast. Presumably the damage has already been done. All the
safety methods have been tried. These devices should not be triggered
may be used.
SCRAM
MAGNET\
RELEASE
POINT
TO SCRAM MAGNET
CURRENT SUPPLY
SPRING TYPE
-MECHANISM FOR
FAST INITIAL
MOVEMENT
-CONTROL
ROD
nism.
274
sists of releasing the control rod from the direct drive and causing the rod to
OPERATIONAL CONTROL PROBLEMS SHUTDOWN
275
is the scram magnet, and this magnet is normally energized. Either the
Each control rod generally has its own magnet releasing system. How-
ever, these rod magnets may be connected together on one or more com-
described: the relay type, the magnetic-amplifier type, and the electronic
type.
MAGNET
POWER
SUPPLY
SCRAM BUS J |_
SCRAM
SIGNAL
NO. 1
AUX. 1
RELAY
lur
SCRAM
Generated on 2011-10-24 11:59 GMT / Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google
SIGNAL
NO 2
AUX. 1
RELAY
SCRAM
SIGNAL
NO. 3
AUX. ,
RELAY
y.
v,
MANUAL
SCRAMS
1'
t^O-
POWER
RELAY
RELEASE
MAGNETS
ROD
ir
RELAY
POWER
SUPPLY
..
release magnet feed circuits is the relay parallel feed arrangement. This
and may be used with any number of rods or any number of input scram-
ming signals. Because of the fact that the rod release magnets will
ber of devices may be connected to close the power relay circuit, thus
Fig. 10-12 normally have current through them and thus keep open a
This is a poor type of circuit for several reasons. From a safe failure
point of view, the power relay circuits, requiring current to release the
control-rod magnets, are backward. Safe failure requires that when the
power fails for any reason, the rods should scram. In Fig. 10-12 the auxil-
iary relays are connected properly in that if their power supply fails, they
close the scramming circuit. Back contacts are used throughout this
sort of circuit. This is not considered good relay practice, as back con-
tacts generally have less pressure available than front contacts. The
over-all circuit is quite slow because the auxiliary relays and the power
relay are effectively in series. The time delays in both relays add up,
SCRAM
SIGNAL
NO. 1
SCRAM
SIGNAL
NO. 2
and as has been pointed out in this chapter, time delays must be kept to
power relay must be closed rather than opened to scram the circuit. For
usual relay design, opening a circuit is slightly faster than closing it.
A much better relay circuit is shown in Fig. 10-13. Here all the auxil-
iary relay contacts are normally closed and are connected in series. The
power relay is normally operated with current through its coil, and a
Generated on 2011-10-24 11:59 GMT / Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google
power failure of any sort releases the magnets. Front contacts are used
throughout, but two relays are still in series, which means that this is
276
Vacuum-tube circuits are very fast and very versatile, but they are sub-
OPERATIONAL CONTROL PROBLEMS: SHUTDOWN
277
CONTROL
SIGNAL
LOAD
Many more complex circuits are also available.3 The circuit of Fig. 10-14
saturation. The load current is caused to shift abruptly from one con-
whereby the relay may be placed as close to the tripping point as desired.
CONTROL CURRENT
indicated in Fig. 10-16. Here the current to the rod release magnets is
either all on or nearly off, depending upon the state of the magnetic-
amplifier power relay. The control winding of the power relay is fed
scramming signals are fed into a scramming bus via isolating diodes.
ROD
RELEASE
MAGNETS
SCRAM
BUS
MAGNETIC
AMPLIFIER
RELAY
POWER
^1
1 r—
Generated on 2011-10-24 12:00 GMT / Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google
IONIZATION
NORMAL CONDITION
ZERO CURRENT ~\
TO ALARM
INDICATION
CIRCUIT
MAGNETIC S||
AMPLIFIER T
*~ POWER D
RELAY pll
MANUAL
SCRAMS
*X0Jv-vD__^
DETECTOR
CURRENT
.5 AMPLIFIERS'
5 MAGNETIC C
MAGNETIC
"J AMPLIFIERS"
5 AUXILIARY f
_I_|-Y~«'V"Y-V_J_
CONVENTIONAL!
RELAY
ROD
RELEASE
MAGNETS
These diodes are provided to prevent a scram in the event that any of the
supplies should be used and 400- and 800-cycle supplies are common.
278
279
The normal condition for the auxiliary relay is to supply minimum load
current. The auxiliary relay feeds a scramming bus through the coil of
which channel has caused the scram. The auxiliary relay is also isolated
now feeds the power magnetic-amplifier relay which is biased so that its
load current is normally on. This load current feeds the rod release
magnet, and manual scrams are provided in the output circuit. A chan-
SCRAMMING
"SIGMA"
"<t
.Z
in
SIGNAL
NO. 1
AMPLIFIER
SCRAMMING
"SIGMA"
ELECTRONIC
Generated on 2011-10-24 12:00 GMT / Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google
AMPLIFIER
"^||
5ll
ROD
RELEASE
MAGNET
SIGNAL
NO. 2
AMPLIFIER
SCRAMMING
"SIGMA"
MAGNET
AMPLIFIER
SIGNAL
NO. 3
AMPLIFIER
POWER
SUPPLY
ing at 400 cycles/sec and will have an over-all time delay from detector
the high speed obtainable from these circuits, special magnets with quick
Figure 10-18 shows a simplified block diagram of the basic type of cir-
cuit. Here, using the terminology originating at the Oak Ridge National
the large current requirements, more than one magnet is usually not
amplifier are fed from a separate transformer. Manual scrams and other
scrams.
SIGMA BUS
MAGNET
FIG. 10-19. Research reactor scramming system block diagram. (Courtesy of Gilbert
fed into the mixer which can accept signals from several sources. The
the output of the mixer becomes proportional to the largest signal applied
to it. The current through the release magnet is set to release the rods
280
safety circuit of this type for the Oak Ridge "swimming pool" research
reactor. One period circuit and two level circuits are shown. The POP
10-19 have been constructed with an over-all time delay from the detec-
REFERENCES
of the IRE 1954 National Convention," pt. 9, " Medical and Nuclear Electronics,"
4. Cole, T. E.: Design of a Control System for a Low-cost Research Reactor, Nucle-
8. Dahl, A., and G. Renders: Heavy Water Reactor at Kjeller, Norway, Nucleonics,
10. Breazeale, W. M.: The "Swimming Pool," a Low-cost Research Reactor, Nucle-
Generated on 2011-10-24 12:00 GMT / Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google
11. Goss, Clinton G.: Reactor Control Instruments, "Proceedings of the 1953 Con-
1953.
12. Lansing, N. F., (comp.): The Role of Engineering in Nuclear Energy Develop-
SIMULATORS
machines for the solution of reactor plant problems are called simulators
plant. Simulators are also useful to test the controls of a given plant in
though it were the plant and actual controls connected to it, sometimes
studied.
It must also be kept in mind that the simulators will be no better than
the actual equations which they simulate. If the plant is treated kinetic-
equations of these actions exhibit the same form, and any one can be used
282
SIMULATORS 283
amplifier techniques.
ElN°-
« E OUT « KE IN
*«*
passive elements: the resistor, the capacitor, and the inductor. The cur-
rents through these elements as a function of the voltage across them are
logue computing networks avoid the use of inductance and only resistive
tion generators.
quently the equations must be modified to take this loading into account.
prevent loading effects, but because of its unique properties, other val-
customary that the amplifier have no zero offset; that is, the output for
zero input is zero volts. The output is usually able to swing through a
range of at least 100 volts centered around zero. For the purposes of
Generated on 2011-10-24 12:00 GMT / Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google
the most critical features is that the amplifier should have as high an
obtained through the stability of all its components and power supplies
for correcting amplifier drift and is used when higher precision is needed.
In the amplifier of Fig. 11-2 the first stage is cathode coupled, giving the
second stage is used for two purposes. First, by applying a small volt-
for any zero offset caused by variation in tubes and component toler-
ances. Second, since this point is separated from the output by an even
here to raise the effective gain of the amplifier to a very high value.
computation.
285
voltage divider, and a similar network is inserted between the third stage
and the output stage. Capacitors are used to compensate for the input
INPUT
+300 V
OUTPUT
-300V
+ 300V
680K
6.3VAC
If
OPERATIONAL
1'1
AMPLIFIER ^^
GAIN A^XXX^
IN I
tive feedback in conjunction with this type amplifier will usually further
Zi(s) and Zf(s). Kirchhoff's law for Fig. 11-3 can be written as
(11-4)
(11-5)
''A
its associated circuits, and usually the gain of the amplifier A is very high,
Z,(«)
="
(.
of discrete networks for Z/(s), and Zi(s) and/or by using multiple inputs.
Generated on 2011-10-24 12:00 GMT / Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google
n rr f \
k=l
By proper use of Eqs. (11-6) and (11-7) many basic forms of computa-
tion can be accomplished. Figure 11-4 indicates some of the more com-
will be noticed again that the gain and the characteristics of the amplifier
have nothing to do with the process, and the accuracy of the computation
depends only upon the accuracy of the individual resistances and input
usually occurs for most output processes. The sign may be reversed
287
resistances. The output of the second amplifier will then give an inverted
ing the drift of the amplifier and the leakage of the capacitor Ct are low.
R<
o—V\AAA.—i
ETC. -
-AA/WV
(a)
= -R .
R<
E,<
R2
E2o—VWW-
ETC.-1
(b)
ETC
.VWW-
oEouT=-Rf^:(C1E1+C2E2....)
ETC.
FIG. 11-4. Elementary forms of computation circuits using operational amplifier techniques,
(a) Summation, (b) Integration, (c) Differentiation, (d) Generalized multiple input
Generated on 2011-10-24 12:00 GMT / Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google
solution.
Care must be taken in selecting only the highest quality capacitors for
RI
R= l +
Eour
RlN ,.ClN
\.
Rf
oEour
E OUT
P IN
network.
reactor that is used. For servo transient problems it is usually not neces-
detail. Instead, the basic reactor kinetic equations that we have been
which the power level of a reactor will increase or decrease depends upon
EM
f The derivations in the next three sections follow closely the original work of W.
2*9
the past history and the effective multiplication factor of the reactor.
We may use the familiar reactor kinetic equations to describe this kinetic
and put into a more suitable form for computer operation as follows:
dn Sk -
v „
n + 2, XA
dd _ PJI
dt " "F
(11-8)
(11-9)
,e
dt Li dt
8k-
that will be used to simulate a reactor, Fig. 11-6 indicates in block form
double ended, giving out two voltages which are equal in magnitude and
input grid of the amplifier. When the gain of the amplifier is very high,
any voltage on the adding bus causes the output voltage to vary in such
Figure 11-7 indicates the elementary circuit networks that perform the
R9 <
I—VWV\—>^P9
I" i
GAIN
OPERATIONAL
AMPLIFIER
£ has the dimensions of volts per neutron. From Fig. 11-7 we can sim-
7, = h -
(11-11)
We can now proceed to derive the simulator constants and scaling factors
T dE /fC7\ ,.,, dn
dt \l I dt
(11-12)
290
neutron.
SIMULATORS 291
The first term on the right-hand side of Eq. (11-10) is the reactivity
the +E end, and —1 at the —E end. Because of the amplifier action,
tap is
imum for any given E, The maximum current in Eq. (11-13) is given
tion of the simulator and the proper scaling factors by the following
method : If we set the ratio Sk/Sk^.m equal to /s/ /a.m.* and note that
(R/P)}
*=f+^f (11-17)
Generated on 2011-10-24 12:01 GMT / Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google
Pi Pi at
Matching Eqs. (11-17) and (11-18) term for term and again introducing
292 CONTROL OF NUCLEAR REACTORS AND POWER PLANTS
PiK\ «t
then becomes
dt ~ we,-/ dt
where ftiRi =
The last term of Eq. (11-10), the source term, is simulated by introduc-
ing an independent current, h, into the adding bus. From Fig. 11-7,
/, = ?f (H-23)
Kg
Now using Eq. (11-11) we can sum up the currents into the adding
/rcA dn
Generated on 2011-10-24 12:01 GMT / Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google
\l*J dt
.,
R[l ^ ,
/ r \ » y dCi
i dt
Then
In order for the output current of each term to be of the correct mag-
nitude the scaling factors of each term must equal each other. That is
I* R[l +
m 26)
('
When these electrical quantities are set up, the scaling factors in Eq.
(11-25) drop out and the equation of the simulator is the same as that of
the reactor. The scaling factor, of course, can be set at any theoretical
value, but limiting values are set to a large extent by such practical con-
rent. The currents caused by these effects must be small compared with
reactor kinetic simulator. Many such devices have been built, and com-
plete circuits as well as the details of practical operation are given in the
literature.3~6
obtain kinetic response curves such as those of Figs. 3-3 and 3-4. In the
easily be obtained.
ing negative reactivity only are being studied. For these cases a very
This type of simulator is useful for shutdown problems, and the simulator
of Sec. 11-2 may be quickly modified to this form by removing the opera-
inating the amplifier, only passive elements remain in the circuit, noise is
Figure 11-8 shows the form of this simplified subcritical reactor sim-
ulator. The delay networks and prompt neutron capacitor are simulated
through a switch Si. The equations of this circuit with the switch Si
open are
Wk + ° ~dt +
where
and the symbols have the same meaning as given in Sec. 11-2.
294 CONTROL OF NUCLEAR REACTORS AND POWER PLANTS
E = fn
Qi = Yfl
Y(-Sk)
c = Yl*
c--Y&
Ct ~ Y \<
(11-29)
The symbols Y and f are the arbitrary scaling factors. It will be noted
the adding bus. Substituting the relationships of Eq. (11-29) into Eq.
equations obviously hold for negative value of Sk, since the value of Rk,
<:
«
;R, <
.J
Generated on 2011-10-24 12:01 GMT / Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google
:«2 <
;3
:RJ «
:«4 •
.J
JR5 \
\ '•
IR6
»
Si
*.
STABLE
HIGH VOLTAGE
^ DC POWER
C SUPPLY
•* x
c,
N X"
C2
.\ *.
C3
"N *
C4
.*. *
C5
•s
ce
tive step changes in Sk, Rk is first adjusted to the desired reactivity step
value. The switch Si is closed, and all the capacitors are charged up
from the high-voltage supply to some initial voltage. When the switch
From Figs. 3-5 and 3-6 neutron decay is rapid at first but decreases slowly
after the first few decades. By using a sensitive stable d-c amplifier in a
will now be discussed. It will be recalled that the effect of poisoning was
and decay of xenon were described, and these are the equations which will
be simulated.
but for periods of many hours a mechanical device such as a gearbox may
rotation.
form.
~ / X,7 (11-30)
Equation (11-30) describes the rate of production of iodine and its sub-
of xenon from nuclear fission and iodine decay and its subsequent elim-
resents these equations is shown in Fig. 11-9. The action of the circuit
the servo amplifiers are held effectively at zero volts with respect to
put of each of the two servo amplifiers supplies the voltage for one phase
duction from the direct fission process. The currents representing xenon
CONTROL OF NUCLEAR REACTORS AND POWER PLANTS
-««
i!
•L
. GEAR
RATIO
Ni
+ £,-«—
11
AC
LINE.
SERVO
MOTOR
TACHOMETER J, |
Again using Kirchhoff's law for currents flowing into a junction the
Rl
Generated on 2011-10-24 12:01 GMT / Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google
r d6 =
dt
(11-32)
of Pi
eter PI
296
we have
Then adjusting Ri, R*, R3, and Ni so that the quantities in the paren-
theses are equal to each other, Eq. (11-34) will be like Eq. (11-30). Sim-
6 E2 cLE1_E1, « #3 , NtKzda _
grounded end
— - I- (H-36)
dX / f \ / Kt \ , / X4
Generated on 2011-10-24 12:01 GMT / Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google
^ (H-37)
By adjusting Rt, R6, Rs, R», and Nz so that the quantities in the paren-
theses are equal, Eq. (11-37) becomes the same as Eq. (11-31).
From the above simulator equations it can be seen that the derivatives
constants involved, rather large gearing ratios are required to make use
of the full speed range of the motors. Once these gearing ratios are set,
CONTROL OF NUCLEAR REACTORS AND POWER PLANTS
however, curves of X and / similar to those of Figs. 2-8, 2-9, and 8-16 can
easily be obtained.
This type of xenon simulator may be combined with the reactor kinetic
indicate transient response and not stability, the use of this combination
transients do not affect the xenon picture, and consequently the reactor
TEMP. COEFF.
REACTIVITY
THERMAL SYSTEMS. I i ^ J
/\
\ TEMPERATURE
FEEDBACK LOOP
simulators look like and having examined some of the principles behind
be recalled from this figure that the plant consists of two basic feedback
Generated on 2011-10-24 12:01 GMT / Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google
tor gives out a power level n which is changed into a thermal energy Q.
298
SIMULATORS
299
coolant from the reactor to the boiler and from the boiler to the reactor.
Coolant mixing is assumed in both the boiler inlet and the reactor inlet.
THROTTLE
MIX
first method consists of writing the differential equations for all the
the case of the reactor kinetic simulator. The second method is to derive
the transfer functions for each of the components shown in the block
cuits that match each of the transfer functions. For this example the
in Chap. 6 for each of these blocks. It will be noted that a single point
formula for the time delays between the boiler and the reactor are stipu-
CONTROL OF NUCLEAR REACTORS AND POWER PLANTS
lated. That is, it will be recalled that these transport time delays were
a pure time lag having a transfer function of the form e~â„¢. Upon series
expansion of this function and by using only the first two terms of the
FUNCTION ,+jQOV
6/ GENERATOR T
f 00001
^AC *
LINE
REACTOR
NEUTRON SECTION
each amplifier has a small amount of drift, the cumulative effect of this
numbered, but the resistors and capacitor values are only indicated.
Their values, of course, are dependent upon the specific design of the
plant involved.
300
Generated on 2011-10-24 12:01 GMT / Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google
The simulation starts with the basic diagram of the reactor kinetic
neutron simulator. The design for this section is similar to that set
forth in Sec. 11-2, with the exception that two resistance networks are
is labeled "Control rods," and the second one is a similar resistance net-
reactor feeds the (?/,(«) reactor output thermal section which is represented
in this diagram as two simple lag terms. These are shown as the circuits
and the mixing are also indicated as elementary resistance and capac-
and 10. The boiler is indicated as a lead network feeding a lag network
its feedback loop. The purpose of this function generator is to change the
added together in amplifier 13. This output is then tied back to Gh(s).
because the direct differentiation involved tends toward less stable opera-
Secs. 11-2, 11-3, and 11-4, and the simulated plant is then ready for opera-
ulation either from the direct differential equations or from the transfer
considerable time can be saved if the engineer can rely on intuitive judg-
REFERENCES
3. Bell, P., and H. Straus: Electronic Pile Simulator, Rev. Sci. Instr., vol. 21, no. 8,
August, 1950.
4. Pagels, Walter: A Portable Kinetic Simulator, AIEE Trans., vol. 70, paper 51-262,
1951.
5. O'Meara, F. E.: Reactor Simulators, /. Appl. Phys., vol. 24, no. 9, September,
1953.
Chap. 2
initially at a level n = n0, what is the kinetic equation of neutron level as a func-
tion of time?
2-3. Plot Eq. (2-20) for a reactor having a mean lifetime of 10~6 sec when a step
of reactivity of 0.002 is inserted into the reactor. How far has the reactor level
2-4. Why is the transient period of a reactor longer for a negative change in
2-5. What is the stable period of a reactor in which I* = 10~3 sec and Sk =
0.0012?
2-6. Given a reactor with a negative temperature coefficient, indicate the posi-
tion of the control rods as a function of power level assuming no coolant flow.
How do these control-rod positions compare with those of the case of constant
2-7. Derive the equations for the steady-state and peak poisoning after shut-
2-8. A given reactor has a designed excess reactivity of 5 percent and control
rods worth 7 percent. When the reactor is hot, the negative temperature coeffi-
cient affects reactivity by 3 percent. If the peak xenon poisoning after a given
shutdown is worth 9 percent, within what period of time after the shutdown
can a hot startup be accomplished? If the reactor is not started during this
Generated on 2011-10-24 12:02 GMT / Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google
time interval, how long a wait is involved before it can be started cold? Hot?
Chap. 3
3-1. What is the equation for the response of a critical reactor having I* = 10~6
3-2. How high does the level in the above problem reach at the end of one
second?
3-3. Compare the response of Prob. 3-1 with the response obtained using the
3-4. Plot the response of a reactor having I* = 10~3 sec to a linear reactivity
change of the form —0.1 + O.Qlt between 0 and 50 sec. Compare the levels
303
304 CONTROL OF NUCLEAR REACTORS AND POWER PLANTS
3-5. Plot the approximate response of a critical reactor having I* = 10~6 sec to
3-6. Derive the equation of the transfer function of a reactor having I* = 10~3
sec.
Chap. 4
4-1. What are the values of the zero frequency gain and phase shift for a reactor
I* = 10~4 sec, feedback time constant T = 0.1, and feedback gain factor KTC =
0.01?
4-2. What is the zero frequency gain of the xenon poisoning feedback-factor
4-3. Develop an electronic circuit for a comparator which is the analogue of the
4-4. Develop an expression for the gain of an ionization chamber and compare
4-6. For the simulator example of Sec. 4-10, plot the peak power attained by the
4-7. What is the transfer function of the samarium 149 poisoning effect?
Chap. 5
5-1. Given an elevated tank of water of 500-gal capacity, at what height must
the tank be placed to equal the stored energy in a steel flywheel of diameter 2 ft
Generated on 2011-10-24 12:02 GMT / Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google
5-3. What is the power required to raise a 200-lb control rod at a rate of 1 ft/sec?
5-4. A reactor is 100 in. tall and has a 50-lb safety rod initially all the way out.
Scramming is by gravity plus a 50-lb spring which acts over the first foot of rod
How long does it take for the rod to be completely inserted? How fast is the rod
reactor in 0.1 sec by moving a 25-lb regulator rod 6 in. How much horsepower is
Chap. 6
6-1. Indicate the temperature and pressure programs for a pressurized water-
cooled reactor in which the reactor-inlet temperature is held constant, and there
6-2. How much power can be transferred from the primary loop of a water-
cooled reactor to the metal in the boiler tubes under the following conditions:
PROBLEMS 305
copper boiler tube area, 10 sq ft; boiler tubes, 1 in. inside diameter; water flow,
6-3. A plate-type uranium fuel element 0.125 in. thick is operating at an output
of 1,000 cal/sec per square foot of surface. What is the rate of temperature rise of
6-4. Plot the response of the inlet temperature to a boiler as a function of time,
using the approximate delay formula of Eq. (6-11), if the reactor outlet tempera-
ture is suddenly raised 500°F and the delay time between reactor and boiler is
3 sec. Compare this curve with the current rise through a coil when a sudden
6-5. Indicate in block form the transfer function of a reactor thermal system for
a four-region reactor in which the region closest to the coolant inlet puts out one-
6-6. Graphically determine 1/[1 — KL,GL(S)] for the example used in Sec. 6-8
cient and no external control system is operating at full power output with its con-
trol rods fixed. Suddenly the pumps fail and flow stops. Sketch the behavior
and no external control system is operating at full power conditions with fixed
coolant flow. Suddenly a control rod is dropped into the reactor. Sketch the
Generated on 2011-10-24 12:02 GMT / Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google
Chap. 7
7-1. The sensitivity of a BF3 counter using enriched B10 gas is 10 counts per
unit of flux. What is its sensitivity if it is now filled with BF3 made from naturally
occurring B?
7-2. A cylindrical enriched fission counter and a B10-lined counter have the same
surface area and volume. Which tube has the highest sensitivity? Which one
7-3. A neutron has an energy of 4 ev. What is its effective velocity? With
7-4. What is the maximum counting rate that may be expected from a cylin-
drical BF3 counter having a 2-in. diameter and a 12-in. length, operating in a
7-5. The thermal flux at the surface of a 2-ft-diameter reactor core is 1012
e~°-lr/r, where r is the distance into the reflector in centimeters. Where should a
thermopile.
306 CONTROL OF NUCLEAR REACTORS AND POWER PLANTS
Chap. 8
8-1. Assume that one out of every 10' alpha particles from radon will produce
most sensitive detector indicates 30 counts per second when the plant is shut down.
How long will this instrument be able to function properly if the background
control-rod position during a reactor startup. At what rod withdrawal will the
reactor go critical?
in. counts/min
0 924
4 1,990
5 2,600
6 3,540
7 7,580
7.5 16,660
7.74 48,130
control rod is being removed from the reactor in such a manner as to produce a
linear change of reactivity of 0.0025& sec"1. On what period is the reactor after
8-5. Show that the shortest period a reactor can be on is greater than
27 In
8-6. Devise an electronic circuit which will take its inputs from three period
meters and initiate a scram if any two of them indicate a shorter period than a
demand reference.
Chap. 9
9-1. The boiler in a given reactor plant limits a change in reactor power level
from 1 percent full power to full power to a time longer than 30 sec. What is the
fastest reactor period that can be used for this level change?
9-2. A thermal reactor has an excess reactivity of 5 percent. At full power out-
put at a neutron flux of 1013, it burns up reactivity at the rate of 1 percent per year.
How long can this reactor be operated continuously at full power output?
9-3. Draw a block diagram for a control system for a shim rod bank which
Chap. 10
10-1. Given four safety rods whose reactivity values are 1, 2, 2, and 3, with
one of the rods of value 2 initially out of the reactor, devise an interlocking system
which prevents the other rods from being withdrawn unless there is more reactiv-
10-2. A power reactor has been operating at a steady output level of 100,000 kw
for a long time. Suddenly the coolant flow stops, and the reactor scrams. The
reactor and stagnant coolant can be considered to behave thermally like 1 ton of
copper. Plot the average temperature of the reactor and local coolant as a func-
10-3. How large a spring is needed to assist gravity to drop a 100-lb rod into a
0.1-sec period goes through a tripping level which is set at 120 percent of full
power, the peak power reached is 180 percent. What peak power is reached if the
10-5. For the plant of Fig. 6-2, propose a safety system of alarms, cutbacks,
and scrams.
Chap. 11
11-1. Indicate in schematic diagram form how you would perform the follow-
mixing process whose time constant is 1 sec. What is a suitable set of constants
Accumulators, 116
Alarms, 272-273
logarithmic, 210-211
equations, 139-140
level, 133
simulation, 301
temperatures, 139
Buffers, 118
automatic, 8
120-123
magnetic-feed-through, 104
rack-and-pinion, 103
105-108
winch-and-cable,104
267-269
ganging, 99
shadowing, 201-204
simulation, 300
simulation, 301
199-200
CP-2 reactor, 46
CP-3 reactor, 70
hot, 22, 26
prompt, 15
Cutbacks, 272-273
170
source, 217
Generated on 2011-10-24 12:02 GMT / Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google
table of, 14
185
Deuterium, 215-216
Differential reactivity, 27
devices, 115-118
Enthalpy, 3-5
Ergen, W. K., 54
spiral, 193-194
Grace, J. N., 71
Heat-transfer material, 11
311
Kochenburger, R. J., 73
Lipkin, H. J., 54
Loading, 261
relay, 276-279
Monitoring, 259-260
hydraulic, 109-111
spring-wound, 117
definition, 21
feedback, 50-54
20
fast, 12
intermediate, 12
thermal, 12
Nichols charts, 52
generator, 2-5
Nyquist plots, 69
stability of, 72
nonstabilized, 284
definition, 14
demand, 241-243
Primary coolant, 49
Programming, 128-135
compromise, 132-133
time-constant, 133-134
183
limitations, 246
Reactivity, 12-13, 26
259
excess, 20, 26
shutdown, 26-27
Generated on 2011-10-24 12:02 GMT / Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google
description, 10-11
fast, 12, 48
intermediate, 12, 48
prompt critical, 15
state, 14
supercritical, 17-20
discontinuous, 62
67-70
requirements, 60-62
stability, 67-69
94-95
manual, 61
mechanisms, 98-123
Reactor stability, 50
54
equations, 137-138
simulation, 301
approximate form, 46
INDEX
313
follow-up, 253-254
Shutdown, 256-281
electromechanical, 295
techniques, 83-92
xenon, 294-298
radium-berylium, 214
range, 190
removable, 219
Stability, 83
absolute, 83
plant, 166-167
relative, 83
Stable period, 19
accident, 227-236
requirements, 225
subsequent, 220-225
Steady period, 33
Steam turbine, 3
coefficient)
159, 162
Transient period, 19
170, 177
simulation, 301