Download as pdf or txt
Download as pdf or txt
You are on page 1of 93

See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/352508895

Exercises in Nuclear Reactor Analysis

Book · January 2022

CITATIONS READS

0 8,846

1 author:

W.F.G. van Rooijen


University of Fukui
67 PUBLICATIONS   499 CITATIONS   

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

IAEA CRP EBR-II SHRT benchmark analysis View project

Evaluation of self-sustaining fuel cycle in Generation IV Gas Cooled Fast Reactor View project

All content following this page was uploaded by W.F.G. van Rooijen on 23 April 2022.

The user has requested enhancement of the downloaded file.


Exercises in Nuclear
Reactor Analysis

Solutions to the exercises of the text book “Nuclear


Reactor Analysis” by James J. Duderstadt and Louis J.
Hamilton

Prepared by W.F.G. van Rooijen

Amasaka Publishing
Disclaimer!

This solution manual represents the personal efforts of the author. The author extends no
guarantee, in whatever form, of the correctness of the solutions presented in this manual.
Part I

Introductory concepts of nuclear


reactor analysis

Preview version © W.F.G. van Rooijen. Reproduction prohibited. 1


Chapter 1

An introduction into nuclear power


generation

No exercises.

Preview version © W.F.G. van Rooijen. Reproduction prohibited. 3


Chapter 2

The nuclear physics of fission chain


reactions

2.1 Exercise 2-1


What target isotope must be used for forming the compound nucleus 11Na
24
when the incident
projectile is: (a) a neutron, (b) a proton, or (c) an alpha particle?

Solution
(a) 11Na + 0n
23 1
−−→ 11Na
24

(b) 10Ne + 1p
23 1
−−→ 11Na
24

(c) 9F + 2He
20 4
−−→ 11Na
24

2.2 Exercise 2-2 (Japanese version: 2-1)


A very important type of radioactive decay process in nuclear reactors is one in which fission
products decay by neutron emission since such processes strongly influence the time behavior of
the fission chain reaction. The slowest such decay process in most reactors is one characterized by
a decay constant of 0.0126 s−1 . Assuming that such a process controls the rate at which one can
decrease the power level of a reactor, calculate the time necessary to decrease the power level from
3800 MW (thermal) to 10 MW (thermal).

Solution
Under the assumptions of the exercise, the decay of the power in the reactor is described by:

dP
= −λP
dt
which has the solution:

P (t) = P0 e−λt
Thus, the time to reach a certain power P1 is found from:

1 P0
 
P1 = P0 e−λT ⇒ T = ln
λ P1
For P0 = 3800 MW and P1 = 10 MW, one finds that T = 471 s (approximately 8 minutes (!)).

Preview version © W.F.G. van Rooijen. Reproduction prohibited. 5


CHAPTER 2. THE NUCLEAR PHYSICS OF FISSION CHAIN REACTIONS

2.3 Exercise 2-3 (Japanese version: 2-2)

Consider an initially pure sample of radioactive material whose successive decay products are
themselves radioactive with different half-lives. Write the isotopic rate equations characterizing the
concentration of various isotopes in the sample if the chain is of the form 1 → 2 → 3 → 4 → . . . N .
Solve this set of equation in a step-wise fashion for the isotopic concentrations N1 (t), N2 (t), . . .. In
particular, determine the long-time composition of the sample if the half-life of one of the isotopes
is very much longer than those characterizing other isotopes in the chain.

Solution

The system under consideration can be written as:

dN1
= −λ1 N1
dt
dN2
= λ1 N1 − λ2 N2
dt
dN3
= λ2 N2 − λ3 N3
dt
..
.
dNN
= λN −1 NN −1 − λN NN
dt

Using a Laplace transform the solution for N1 (t) is found from:

sÑ1 (s) − N1 (0) = −λ1 Ñ1 (s)


N1 (0)
Ñ1 = ⇒ N1 (t) = N1 (0) exp (−λ1 t)
s + λ1

Since the sample is initially pure, the initial concentrations for all other isotopes are zero. Thus,
the solution for N2 (t) is found as:

sÑ2 = −λ2 Ñ2 + λ1 Ñ1


λ1
(s + λ2 )Ñ2 = N1 (0)
s + λ1
λ1 N1 (0)
Ñ2 =
(s + λ2 )(s + λ1 )
λ1 N1 (0) 1 1
 
Ñ2 = −
λ2 − λ1 s + λ1 s + λ2

Proceeding for Ñ3 gives:

6 Preview version © W.F.G. van Rooijen. Reproduction prohibited.


2.3. EXERCISE 2-3 (JAPANESE VERSION: 2-2)

λ2 Ñ2
Ñ3 =
s + λ3
λ2 λ1 N1 (0) 1 1
 
Ñ3 = −
λ2 − λ1 (s + λ3 )(s + λ1 ) (s + λ3 )(s + λ2 )
λ2 λ1 N1 (0) 1/(λ1 − λ3 ) 1/(λ1 − λ3 ) 1/(λ2 − λ3 )
(
Ñ3 = − − + ...
λ2 − λ1 s + λ3 s + λ1 s + λ3

1/(λ2 − λ3 )
)
... +
s + λ2

1 1
(
Ñ3 = λ2 λ1 N1 (0) + ...
(λ2 − λ1 )(λ3 − λ1 ) s + λ1

1 1 1 1
)
... +
(λ1 − λ2 )(λ3 − λ2 ) s + λ2 (λ1 − λ3 )(λ2 − λ3 ) s + λ3
Continuing this chain for N nuclides yields the following expression for the density of isotope i
in the chain:
i
e−λj t
Ni (t) = λ1 λ2 . . . λi−1 N1 (0)
X
i
(λk − λj )
Q
j=1
k=1,k6=j

The set of coupled Ordinary Differential Equations (ODEs) for the decay of a radioactive isotope
and its radioactive daughter isotopes was first formulated by Ernest Rutherford in 1905 and solved
by Harry Bateman in 1910 using Laplace transforms, just as we have done in this exercise. The set
of ODEs is called the Bateman Equation (sometimes also called Bateman Equations). Sometimes,
the solution is (also) called the Bateman Equation.
The Bateman Equation can be extended to include so-called branch decays, where one isotope
in the chain has more than one mode of decay (for example, some radioactive isotopes may undergo
β − or β + decay). The Bateman Equation is an example of a so-called Master Equation: in physics,
chemistry and related fields, master equations are used to describe the time evolution of a system
that can be modeled as being in a probabilistic combination of states at any given time and
the switching between states is determined by a transition rate matrix. The equations are a set
of differential equations – over time – of the probabilities that the system occupies each of the
different states. In the specific case of the Bateman Equation, the transitions in the system are
only from one species to the next, i.e. from isotope i to isotope i + 1, but never in the opposite
direction.
If there is one long-lived isotope in the chain, the nuclides will ’pile up’ at that isotope. Decay
from the initial isotope to the long-lived daughter is quick, and once a nuclide decays beyond the
long-lived daughter, decay until the final (stable) isotope is also quick. Thus, if there is one long-
lived isotope in the chain, that isotope will dominate the overall composition of the sample in the
long run.
Mathematically it can be inferred from the solution of the Bateman equation: if there is one
decay constant λi which is much smaller than all other decay constants λ in the equation, then,
after a sufficiently long time, only the terms in the summation with exp(−λi t) will remain. Thus,
in the long run, the sample will contain mostly isotope i. This is illustrated in Fig. 2.1. Also note
that the time scale of decay of the total amount of radioactive nuclei in the sample is governed by
the time scale of the slowest decaying isotope.

Preview version © W.F.G. van Rooijen. Reproduction prohibited. 7


CHAPTER 2. THE NUCLEAR PHYSICS OF FISSION CHAIN REACTIONS

Exercise 2-3
1.0 Nuclide 1
Nuclide 2
Nuclide 3
0.8 Nuclide 4
Nuclide 5
Nuclide 6
Nuclide 7
Nuclide density [-]

0.6 Nuclide 8
Nuclide 9
Nuclide 10
0.4

0.2

0.0
0 2 4 6 8 10
Time [-]

(a) The case where the decay constants of all isotopes are simi-
lar.
Exercise 2-3
1.0 Nuclide 1 Nuclide 6
Nuclide 2 Nuclide 7
Nuclide 3 Nuclide 8
0.8 Nuclide 4 Nuclide 9
Nuclide 5 Nuclide 10
Nuclide density [-]

0.6

0.4

0.2

0.0
0 2 4 6 8 10
Time [-]

(b) The case where the decay constant of one isotope (nuclide 5
in this case) is much smaller than the other decay constants, that
is to say, this isotope has a much longer half-life than the other
isotopes in the decay chain. It is clear that nuclide 5 becomes
the dominant isotope in the mixture.
Figure 2.1: Illustration of Exercise 2-3. A decay chain with 10 isotopes.
Initial condition: sample contains 100% of isotope 1.

8 Preview version © W.F.G. van Rooijen. Reproduction prohibited.


Chapter 3

Fission chain reactions and Nuclear


Reactors - an Introduction

3.1 Exercise 3-1


What is the maximum value of the multiplication factor that can be achieved in any conceivable
reactor design?

Solution
The most extreme multiplication factor would be obtained if each fission neutron induces another
fission, and then k = ν. This would be a limiting case. Let’s use equation (3-13):

νΣf ΣFa
k = ηf PN L = ηf = k∞ = · ,
ΣFa Σa
under the assumption that we can make the leakage small enough. The factor η only refers to
the isotopes which are considered fuel. Obviously, we have f = 1 if there is only fuel material in
the reactor, and then we find:

νΣf
k=η= .
Σa
If there is only one fuel isotope, the maximum multiplication equals:
νσf
k=η= .
σa
In the extreme case that the fuel isotope has no capture, then one finds σa = σf , and k = ν.

3.2 Exercise 3-2


Using the alternative definition of the multiplication factor based on the concept of neutron balance,
repeat the derivation of the six-factor formula.

Solution
It is not clear what Duderstadt & Hamilton expect from this exercise. The discussion in the text
book of the four-factor formula is based on the number of neutrons from one generation to the
next. It seems reasonable to assume that this exercise is asking about a definition of the four-factor
formula (six-factor formula) in terms of reaction rates, i.e. the rates of production and destruction
of neutrons, that is to say:

Preview version © W.F.G. van Rooijen. Reproduction prohibited. 9


CHAPTER 3. FISSION CHAIN REACTIONS AND NUCLEAR REACTORS

Neutron production rate


k=
Neutron loss rate
There are many different ways to describe the processes occurring in the neutron chain reaction.
The following discussion is largely based on the discussion in [1, Chapter VII].
The four-factor formula was originally used in the design of the plutonium production reactors
constructed at the Hanford site in the state of Washington, USA, as part of the Manhattan Project.
These reactors were very large graphite moderated reactors with natural uranium fuel. In such
reactors, 95% to 99% of the fissions are due to fission of U-235 induced by thermal neutrons. The
remainder of the fission reactions, 1% to 5% of the total number of fission reactions, are due to fission
of U-238 induced by fast neutrons (fission in U-238 is a threshold reaction with E > 2 MeV). Since
these reactors were intended to produce plutonium from neutron capture in U-238, the resonance
escape probability was an important parameter: the higher the resonance escape probability, the
lower the capture rate of neutrons in resonances of U-238, and thus the lower the production of
plutonium. It was therefore important to distinguish between the processes occurring at high
energy (mainly fast fission in U-238), the number of neutrons entering the slowing down process,
the number of neutrons captured during the slowing down process (plutonium production), and
the number of neutrons emerging as thermal neutrons from the slowing down process. Since the
production reactors at Hanford were physically of very large dimension, neutron leakage could
be mostly neglected. To design reactors of physically small size, such as research reactors or
experimental piles, leakage cannot be neglected and the non-leakage probability must be included
in the derivations. The diagram given in Fig. 3.1 shows the fission chain reaction and indicates the
various factors of the four-factor (six-factor) formulas.
In a first approach, let’s assume that the reactor is sufficiently large that leakage can be ne-
glected. Let the absorption rate of thermal neutrons in the fuel be given as:

ΣFa ϕth
then the total neutron production due to thermal fissions is:

νΣFf F
Σ ϕth
ΣFa a
The newly created neutrons are fast neutrons, thus the production rate of neutrons due to fast
fission is:

νΣFf F
" #
νΣN F
Σ ϕth
f
ΣFa a

where ΣNf
F indicates the fission cross section of “Non Fuel” isotopes. Thus the total rate of

neutron production is:

νΣFf F νΣFf F  νΣF


" #

f
Σ ϕ th + νΣNF
Σ ϕth = 1 + νΣNF
ΣF ϕth
ΣaF a f
Σa
F a f
Σa a
F

As for the total absorption rate of neutrons, consider the following. The rate of absorption of
thermal neutrons is:

Σa ϕth
where Σa concerns all absorption of thermal neutrons, i.e. absorption of thermal neutrons in
the fuel and in other materials. At high energy, the absorption rate in the resonance region is a
function of the number of neutrons entering the slowing down process. Thus, for the resonance
absorption rate, one can write:

10 Preview version © W.F.G. van Rooijen. Reproduction prohibited.


3.2. EXERCISE 3-2

Thermal neutrons
absorbed in fuel

Fission neutrons due


to thermal fission

All fission neutrons

ηPF N L

Neutrons entering the


Multiplication factor
slowing down process

pηPF N L

Neutrons emerging
as thermal neutrons

pηPF N L PT N L

Thermal neutrons
absorbed in reactor

pf ηPF N L PT N L

Thermal neutrons
absorbed in fuel

Figure 3.1: The neutron chain reaction to illustrate the six factors of the
six-factor formula.

 νΣF
" #

f
Σres 1 + νΣf NF
ΣFa ϕth
a
ΣFa

Thus one finds for the total absorption rate:

 νΣF νΣFf ΣFa


" # " !#
  
f
Σa ϕth + Σres 1+ νΣN F
ΣF ϕth = Σa ϕth 1 + Σres 1+ νΣN F
a f
ΣFa a a f
ΣFa Σa

Now write the ratio of neutron production rate and neutron loss rate:

  νΣF
f
1+ νΣN F
ΣF ϕth
f
ΣFa a
k="
 νΣF ΣF
!#

f
1 + Σres 1 + νΣf
NF a
Σa ϕth
a
ΣFa Σa

Now we can identify the various factors:

Preview version © W.F.G. van Rooijen. Reproduction prohibited. 11


CHAPTER 3. FISSION CHAIN REACTIONS AND NUCLEAR REACTORS

νΣFf
η≡ F
Σa
 
 ≡ 1 + νΣN
f
F

ΣFa
f≡
Σa
!#−1
 νΣF ΣF
"

f
p≡ 1+ Σres 1+ νΣN F a
a f
ΣFa Σa

Thus we have established expressions for the four factors based on reaction rates. The non-
leakage probabilities cannot be expressed in reaction rates in a straightforward manner, thus we
will finish this exercise by noting that, in general, the non-leakage probabilities depend on the size
of the reactor, the shape of the reactor, and parameters such as the neutron mean free path.

3.3 Exercise 3-3


A spherical reactor composed of 235U metal is operating in a critical steady state. Discuss what
probably happens to the multiplication factor and why, if the system is modified in the following
ways (treat each modification separately, not cumulatively):

Solution
(a) The reactor is rapidly compressed to one-half its original volume. The reactor is a sphere of
pure 235U. By compressing the volume by a factor of two, the number density in the sphere
doublew. It can be assumed that the average energy E of the neutrons, and thus their speed v,
is not changed by this compression. By doubling the number density of material, the neutrons
suffer twice as many interactions per unit path-length as before. The outside surface area
of the sphere on the other hand is reduced by a factor of less than two (to be precise: if
the surface area of the original reactor is S1 , the surface area of the compressed reactor is
S2 = 3 1/4S1 ≈ 0.63S1 ). Assuming that the compression is instantaneous, such that the
p

neutron flux does not change, the fission rate doubles, thus p the production rate of neutrons
doubles, but the neutron leakage increases only by a factor 1/ 3 1/4 ≈ 1.59. Thus, the increase
of the neutron production rate is larger than the increase of the neutron leakage rate, and as
a result, k > 1.
Note: there is one more thing to consider in this exercise. If it were possible to compress the
reactor to half of its original volume, and to do this so quickly that the neutron flux does not
have time to change, then the question is what will happen to the neutrons that were present in
the part of the sphere which has disappeared due to the compression. Neutrons in the reactor
are propagating in between the atoms of the material, thus upon the compression, the neutrons
that were present in the outer shell are “squeezed out” of the reactor and instantaneously end
up in the vacuum surrounding the reactor. The reactor still becomes super-critical, but the
total number of neutrons in the reactor immediately after the compression is lower than in the
original reactor.

(b) A large, fat reactor operator accidentally sits on the reactor, squashing it into an ellipsoidal
shape. The sphere has the smallest surface-to-volume ratio. Thus, if the reactor is shaped
in any other shape than a sphere at constant volume, the neutron leakage will increase while
neutron production stays the same. Thus, k < 1 is the result. However, since the operator
is large and fat, we may assume that he is composed of mainly H, C, and O. Hydrogen and
graphite are good moderators and scatterers, so the operator may introduce reflection and

12 Preview version © W.F.G. van Rooijen. Reproduction prohibited.


3.3. EXERCISE 3-3

moderation into the system. The effects of both may effectively reduce the leakage, and so,
k > 1 would be the result. Without further info, it is impossible to say which effect is dominant.

(c) A thick sheet of cadmium is wrapped around the outside of the reactor. Assuming that the
reactor was originally in air, or in vacuum, it is a fair assumption that any neutron leaking
from the reactor would never return to the reactor. Wrapping any material around the reactor
will introduce some scattering, thus introducing reflection - even if the material is a strong
absorber such as cadmium. Even in a strong absorber there is a finite chance that a neutron
suffers a scattering collision. Thus some neutrons, which would otherwise have been lost from
the core will scatter in the outside layer, and some of the scattered neutrons will be scattered
back into the reactor. Thus, wrapping a sheet of cadmium around the reactor will reduce the
net leakage, and thus, paradoxically, the multiplication factor will become (very slightly) larger
than one.

(d) The reactor is suddenly immersed in a large container of water. If this happens, the water
will act as a reflector and as a moderator. The reflector effect is to reduce the leakage. The
moderator effect is to introduce low-energy neutrons into the reactor. For low energies, the
fission cross section of 235U is larger, yielding an increase of the neutron production rate. Thus,
k > 1 for this case.

(e) A source of neutrons is placed near the reactor. The source of neutrons does not change the
critical state of the reactor. Thus, k = 1. Each extra neutron which is added by the source is
multiplied, with k = 1, thus the presence of the source causes the neutron population in the
reactor to increase linearly (see also Exercise 3-9).

(f) Another identical reactor is placed a short distance away from the original reactor. Observe one
of the reactors. The reactor is exactly critical, and the addition of neutrons from the outside
world does not change the multiplication factor. It may be tempting to view the neutrons
leaking from the other reactor as a simple source of neutrons, but this is not true. Assume one
introduces one extra neutron in reactor 1. Because reactor 1 is critical, the neutron population
is permanently increased by one neutron. As a result, the leakage increases, and this will
introduce more neutrons in reactor 2. Thus, the leakage from reactor 2 will increase, which
will, in turn, increase the neutron population in reactor 1.
Let’s make a balance equation for both reactors. Since both reactors are critical, we can write:

dN1 (t)
= S1 (t) = f2→1 N2 (t)
dt
dN2 (t)
= S2 (t) = f1→2 N1 (t)
dt

with fi→j the fraction of the leaking neutrons from i arriving in j. Take the d/dt of the first
equation to find:

d2 N1 dN2 (t)
2
= f2→1 = f2→1 f1→2 N1 (t) = CN1 (t)
dt dt
The solution is then found to be
√ 
N1 (t) = exp Ct

with C the coupling coefficient between the two reactors. Thus the neutron population will
increase exponentially, with the rate of increase determined by how closely the two cores are

Preview version © W.F.G. van Rooijen. Reproduction prohibited. 13


CHAPTER 3. FISSION CHAIN REACTIONS AND NUCLEAR REACTORS

coupled. As also indicated in Exercise 3-9, an exponential increase of the neutron population
means that k > 1. Thus, in the present case, each core individually is exactly critical, but the
combined system consisting of both cores together is super-critical.

(g) One simply leaves the reactor alone for a period of time. If one does this, at some point in time
so many of the original 235U will have fissioned that the production rate of neutrons starts to
decrease. At the same time, the leakage of neutrons remains more or less constant because
leakage is mainly a geometrical effect. Thus after a while, production of fission neutrons
decreases while leakage stays more or less the same, thus k < 1.

14 Preview version © W.F.G. van Rooijen. Reproduction prohibited.


Part II

The one-speed diffusion model of a


nuclear reactor

Preview version © W.F.G. van Rooijen. Reproduction prohibited. 15


Chapter 4

Neutron Transport

4.1 Exercise 4-1


We have defined the angular neutron density n(r, E, Ω̂, t) in terms of the neutron energy E and the
direction of motion Ω̂, but one could as well define an angular density that depends instead on the
neutron velocity v, n(r, v, t). Calculate the relationship between these two dependent variables.

Solution
It is important to realize that the velocity v is a vector; as given in the exercise, both the energy
variable E and the direction variable Ω̂ have disappeared. Thus the number of neutrons with
energy between E and E + dE and direction of propagation between Ω̂ and Ω̂ + dΩ̂ should be the
same as the number of neutrons between v and v + dv, where the magnitude of v corresponds to
the energy E:

n(r, E, Ω̂, t)dEdΩ̂ = n(r, v, t)dv


We can write the velocity as:

v = v V̂
where V̂ is a unit vector expressing the orientation of the velocity vector and v the magnitude
of the velocity, which is related to the kinetic energy as:
s
1 2E
E = mv 2 ⇒ v =
2 m
Thus, we can write:

n(r, E, Ω̂, t)dEdΩ̂ = n(r, V̂ v, t)dV̂ dv


We can identify dΩ̂ and dV̂ , thus:

dE
n(r, E, Ω̂, t)dE = n(r, v, t)dv ⇒ n(r, V̂ , v, t) = n(r, E, Ω̂, t) = n(r, E, Ω̂, t)mv
dv

4.2 Exercise 4-2


Two thermal neutron beams are injected from opposite directions into a thin sample of 235U. At
a given point in the sample, the beam intensities are 1 × 1012 neutrons / cm2 s from the left and
2 × 1012 neutrons / cm2 s from the right. Compute: (a) the neutron flux and current density at
this point and (b) the fission reaction rate density at this point.

Preview version © W.F.G. van Rooijen. Reproduction prohibited. 17


CHAPTER 4. NEUTRON TRANSPORT

Solution
(a) Define êx as the unit vector in x-direction. The angular flux is given as:
h    i
ψ(Ω̂) = 1 × 1012 δ Ω̂ − êx + 2δ Ω̂ + êx

Then the scalar flux can be calculated as:


Z
ϕ= ψ(Ω̂)dΩ̂ = 3 × 1012

and the current is calculated as:


Z
J = 1 × 1012 Ω̂ψ(Ω̂)dΩ̂ = −1 × 1012 êx

(b) The reaction rate for reaction x is given as:

Rx = Σx ϕ ⇒ Rf = 3 × 1012 × Σf

4.3 Exercise 4-3


Suppose that the angular neutron density is given by
  n0
n r, Ω̂ = (1 − cos θ)

where θ is the angle between Ω̂ and the z-axis. If A is the area perpendicular to the z-axis,
then what is the number of neutrons passing through the area A per second: (a) per unit solid
angle at an angle of 45° with the z-axis, (b) from the negative z to the positive z direction, (c) net,
and (d) total?

Solution
An illustration of the geometry for this exercise is given in Fig. ??. In general, the direction vector
Ω̂ is given by two angles θ and χ, by definition (taking the angle θ from the positive z-axis):

Ωx = cos χ sin θ
Ωy = sin χ sin θ
Ωz = cos θ

and the differential surface element dΩ̂:

dΩ̂ = sin θdθdχ


In the current exercise, note that the flux is “downward” oriented: the neutron density is zero
for θ = 0 (positive z-direction) and maximal for θ = π (negative z-direction).

(a) Requested is the number of neutrons passing through the area A per unit solid angle at an
angle θ0 = 45°. From the definition in the text book, e.g. Eq. (4-32), the number of neutrons
of direction Ω̂ leaking through an elementary surface dS equals Ω̂ · dS. In this exercise, dS
is oriented along the positive z-axis, thus Ω̂ · dS = Ωz = cos θ. Since the surface is flat, and

18 Preview version © W.F.G. van Rooijen. Reproduction prohibited.


4.3. EXERCISE 4-3

χ
θ Ω̂

x
Figure 4.1: Geometry for Exercise 4-3.

none of the quantities depend on x or y, one can integrate over S, and S = A. Thus, the
R

number of neutrons passing through the (unit) area A per unit solid angle is found as:

n0 n0 1 n0 √ 
n= (1 − cos θ0 ) Ω̂ · dS = (1 − cos θ0 ) cos θ0 A = 2−1 A
4π 4π 2 4π

(b) It is a bit unclear what is meant by from the negative z to the positive z direction, but I have
interpreted it to mean the number of neutrons passing through the area A with a direction
of propagation θ < π/2, i.e. the partial current J + . Thus, we can use Eq. (4-22) from the
text book, i.e.

Z  
J+ = A dΩ̂ês · Ω̂n Ω̂
2π +

where, in this case, ês = [001]T , so that we find:

Z  
J+ = A dΩ̂Ωz n Ω̂
2π +

Writing out the integration over space angle:

Z2π π/2
n0 n0
Z
J +
=A dχ dθ cos θ sin θ (1 − cos θ) = A
4π 12
0 0

(c) I interpret this question to mean: calculate J · n̂, with n̂ the normal vector of the surface.
Thus:

Z
J · n̂ = A Ωz n(Ω̂)dΩ̂

and we find:

Preview version © W.F.G. van Rooijen. Reproduction prohibited. 19


CHAPTER 4. NEUTRON TRANSPORT

Z2π Zπ
n0
Z
Jz = A Ωz n(Ω̂)dΩ̂ ⇒ A dχ dθ(1 − cos θ) cos θ sin θ =

0 0
Zπ Zπ
n0 n0 1
A dθ cos θ sin θ − A dθ cos2 θ sin θ = −A n0
2 2 3
0 0

(d) The total number of neutrons passing through the area A can be evaluated as follows. Ac-
cording to Eq. (4-23) in the text book, the net number of neutrons passing through a surface
is defined as:
h i h i
ês · J = J + − J − ⇒ Jz = J + − J −

The total number of neutrons passing through the surface, i.e. not considering whether the
neutrons propagate in positive or negative z-direction is found as:
h i
n = J + + J − = 2J + − Jz

In part (b) J + was evaluated and in part (c) Jz was evaluated, and thus we find:

n0 n0 1
n = 2J + − Jz = 2A +A = A n0
12 3 2

20 Preview version © W.F.G. van Rooijen. Reproduction prohibited.


Chapter 5

The one-speed diffusion theory model

5.1 Exercise 5-1


Compare the derivation of the one-speed neutron diffusion equation with that for the equation of
thermal conduction, taking care to point out the assumptions and approximations used in each
case. Refer to any text on heat transfer such as those listed at the end of Chapter 12.

Solution
Thermal conduction concerns the diffusion of heat, which is a form of energy, whereas the neutron
diffusion equation concerns the diffusion of neutrons, which are particles 1 .
For the current discussion, consider a single crystal of one atomic species: a solid in which
the constituent parts, atoms, are arranged in an ordered microscopic structure, referred to as the
crystal lattice. An example could be a single crystal of iron (Fe). Various three-dimensional crystal
structures exist. For example, in Fig. 5.1 several well known crystal structures are given, e.g.
the hexagonal close packed structure (hcp), the face centered cubic (fcc) structure, and the body
centered cubic (bcc) structure of iron crystals.
There exist attractive and repulsive forces between the atoms in the crystal. These forces
between the atoms can be interpreted as if the atoms are connected to their nearest neighbors with
tiny springs, such as indicated in Fig. 5.2. The forces between the atoms cause the crystal to bond
together.
The atoms in the crystal lattice are not stationary, but rather, the atoms vibrate around their
lattice position, a phenomenon called thermal motion or thermal vibration. Since the atoms are
moving, the atoms possess kinetic energy. The temperature of the crystal is determined by the
thermal motion of the atoms. If heat is added to the crystal, the kinetic energy of the atoms
increases, that is to say, the thermal motion of the atoms increases. When heat is removed from
the crystal (i.e., cooling), the kinetic energy, and thus the thermal motion, of the atoms decreases.
Suppose now that heat is added to one corner of the crystal lattice. The thermal vibration of
the atoms in that corner increases, and since all atoms are connected to their nearest neighbors
with “atomic springs”, the thermal vibration of the nearest neighbors increases. In turn, the
nearest neighbors will influence their nearest neighbors, etc. In this way, the atomic vibration
of the atoms increases, starting from the point where extra heat (energy) is first added, and
with incremental exchange of kinetic energy (thermal vibration) between the atoms, the heat
gradually “diffuses” throughout the material. This process is known as conduction of heat. Thus
the atomic vibration transfers energy from a location with a high concentration of heat energy
(high temperature, strong atomic vibration) to a location with less heat energy, in other words,
heat (energy) propagates through the material going from a location with a high concentration of
heat (i.e. high temperature) to a location with lower concentration of heat (i.e. low temperature).
1
Perhaps one should be more precise: for the purposes of nuclear reactor physics, neutrons behave like particles.

Preview version © W.F.G. van Rooijen. Reproduction prohibited. 21


CHAPTER 5. THE ONE-SPEED DIFFUSION THEORY MODEL

B C
B
A A A

(a) Two well known crystal structures, hexagonal close packed


(hcp) on the left and face centered cubic (fcc) on the right. Note
that these structures are very similar, the difference being to
location of the atoms in the third layer. In hcp, the atoms in the
third layer are at the same location as the atoms in the first layer,
in fcc that the atoms in the third layer are in a different location.
Thus, hcp can be described as layers of . . . ABABAB . . . whereas
fcc is . . . ABC . . ..

(b) The body centered cubic (bcc) structure of iron.


Figure 5.1: Exercise 5-1: three well known crystal structures.

The magnitude of the heat transport between two locations, that is to say, the difficulty with
which heat can move throughout the material, depends on the difference of the concentration
of heat between the two locations and the specific nature of the atoms and the crystal lattice;
the “capability to exchange vibration energy” is a material property, related to the the thermal
conductivity of the material. One important assumption must be stressed: in the model of thermal
vibration, it is assumed that the displacement of the atoms due to thermal vibration is (much)
smaller than the lattice spacing between the atoms.
As discussed in the text book, neutron diffusion theory hinges on a similar assumption, that is
to say, Fick’s Law, which states that the net transfer of neutrons (particles) between two locations
depends on the difference of the neutron density (neutron concentration) between the two locations
and the ease (or difficulty) with which neutrons (particles) can travel through the material. Suppose
one “injects” a neutron into a crystal, and assume that the most likely nuclear reaction is scattering.
Each scattering reaction causes the neutron to change its direction, in a random fashion. Suppose
now that one injects not one, but a group of neutrons into the crystal. Again, each neutron will
undergo random scattering reactions. There is a finite probability for each neutron to scatter into a
direction taking the neutron into a part of the crystal where there are not yet many other neutrons.
Thus, neutrons can travel into the area of low neutron density of the crystal. But at the same
moment, the number of neutrons traveling from the area with low neutron density towards the area

22 Preview version © W.F.G. van Rooijen. Reproduction prohibited.


5.1. EXERCISE 5-1

Figure 5.2: Exercise 5-1: the “atomic spring model” of a crystal.

of high neutron density is very low, simply because there are not many neutrons in the area of low
neutron density. Thus, on average, more neutrons will travel from the area of high neutron density
to the area of low neutron density than the other way round. That is to say, on average, starting
at the point of injection, the neutrons will spread out through the crystal in much the same way
as heat spreads through the crystal. If the scatter cross section of a material is relatively small,
neutrons can travel a large distance before suffering a new scattering reaction. Thus, movement
of neutrons from one place to another is easy; if the scatter cross section is large, a neutron can
travel only a small distance between collisions, and thus the movement of neutrons is impeded.
We therefore expect that the neutron diffusion coefficient is a material property, and inversely
proportional to the scatter cross section2 . It should be noted here that it is assumed that neutrons
collide with the atomic nuclei of the atoms in the lattice, in other words, collisions between two
neutrons are neglected. Another way of saying is that it is assumed that the density of neutrons is
(much) lower than the density of atoms in the crystal.
The model of heat conduction can be written as

J = −λ∇T

and for neutron diffusion:

J = −D∇φ

There are some differences between neutron diffusion and heat diffusion. One difference is that
neutrons may be absorbed while moving through the material. If a neutron is absorbed, it does
no longer move throughout the material and thus the presence of absorption reduces the neutron
current J . On the other hand, heat cannot be “absorbed”. Another difference is that heat cannot
be exchanged to a vacuum, since heat is described as the thermal vibration of atoms, and if there
are no atoms (vacuum), there is thus no transfer of heat. Neutrons can enter from a crystal into a
2
A more detailed analysis would also take into account the effects of absorption, and then the diffusion coefficient
is found to be inversely proportional tot the total cross section.

Preview version © W.F.G. van Rooijen. Reproduction prohibited. 23


CHAPTER 5. THE ONE-SPEED DIFFUSION THEORY MODEL

vacuum. But note that neutrons in a vacuum move in straight lines, that is to say, the propagation
of neutrons in a vacuum is not a diffusion process but a transport process.
It should be noted that the above models of heat conduction and neutron diffusion are based on
a single crystal. In reality, single crystals hardly ever occur. Most materials are poly-crystalline, i.e.
composed of many, randomly ordered single crystals, or amorphous, i.e. without a specific crystal
structure. However, in all cases, the basic model remains that a material is made of microscopic
bodies, atoms or molecules, and that these microscopic bodies exchange kinetic energy due to their
mutual attractive and repulsive forces. Thus, the basic model of heat conduction remains, even
though the exact description of the exchange of kinetic energy between the microscopic bodies
is more complicated than for single crystals. In the case of fluids, i.e, materials in the liquid or
gas phase, the transport of kinetic energy of the microscopic bodies throughout the material is
complicated even further by the fact that the microscopic bodies (atoms or molecules) themselves
are moving throughout the material. As far as neutrons are concerned, the basic interaction is with
atomic nuclei, and the ordering of such nuclei (or lack thereof) on a lattice does not fundamentally
change the interaction phenomena, and as long as the velocity of the neutrons is (much) higher
than the velocity of the atoms in a gas or liquid, the atoms can be treated as being stationary with
respect to the neutrons and thus the interaction phenomena are the same as in a solid.

5.2 Exercise 5-2


By considering a plane source or absorber of neutrons located at the origin of an infinite medium,
derive the interface condition Eq. (5-15) on the neutron current density by modeling the source
term in the one-dimensional diffusion equation as Sδ(x) and then integrating this equation over an
infinitesimal region about the origin.

Solution
Consider Eq. (5-15) of the text book:
h   i
ês · J rs+ − J rs− =S

In the present case, given the fact that the geometry is one-dimensional, one has ês = [1, 0, 0]T ,
and since the interface is located at x = 0, we are asked to show:
h   i
Jx 0+ − Jx 0− =S

with Jx 0+ pointing in positive x-direction and Jx (0− ) pointing in the negative direction. Use


the fundamental theorem of calculus:

Zb
dF (x)
f (x)dx = F (b) − F (a) where = f (x)
dx
a

This gives the special case:


a+h
Z
lim f (x)dx = lim [F (a + h) − F (a)] = hf (a)
↓0 ↓0
a

where one uses the definition of the derivative:

df (x) f (x + h) − f (x)


≡ lim
dx ↓0 h
Thus we integrate the diffusion equation from x = −h to x = h to find:

24 Preview version © W.F.G. van Rooijen. Reproduction prohibited.


5.3. EXERCISE 5-3

+h +h +h


d dφ
Z Z Z
− lim D dx + lim Σa φ(x)dx = lim Sδ(x)dx
↓0 dx dx ↓0 ↓0
−h −h −h

which gives:

dφ dφ
−D |+h + D |−h + lim 2hΣa φ(0) = S
dx dx ↓0

Using the definition of the current, i.e. Jx = −Ddφ/dx, and under the condition that φ(0) is
finite (which it must necessarily be for a physically reasonable system) one then finds:
 
Jx 0+ − Jx 0− = S


QED. Note: mathematically, the current is a vector, i.e. J , i.e. the current has a magnitude
and a direction. In a one-dimensional geometry, the current vector has only one component. If one
then takes care to distinguish the direction of the current, the current can be written as a scalar
as in this solution.

5.3 Exercise 5-3


Compute the rms distance hx2 i a neutron will travel from a plane source to absorption using
p

one-speed diffusion theory. Compare this result with the rms distance to absorption in a strongly
absorbing medium (in which neutron scattering can be neglected). In particular, plot the rms
distance to absorption in water in which boron has pbeen dissolved against the boron concentration
to determine whether the diffusion theory result for hx2 i ever approaches the result characterizing
a purely absorbing medium. (Use the thermal cross section data in Appendix A.)

Solution
Consider an infinite, one-dimensional slab configuration with a plane source of neutrons. For such
a configuration, the flux is given by Eq. (5-29):

S0 L −x/L
e ϕ(x) =
2D
Thus the total rate of absorption between x and x + dx is found as:

S0 L −x/L
e Ra = Σa
dx
2D
which gives the probability of absorption between x and x + dx as:

Ra Σa L −x/L 1 −x/L
p(x)dx = = e dx = e dx
S0 2D 2L
Thus one can calculate3 :
Z∞
D E 1 1 q
x 2
= x2 e−x/L dx = 2L3 = L2 ⇒ hx2 i = L
2L 2L
0
3
See www.integrals.com:
Z∞
∞
x2 e−x/L dx = −Le−x/L 2L2 + 2Lx + x2

0
0

Preview version © W.F.G. van Rooijen. Reproduction prohibited. 25


CHAPTER 5. THE ONE-SPEED DIFFUSION THEORY MODEL

The second part of the exercise seems more complicated. From Exercise 4-19 we know the flux
at a distance x from an isotropic plane source in a purely absorbing infinite medium:

SA
ϕ(x) = E1 (Σa x)
2
Thus, using the same approach as before, the probability of absorption between x and x + dx
is found as:

Σa
p(x)dx = E1 (Σa x) dx
2
and one can then define similarly:

Z∞
D E Σa x2
x 2
= E1 (Σa x) dx
2
0

The above equation can be rewritten as:

Z∞
D E 1 1
x 2
= β 2 E1 (β) dβ, β ≡ Σa x
2 Σ2a
0

The integral over E1 can be evaluated using the recurrence relations for exponential integrals:

Z∞
1 1 h i∞
β 2 E1 (β) dβ = −β 2 E2 (β) − 2βE3 (β) − 2E4 (β)
2 Σ2a 0
0

To evaluate the value of the integral, required values of the exponential integral are shown in
Table 5.1. Note that lim xe−x = 0. We find:
x→∞
D E 1 1 2 1
x2 = =
2 Σa 3
2 3Σ2a
This leads to
a rather paradoxical situation. In a purely absorbing medium, the average distance
to absorption is x2 = 1/3Σ2a . If diffusion theory is assumed to be a correct description of neutron

behavior, then in a medium with the same absorption cross section and a non-zero scatter cross
section, the average distance to absorption is given by:
D E D 1 1
x2 = = 
Σa 3Σa [Σa + Σs (1 − µ0 )] 3Σ2a
It is counter-intuitive that a neutron would travel, on average, a larger distance from the source
until absorption in a purely absorbing medium than in a non-purely absorbing medium. The
important aspect to realize is that L measures the distance from birth of a neutron (source) until
absorption on the average. In a medium with a high scatter cross section (a necessary condition for
diffusion theory to be valid), a neutron will suffer, on average, many scattering interactions before
being absorbed. Thus the neutron will travel along a random path, sometimes moving away from
the source, and sometimes moving towards the source. As a result, a neutron will, on average,
stay relatively near the original point of emission (the source). On the hand, in a purely absorbing
medium the neutrons travel from the source until their first interaction; that first interaction is
always an absorption, so that neutrons can suffer at most one interaction. As a result, there are
only neutrons traveling in a straight line from the source until the point of absorption; as a result,
the average distance to absorption from the source is longer.
In Fig. 5.3 the requested graph is given. The graph shows the following two parameters:

26 Preview version © W.F.G. van Rooijen. Reproduction prohibited.


5.3. EXERCISE 5-3

Table 5.1: Data about the exponential integral for Exercise 5-3.

Value Value Series expansion for x → ∞


h  i
E2 (0) = 1 E2 (∞) = 0 e−x 1
x − 2
x2
+ 6
x3
− 24
x4
+ 120
x5
+O 1
x6

h  i
E3 (0) = 1
2 E3 (∞) = 0 e−x 1
x − 3
x2
+ 12
x3
− 60
x4
+ 360
x5
+O 1
x6

E4 (0) = 1
3 E4 (∞) = 0

Exercise 5-3
diffusion
pure absorber

101
x2 [cm]

100

0.0 0.1 0.2 0.3 0.4 0.5


Concentration B-10 in water [%]

Figure 5.3: Plot of x2 diffusion and x2 absorber , as requested in exercise 5-3.




Note that even for a high concentration of B – 10, that is to say, for
a medium
characterized by scattering
and strong absorption, the value of x2 is still
different from the value of x for a purely absorbing medium.
2

D E D 1
x2 = =
diffusion Σa 3Σa [Σa + Σs (1 − µ0 )]
D E 1
x2 =
absorber 3Σ2a

where the absorption cross section is defined as:


h i
Σa ≡ NH2 O (1 − e)σaH2 O + eσaB−10

For this exercise, NH2 O was calculated using the data in Appendix A as NH2 O ≡ ΣH s
2 O /σ H2 O ,
s
and the concentration of B – 10 e < 5000 ppm = 0.5% (in PWR power plants, for various reasons,
the concentration of natural boron dissolved in the coolant is limited to about e < 2000 ppm =
0.2%). As seen in the graph, even for the highest values of e, the value of L in diffusion theory is
very different from the value of L in a pure absorber.

Preview version © W.F.G. van Rooijen. Reproduction prohibited. 27


Chapter 6

Neutron Reactor Kinetics

6.1 Exercise 6-1 (Japanese version: 6-1)


At time t = 0, a decaying point source emitting S0 e−λt neutrons per second is placed at the center
of a homogeneous bare spherical reactor which is maintained in a subcritical state. Determine the
time-dependence of the neutron flux that would be measured be a detector placed outside of the
reactor. Use one-speed diffusion theory and ignore delayed neutrons.

Solution
Although this is the first exercise of the chapter, the solution is quite complicated. The exercise
mentions that the reactor is in a subcritical state, and that we can ignore delayed neutrons. That
implies that Σf 6= 0. However, if Σf ≈ 0, the spatial distribution of the neutron flux will be similar
to the distribution due to a point source in the center of a sphere. However, it is also possible that
Σf is quite large, so that the reactor is nearly critical, and in that case the spatial distribution of
the neutron flux will resemble the distribution of neutrons in a critical reactor of spherical shape.
Following the assumptions of the exercise: spherical reactor, homogeneous composition, no
delayed neutrons, in one-speed diffusion theory the reactor is described by the following diffusion
equation:

1 ∂ϕ 1 d dϕ
 
−D 2 r2 + Σa ϕ = νΣf ϕ + S(r, t)
v ∂t r dr dr
The source is described by the following expression:

S(r, t) = S0 δ(r)e−λt
If initially there are no neutrons in the reactor, then when the source is placed in the center
of the sphere, there will be a very rapid transient in which the neutrons from the source diffuse
into the sphere, causing fission reactions, and subsequently the fission neutrons will cause even
more fission reactions. Since there are no delayed neutrons, these transients are all due to prompt
neutrons, and thus these transients have a time scale on the order of 1.0 × 10−15 s to 1.0 × 10−12 s.
If we assume that the decay constant of the source is much smaller, i.e., the source decays much
more slowly then the prompt neutron transients, the spatial distribution of the neutrons will be
constant, and only the absolute number of neutrons in the reactor will decay, with a decay constant
determined by the decay constant of the source. We can use this to assert that ϕ(r, t) is separable
in time and space:

ϕ(r, t) = vn(t)ψ(r)
The flux measured by a detector outside of the sphere is proportional to the current leaking
from the sphere at the boundary. This leakage current is given as:

Preview version © W.F.G. van Rooijen. Reproduction prohibited. 29


CHAPTER 6. NEUTRON REACTOR KINETICS

1 1 ∂ϕ 1 1 dψ
 
j (R) = ϕ(R, t) − D
+
= vn(t) ψ(R) − D ≡ γvn(t)
4 2 ∂r 4 2 dr
Substituting the separable form for the flux, we find:

∂n 1 d dψ(r)
 
ψ(r) − vDn(t) 2 r2 + vΣa ψ(r)n(t) = vνΣf ψ(r)n(t) + S(r, t)
∂t r dr dr
Now we can integrate over the volume of the reactor, and define the following two parameters:

ZR
4πr2 ψ(r) dr ≡ κ
0
ZR
d dψ(r)
 
4πr2
r2 dr ≡ µ
dr dr
0

so that the equation reduces to1 :

∂n
κ − vDn(t)µ + vΣa n(t)κ = vνΣf n(t)κ + S0 e−λt
∂t
Divide by κ to find:

∂n µ
 
+ vΣa − vνΣf + vD n(t) = S1 e−λt
∂t κ
where S1 ≡ S0 /κ. This equation can be solved using the technique indicated in Appendix B of
the text book. Introducing α ≡ vΣa − vνΣf + vD µλ , we need to solve:
 

dn
+ αn(t) = S1 e−λt
dt
The solution is then found as:

S1 h −λt i
n(t) = e − e−αt
α−λ
In the case that the reactor
 is nearly critical, the solution can be written as follows. In this
1 d 2 dϕ
case, one has D r2 dr r dr ≈ −DB 2 ϕ; substitute the separable form of the flux to find:

dn(t)
ψ(r) + vDB 2 n(t)ψ(r) + vΣa n(t)ψ(r) − vνΣf n(t)ψ(r) = S0 δ(r)e−λt
dt
1
Notice the following: when integrating the source term, an integral over the δ-function appears:

ZR
−λt
S0 e 4πr2 δ(r) dr
0

The first intuition would be that this integral is equal to zero, but that is not the case. Consider that the δ-function
in spherical geometry can be defined as follows:
i−1
4 3
h
δ(r) ≡ lim π
↓0 3
Thus the integral becomes:
Z i−1
4 3
h
lim S0 e−λt 4πr2 π dr = S0 e−λt
↓0 3
0

30 Preview version © W.F.G. van Rooijen. Reproduction prohibited.


6.1. EXERCISE 6-1 (JAPANESE VERSION: 6-1)

Exercise 6-1
0.04

0.03

n(t)
0.02

0.01

0.00
0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4
Time [s]

Figure 6.1: The solution n(t) of exercise 6-1.

Now we can integrate over the volume of the reactor:

dn(t)
κ + vDB 2 n(t)κ + vΣa n(t)κ − vνΣf n(t)κ = S0 e−λt
dt
Now divide by κ:

dn   S0
= −DB 2 − Σa + νΣf vn(t) + e−λt
dt κ
Define S1 ≡ S0 /κ, introduce the definitions of diffusion theory and the definition of the neutron
lifetime:

dn νΣf
 
= n(t)vΣa − 1 − L2 B 2 + S1 e−λt
dt Σa
dn   
= n(t)vΣa k∞ − 1 + L2 B 2 + S1 e−λt
dt
dn  
= n(t)vΣa 1 + L2 B 2 (k − 1) + S1 e−λt
dt
dn k−1
 
= n(t) + S1 e−λt
dt l

This last equation can be solved using the technique indicated in Appendix B of the book.
Introducing α = (1 − k)/l, we need to solve:

dn
+ αn(t) = S1 e−λt
dt
The solution is then found as:

S1 h −λt i
n(t) = e − e−αt
α−λ

The shape of this curve is given in figure 6.1 for k = 0.98, l = 10−3 s, λ = 1 s−1 .

Preview version © W.F.G. van Rooijen. Reproduction prohibited. 31


CHAPTER 6. NEUTRON REACTOR KINETICS

6.2 Exercise 6-2


Consider a spherical assembly operating at a critical steady-state power level at time time t = 0. At
this instant a neutron burst is suddenly inserted at the center of the reactor. Derive an expression
for the length of time before the reactor flux will once again assume its fundamental mode shape
to within 10%. Plot this time versus assembly radius R for a thermal assembly moderated with
H2 O. Ignore delayed neutrons.

Solution
The effect of a neutron burst at the center of the core is that of a source S(r, t) = S0 δ(r)δ(t). Before
the burst is introduced, the reactor is operating at critical steady state, and thus the flux shape is
the fundamental mode shape (or critical flux shape). We can expand the source in eigenfunctions,
and determine the coefficient sn of each eigenfunction:

ZR
sn = c δ(r)ψn (r)dr = ψn (0), ∀n
0

Thus, the effect of the source is to excite all eigenfunctions in the reactor with the same source
term at t = 0. Since the reactor is critical, we know that after sufficiently long time, the amplitude
of all eigenfunctions will have decreased and only the fundamental mode will remain. The decay
constants of the eigenmodes are ordered, and thus, in due time, only the second mode will distort
the flux shape from the true fundamental mode, and the question then is when the amplitude of
the second mode is sufficiently small to be neglected. We furthermore know the eigenvalues for
each mode shape, because the reactor is spherical:


 
Bn2 =
R
For the decay constant of the higher order modes we use Eq. (5-202) of the text book:

λn = vDBn2 + vΣa − vνΣf


Since the reactor is critical, we know that λ1 = 0, and we find:

λ1 = vDB12 + vΣa − vνΣf = 0 ⇒ vΣa − vνΣf = −vDB12


We find for the decay constant of the second mode:

λ2 = vDB22 + vΣa − vνΣf = vD(B22 − B12 )


In order for the second eigenmode to have an amplitude smaller than some arbitrary value A:

− ln A − ln A − ln A
e−λ2 t < A ⇒ t > = =
vD(B22 − B12 )
 2 
λ2 2π π 2

vD R − R

A plot can now be prepared if values of v and D are known. For H2 O, use v = 3 × 105 m/s, D
= 0.14. A plot of t as a function of radius R is given in figure 6.2.

6.3 Exercise 6-3


(a) What is the maximum possible reactivity insertion capable in a 235U-fueled reactor? Express
your asnwer in $. (b) What is the maximum possible reactivity of a 235U-fueled reactor having a
nonleakage probability of 0.6 for fission neutrons and 0.7 for delayed neutrons?

32 Preview version © W.F.G. van Rooijen. Reproduction prohibited.


6.3. EXERCISE 6-3

Exercise 6-2
0.0175

0.0150

0.0125

0.0100

Time [s]
0.0075

0.0050

0.0025

0.0000
0 20 40 60 80 100
R [cm]

Figure 6.2: Plot for exercise 6-2.

Solution
(a) The maximum reactivity occurs for the maximum value of the multiplication factor. Thus:

1
ρmax = 1 −
kmax

We know that kmax = η, and since η ≈ 2.07 for 235


U, we find:

1
ρmax = 1 − = 0.517
η

In dollars, ρ = β equals 1$, thus ρ = 0.517 equals some 74$ for U.


235

(b) For this part, propose to use the relation k = ηpf PN L , with  = 1, p = 1, f = 1, PN L as
given in the exercise and η for the prompt neutrons and delayed neutrons; we find:

(1 − β)νσf
ηp = = (1 − β)η
σa
βνσf
ηd = = βη
σa

Thus we find for the multiplication factors:

kp,max = PN L,p (1 − β)ηmax = 0.6(1 − β)ηmax


kd,max = PN L,d βη = 0.7βηmax

and thus kmax = η(0.6(1 − β) + 0.7β) = η(0.1β + 0.6), and thus the maximum reactivity in
dollars:

1 1
 
ρ= 1−
β η(0.1β + 0.6)

Preview version © W.F.G. van Rooijen. Reproduction prohibited. 33


CHAPTER 6. NEUTRON REACTOR KINETICS

6.4 Program listings


Exercise 6-1
0 import numpy a s np
import m a t p l o t l i b . p y p l o t a s p l t
from c y c l e r import c y c l e r

monochrome = ( c y c l e r ( ’ marker ’ , [ ’ ’ , ’ x ’ , ’+ ’ , ’ 1 ’ , ’ 2 ’ , ’ 3 ’ , ’ 4 ’ , ’ . ’ ]) ∗
5 c y c l e r ( ’ l i n e s t y l e ’ , [ ’− ’ , ’−− ’ , ’ : ’ , ’ −. ’ ] ) ∗
c y c l e r ( ’ c o l o r ’ , [ ’ xkcd : dark g r e y ’ , ’ xkcd : g r e y ’ ,
’ xkcd : l i g h t g r e y ’ ] ) )

tmin = 0 . 0
10 tmax = 1 . 5
numsteps = 121
t = np . l i n s p a c e ( tmin , tmax , numsteps )

k = 0.98
15 l = 1 . 0 e−3
l l = 1.0
S1 = 1 . 0

alpha = ( 1 . 0 − k ) / l
20 n = ( S1 / ( a l p h a − l l ) ) ∗ ( np . exp(− l l ∗ t ) − np . exp(− a l p h a ∗ t ) )

fig = plt . figure ()


ax = f i g . a d d s u b p l o t ( 1 1 1 )
ax . p l o t ( t , n )
25 ax . s e t t i t l e ( ” E x e r c i s e 6−1” )
ax . s e t x l a b e l ( ” Time [ s ] ” )
ax . s e t y l a b e l ( ” $n ( t ) $ ” )
ax . g r i d ( )

30 f i g . t i g h t l a y o u t ( )
f i g . s a v e f i g ( ” e x e r c i s e 6 −1 f i g u r e . p d f ” )

fig = plt . figure ()


ax = f i g . a d d s u b p l o t ( 1 1 1 )
35 ax . s e t p r o p c y c l e ( monochrome )
ax . p l o t ( t , n )
ax . s e t t i t l e ( ” E x e r c i s e 6−1” )
ax . s e t x l a b e l ( ” Time [ s ] ” )
ax . s e t y l a b e l ( ” $n ( t ) $ ” )
40 ax . g r i d ( )

fig . tight layout ()


f i g . s a v e f i g ( ” e x e r c i s e 6 −1 f i g u r e b w . p d f ” )

45 p l t . show ( )

34 Preview version © W.F.G. van Rooijen. Reproduction prohibited.


6.4. PROGRAM LISTINGS

Exercise 6-2
0 import numpy a s np
import m a t p l o t l i b . p y p l o t a s p l t
from c y c l e r import c y c l e r

monochrome = ( c y c l e r ( ’ marker ’ , [ ’ ’ , ’ x ’ , ’+ ’ , ’ 1 ’ , ’ 2 ’ , ’ 3 ’ , ’ 4 ’ , ’ . ’ ]) ∗
5 c y c l e r ( ’ l i n e s t y l e ’ , [ ’− ’ , ’−− ’ , ’ : ’ , ’ −. ’ ] ) ∗
c y c l e r ( ’ c o l o r ’ , [ ’ xkcd : dark g r e y ’ , ’ xkcd : g r e y ’ ,
’ xkcd : l i g h t g r e y ’ ] ) )

rmin = 0 . 0 1
10 rmax = 1 0 0 . 0
numsteps = 121
r = np . l i n s p a c e ( rmin , rmax , numsteps )

v = 3 . 0 e5
15 D = 0 . 1 4
t 1 = ( 2 . 0 ∗ np . p i / r ) ∗∗ 2
t 2 = ( np . p i / r ) ∗∗ 2
t = − np . l o g ( 0 . 1 ) / ( v ∗ D ∗ ( t 1 − t 2 ) )

20 f i g = p l t . f i g u r e ( )
ax = f i g . a d d s u b p l o t ( 1 1 1 )
ax . p l o t ( r , t )
ax . s e t t i t l e ( ” E x e r c i s e 6−2” )
ax . s e t x l a b e l ( ”R [ cm ] ” )
25 ax . s e t y l a b e l ( ” Time [ s ] ” )
ax . g r i d ( )

fig . tight layout ()


f i g . s a v e f i g ( ” e x e r c i s e 6 −2 f i g u r e . p d f ” )
30
fig = plt . figure ()
ax = f i g . a d d s u b p l o t ( 1 1 1 )
ax . s e t p r o p c y c l e ( monochrome )
ax . p l o t ( r , t )
35 ax . s e t t i t l e ( ” E x e r c i s e 6−2” )
ax . s e t x l a b e l ( ”R [ cm ] ” )
ax . s e t y l a b e l ( ” Time [ s ] ” )
ax . g r i d ( )

40 f i g . t i g h t l a y o u t ( )
f i g . s a v e f i g ( ” e x e r c i s e 6 −2 f i g u r e b w . p d f ” )

p l t . show ( )

Preview version © W.F.G. van Rooijen. Reproduction prohibited. 35


Part III

The multigroup diffusion method

Preview version © W.F.G. van Rooijen. Reproduction prohibited. 37


Chapter 7

Multigroup diffusion theory

7.1 Exercise 7-1 (Japanese version: 7-1)


Estimate the fast group constants characterizing H2 O if the fast group is taken from E1 = 1 eV to
E0 = 10 MeV and the neutron energy spectrum over this group is taken as ϕ(E) ∝ 1/E.

Solution
H2 O consists of hydrogen and oxygen. These are “light” elements, so that their isotopes have very
few resonances. For a nuclear reactor, the most important nuclear reactions of hydrogen and oxygen
are√capture and scatter. The cross section of the neutron capture reaction has a 1/v-behavior (i.e., a
1/ E-behavior) outside of the resonances, whereas the scattering reaction is more or less constant
outside of the resonances (the so-called potential scattering).
By definition, the group-wise cross section for reaction type x in group g is defined as:
EZg−1

dEσx (E)ϕ(E)
Eg
σxg ≡ EZg−1

dEϕ(E)
Eg

while group-wise scatter cross section from group g 0 to group g is defined as:

EZg−1 Eg0 −1
Z
dE dE 0 σs (E 0 → E)ϕ(E 0 )
0 Eg Eg0
σsg →g ≡ Eg0 −1
Z
ϕ(E 0 )dE 0
Eg0

We find for the case of the capture reaction that σ(E) ∝ C/ E, and thus the group cross
section:
EZg−1
C0
1 1
" #
dE
E 3/2 −2C0 p −p
Eg Eg−1 Eg
σxg = EZg−1
= !
Eg−1
1 ln
dE Eg
E
Eg

Preview version © W.F.G. van Rooijen. Reproduction prohibited. 39


CHAPTER 7. MULTIGROUP DIFFUSION THEORY

and for the scattering cross section from group g to group g:

EZg−1 Eg0 −1
1
Z
σs dE dE 0
E0
0 Eg Eg0
σsg →g = Eg0 −1
= σs [Eg−1 − Eg ]
1
Z
dE 0
E0
Eg0

If it is also necessary to determine the effective neutron velocity, it can be determined as follows:

EZg−1
1 1 1
g
= g ϕ(E)dE
v ϕ v
Eg

In this last equation, we can use v = 2E/m, to find:


p

EZg−1r
1 1 m dE 1 1 1
" #

= E  =  2m p −p
2 E

v g 3/2 E
ln Eg−1
g Eg
ln Egg−1 Eg−1 Eg

In Fig. 7.1 graphs are given of the capture cross sections and the scattering cross sections of
hydrogen and oxygen. From these graphs it is clear that the assumption of 1/v-behavior for the
capture cross sections is quite good; the assumption that the scattering cross section is constant
breaks down for high energy. However, since the flux behaves as 1/E, the cross section at high
energy has a relatively small influence on the final result.

A note on the calculation of the group-wise neutron velocity


For the calculation of the multiplication factor of a nuclear reactor, the neutron velocity does not
play a role and thus there is no need to prepare group-wise neutron velocities. However, in other
types of reactor calculations, the neutron velocities do play a role, for example in the determination
of the so-called α-modes (sometimes called ω-modes) as described in [2], and in the direct numerical
calculation of neutron transients, see for example [3]. In general, the quantity of interest is not so
much the neutron velocity, but the inverse of the velocity, i.e.:

ZE2
1
ϕ(E)dE
v(E)
1 E
= 1 E
vg Z2
ϕ(E)dE
E1

where E1 and E2 are the energy boundaries of the energy group. The question is then what
kind of energy spectrum ϕ(E) should be assumed. In order of decreasing energy, it is common to
assume a fission spectrum χ(E), 1/E, and Maxwellian, where the energy boundaries are commonly
taken as E1 ≈ 1 MeV and E2 ≈ 0.625 eV. For example, the SCALE software uses E1 = 0.821 MeV
and E2 = 0.625 eV. Express the neutron velocity as:
s
2E √
v(E) = = C0 E
mn
where the constant C0 is used to take into account the units, i.e. the velocity is calculated in
cm s−1 and the energy is given in eV. The numerical value of C0 is:

40 Preview version © W.F.G. van Rooijen. Reproduction prohibited.


7.1. EXERCISE 7-1 (JAPANESE VERSION: 7-1)

(a) Capture and scatter cross section of H-1.

(b) Capture and scatter cross section of O-16.


Figure 7.1: Nuclear cross sections for the solution of Exercise 7-1.

s
2 × 1.602 176 × 10−19
C0 = 100 = 1.383 159 × 106
1.674 927 × 10−27

• High energy range: use ϕ(E) = χ(E). An expression that is often used for the fission energy
distribution is a so-called energy-dependent Watt spectrum, see for instance Eq. (2-112) in
the text book. The general form of this spectrum is:

e−E/a √ 
χ(E) = sinh bE
I
where a and b are (energy-dependent) parameters, and I is a normalization factor so that the
integral over energy of χ(E) is unity. Thus, one would the group velocity as:

ZE2 √ 
1 e−E/a
√ sinh bE dE
E I
1 E
= C0−1 1 E
vg Z2 √ 
e−E/a
sinh bE dE
I
E1

Preview version © W.F.G. van Rooijen. Reproduction prohibited. 41


CHAPTER 7. MULTIGROUP DIFFUSION THEORY

The normalization factor I thus drops out of the equation. However, while the denominator
can be written in terms of elementary functions, the primitive of the numerator cannot
be determined. Thus, this problem would be solved with numerical integration. Nuclear
data processing software such as NJOY, CALENDF and FRENDY use highly specialized
algorithms to evaluate such integrals with very high accuracy.
• Slowing down energy range: use ϕ(E) = 1/E:

ZE2
dE

E E
1 E
= C0−1 1 E
vg Z2
dE
E
E1

For consistency with the following derivation, introduce η ≡ E/kT :

Zη2


η η
1 1 η1
g
= √ η
v C0 kT Z 2 dη
η
η1

and thus:
"√√ #
η2 − η1
2 √
1 1 η1 η2
= √
v g
C0 kT ln (η2 /η1 )
• Thermal energy range: use the Maxwellian spectrum:

√ 
E

ϕ(E) = E exp −
kT
Note that in general, the Maxwellian spectrum must be properly normalized, but in the
present work, the normalization constant cancels in numerator and denominator, thus for
brevity it will not be written here. Thus one finds:

ZE2
E
 
exp − dE
kT
1 E
g
= C0−1 E 1
v Z 2√
E
 
E exp − dE
kT
E1

In this case, the numerator can be evaluated in a straightforward manner, and it is the
denominator which requires some work. Introduce η ≡ E/kT :

Zη2
e−η dη
1 1 kT η1
g
= √ η2
v C0 kT kT √
Z
ηe−η dη
η1

42 Preview version © W.F.G. van Rooijen. Reproduction prohibited.


7.2. EXERCISE 7-2 (JAPANESE VERSION: 7-2)


In the denominator, introduce θ ≡ η:

Zη2
e−η dη
1 1 kT η1
g
= √
v C0 kT kT Zθ2
2 θ2 exp(−θ2 )dθ
θ1

The denominator can be evaluated:

Zθ2
1 √ oθ2
 n
2 θ2 exp(−θ2 )dθ = 2 π erf(θ) − 2 exp(θ2 )θ
4 θ1
θ1

So that we find after some re-arrangement:

1 1 e−η1 − e−η2
= √
vg
r
π
 
C0 kT −η1
e θ1 − e−η2 θ2 + (erf(θ2 ) − erf(θ1 ))
4

In the case that the energy boundaries are such that the lower boundary is in the thermal
energy range and the upper boundary is in the slowing-down range, then one calculate the group
velocity as follows, taking the upper limit of the thermal energy range as η0 ≡ E0 /kT :

1 1 f1 + f2
= √
v g
C0 kT f3 + f4
with

"√ √ #
2 η2 − η0
f1 = √ √
C0 kT η0 η2
1
f2 = √ e−η1 − e−η0
C0 kT
f3 = ln (η2 /η0 )
r
π
 
−η1 −η0
f4 = e θ1 − e θ0 + (erf(θ0 ) − erf(θ1 ))
4

7.2 Exercise 7-2 (Japanese version: 7-2)


Estimate the minimum group spacing that will yield directly coupled multigroup equations for C,
12
2
D, 9Be, and 22Na.

Solution
Under the assumption of elastic and isotropic scatter, a neutron with initial energy Ei cannot obtain
an energy below Ef = αEi in any scattering reaction. Directly coupled groups exist if a neutron
scattering in group g can only scatter to group g + 1, but not to g + 2, g + 3, . . .. Consider thus that
a neutron scatters in group g, at an energy just above the lower boundary, i.e. Ei = Eg + . Then
the final energy of the neutron can not be lower than Ef = α (Eg + ). Thus, the lower boundary
of group g + 1 must be such that Eg+1 ≤ αEg . The value of α for the isotopes in this exercise
are given in Table 7.1. Note: for increasing nuclide mass, the group spacing for direct coupling

Preview version © W.F.G. van Rooijen. Reproduction prohibited. 43


CHAPTER 7. MULTIGROUP DIFFUSION THEORY

becomes smaller. For example, suppose there is a group boundary at 1 keV. For directly coupled
groups, in the case of deuterium the next group boundary is at 111 keV, whereas for sodium the
next group boundary would be at 840 keV. (The exercise considers 22Na, which is a bit strange
since natural sodium consists of 23Na. Therefore, in this solution 23Na is assumed.)

Table 7.1: Data for Exercise 7-2.


Isotope α
2
D 0.111
9
Be 0.640
12
C 0.716
23
Na 0.840

7.3 Exercise 7-3


What percentage of the neutrons slowing down in hydrogen will tend to skip groups if the group
structure is chosen that Eg−1 /Eg = 1/100?

Solution
For hydrogen, assume the simple scattering law from Chapter 2, i.e.

Σs (E 0 )
Σs (E 0 → E) =
E0
This scattering law does not depend on the final energy E, thus, for a neutron scattering
at energy E 0 , all final energies E < E 0 are equally likely. Assume we have the following group
structure:

1. group 1 with boundaries at E and E/102

2. group 2 with boundaries at E/102 and E/104

3. group 3 with boundaries at E/104 and E/106

The question then becomes: if a neutron scatters in group 1, what is the probability of the
neutron showing up in group 3 rather than in group 2? We can use the simple model of isotropic
scatter in hydrogen. The reaction rate for scattering reactions where a neutron scatters from an
energy E 0 to an energy E is given by:

Σs (E 0 )ϕ(E 0 ) C
R(E 0 → E) = 0
dE = 02 (7.3.1)
E E
where we have chosen Σs (E 0 ) = C and ϕ(E 0 ) = 1/E 0 . The number of neutrons ending up in
group 3 due to scatters from some energy E 0 in group 1 then equals:

E/104

C
Z
R(E 0 → g3 ) = dE (7.3.2)
E 02
E/106

To find all contributions of all starting energies E 0 in group 1, we need to integrate over group
1 to find:

44 Preview version © W.F.G. van Rooijen. Reproduction prohibited.


7.3. EXERCISE 7-3

E/104
ZE
CdE
Z
R(g1 → g3 ) = dE 0 = 99 × 99 × 10−6 = 0.9801 × 10−3 (7.3.3)
E 02
E/102 E/106

i.e. 0.9801% of the neutrons will skip a group

Preview version © W.F.G. van Rooijen. Reproduction prohibited. 45


Chapter 8

Fast spectrum calculations and fast


group constants

8.1 Exercise 8-1


The lower cutoff energy Ec for the fast region is usually chosen such that there is negligible up-
scattering above this energy. If the moderator is modeled as a free proton gas [cf. Eq. (2-107)],
compute the cutoff energy Ec such that less than 0.1% of the thermal region neutrons will be
upscattered above Ec . Plot this cutoff energy versus moderator temperature T .

Solution
I am very uncertain about this exercise. It is the first exercise of the chapter, and as such, one
expects a rather simple exercise. Also, at the time that the text book was written, the average
student certainly did not have access to computers or software for numerical evaluations, both of
which are needed for this exercise. Perhaps Duderstadt & Hamilton just included this exercise
without thinking too much, not realizing how complicated the actual solution would be. That
being said, perhaps my answer is simply incorrect, perhaps much too far-fetched.
Start out with the scatter cross section for neutrons scattering on a proton gas. Eq. (2-107) in
the text book is the scattering cross section:
 q 
σH Ef
 Esi erf kT , Ef ≤ Ei


σs (Ei → Ef ) = q 
σH Ei −Ef
 
 Esi exp erf Ei
kT , Ef > Ei


kT

The exercise asks to evaluate the number of neutrons “scattering out of the thermal range”,
that is to say, the number of neutrons gaining energy above some energy Ec in a scattering collision.
Thus, assume that the energy distribution of the neutrons in the thermal range is a Maxwellian
distribution (see also Exercise 2-15):

8
r
E E
   
φM (E, T ) = v(E)NM (E, T ) = exp −
mπkT kT kT
In Figure 8.1 the Maxwellian flux distribution is given for several temperatures. Thus, the
reaction rate of up-scattering reactions Ru bringing neutrons from Ei to Ef with Ef > Ei is found
as:
s 
8 σsH
r
Ef Ei 
 
Ru (Ei → Ef ) = exp − erf 
mπkT kT kT kT

Then the total number of up-scattering reactions with neutrons ending at Ef is found as:

Preview version © W.F.G. van Rooijen. Reproduction prohibited. 47


CHAPTER 8. FAST SPECTRUM CALCULATIONS AND FAST GROUP CONSTANTS

Maxwellian neutron flux distribution


T = 300.0 K
0.35 T = 600.0 K
T = 900.0 K
0.30 T = 1200.0 K

0.25

M(E, T) [-]
0.20

0.15

0.10

0.05

0.00
0.0 0.2 0.4 0.6 0.8 1.0
Energy [eV]

Figure 8.1: Maxwellian flux distribution for several temperatures.

 ZEf
s 
8 σsH
r
Ef Ei 

Fu (Ef ) = exp − erf  dEi
mπkT kT kT kT
0
q
The integral can be performed. Define β ≡ Ef /kT :

8 1 −β 2 2
r  
2
Fu (Ef ) = e (2β 2 − 1) erf(β) + √ e−β β
mπkT 2 π
The function Fu (Ef ) is illustrated in Figure 8.2. The fraction of up-scattering reactions bringing
neutrons to above Ec as a fraction of the total number of up-scattering reactions can be defined as:
Z∞
Fu (Ef ) dEf
E
f (Ec ) = Z∞
c

Fu (Ef ) dEf
0

At this point I get confused. It is certainly possible to carry out the integrations.
p The result is
a set of expressions involving the error function and exponentials with argument Ec /kT . Those
expressions then need to be evaluated to find the value of Ec for which f (Ec ) < 1.0 × 10−3 . Even if
it is possible to derive the necessary expressions, one would require (tedious) numerical evaluations
to determine the actual value of Ec . Even with modern numerical software the work is not easy, and
in 1976 only researchers in advanced laboratories would have had access to the necessary numerical
routines1 .
Instead of trying to find the actual value of Ec , one can inspect Figure 8.2. In this figure, Fu (Ef )
is given for various values of T . It is clear that a cut-off of Ec = 0.6 eV would be sufficient for most
practical cases (for example, an LWR reactor with a moderator temperature of Tm = 600 K; note
that in the SCALE software, the default cut-off for the thermal energy range is set at Ec = 0.625 eV).
1
Computer users are perhaps accustomed to the fact that even simple software like Excel nowadays has special
functions such as the error function erf(t) but most users do not realize the details that go on behind the scenes for the
fast, correct and efficient numerical evaluation of such functions. In 1976 there would have been very few computers
in the world with the necessary hardware to evaluate the error-function for arbitrary values of the argument in a
short time. That being said, even in 2020 it still happens that mistakes are found in the fundamental numerical
routines that provide numerical evaluation of special functions, such sine, cosine and exponential functions.

48 Preview version © W.F.G. van Rooijen. Reproduction prohibited.


8.2. EXERCISE 8-2 (JAPANESE VERSION: 8-1)

For high-temperature applications, such as a helium-cooled High Temperature Reactor (HTR), a


cut-off energy of Ec = 1.0 eV would be sufficient.

Up-scatter reactions in proton gas


0.25 T = 300.0 K
T = 600.0 K
T = 900.0 K
0.20 T = 1200.0 K

Fu(Ef, T) [-] 0.15

0.10

0.05

0.00
0.0 0.2 0.4 0.6 0.8 1.0
Energy [eV]

Figure 8.2: Number of up-scattering reactions bringing neutrons to final


energy Ef as a function of temperature.

The following should be noted: in the early days of nuclear reactor design, numerical calculations
were extremely limited and very costly. Analytical methods were used to design nuclear reactors,
and these methods were based on various assumptions and simplications. Depending on the energy
range of the neutrons, different theories were used and thus it was very important to make proper
judgement when to apply a certain theory. The distinction of the “slowing-down energe range” to
the “thermal energy range” was one of those important problems. At the time of writing of these
solutions (2020), the entire problem simply does not exist any more. Multi-group theory is used,
with sufficiently narrow energy groups that any resonances are completely resolved below about
20 eV. Calculations in the slowing down energy range are performed with transport theory and
hundreds of energy groups. The casual reactor physicist is no longer concerned with problems of
energy range. The multigroup approach obviates such considerations in most cases. There are still
a few theoretical and practical problems such as the application of scattering kernels in Monte Carlo
calculations. It has been shown that in certain cases the accuracy of the results of calculations
with the Monte Carlo method is a (strong) function of the scattering kernel.

8.2 Exercise 8-2 (Japanese version: 8-1)


Determine the neutron flux ϕ(E) resulting from an arbitrary source in an infinite hydrogenous
medium by: (a) solving the infinite medium slowing down equation with this general source term,
and then (b) using the solution obtained for a monoenergetic source as a Green’s function for the
more general problem.

Solution
The general slowing down equation in an infinite hydrogenous medium without absorption is given
by:
Z∞ Z∞
Σs (E 0 )ϕ(E 0 ) 0 F (E 0 ) 0
Σs (E)ϕ(E) = dE + S(E) ⇒ F (E) = dE + S(E)
E0 E0
E E
Take the derivative to E and rearrange to find:

Preview version © W.F.G. van Rooijen. Reproduction prohibited. 49


CHAPTER 8. FAST SPECTRUM CALCULATIONS AND FAST GROUP CONSTANTS

dF F dS
+ =
dE E dE
Now we can use the standard solution from Appendix B for this equation:
E  E 
1 dS(E ) 0 
0 1
Z Z
F (E) =  E 0 dE =  E 0 dS(E 0 )
E 0 dE E
∞ ∞
where we have used the boundary condition that lim F (E) = 0. This integral can now be
E→∞
integrated by parts to find:

ZE Z∞
 
1  E 1 1
F (E) =  E 0 S(E 0 ) ∞ − 0
S(E )dE 0
= ES(E) + S(E 0 )dE 0
E E E
∞ E
where we have used lim S(E) = 0. Thus the result is:
E→∞
Z∞
S(E) 1
ϕ(E) = + S(E 0 )dE 0
Σs (E) Σs (E)E
E
If we use the theory of Green’s functions, the flux due to a delta-source is known:

S0 (E0 ) S(E0 )
ϕ(E) = + δ(E − E0 )
Σs (E)E Σs (E0 )
and the solution follows from:
Z∞  Z∞
1 δ(E − E 0 ) 1 S(E)

ϕ(E) = + S(E 0 )dE 0 = S(E 0 )dE 0 +
Σs (E)E Σs (E) Σs (E)E Σs (E)
E E

8.3 Exercise 8-3 (Japanese version: 8-2)


• Derive an expression for the neutron balance in a nuclear reactor in terms of the scalar flux
ϕ(r, E, t) and the net current J (r, E, t).
• Derive an expression for the neutron slowing down density q(r, E, t) in terms of ϕ(r, E, t).
• Prove that in a purely hydrogenous medium and in the steady state one may write

q = E(Σt ϕ + ∇ · J − S)

where S represents the sources. You may use the lethargy variable u rather than the energy
E if you prefer. Throughout, assume E  kT .

Solution
• Hydrogenous medium, no absorption. The energy dependent diffusion equation is then given
by Eq. (7-6) from the book:

1 ∂ϕ(r, E, t)
− ∇ · D∇ϕ(r, E, t) + Σt ϕ(r, E, t) =
v ∂t
Z∞ Z ∞
0 0 0
Σs (E → E)ϕ(r, E , t)dE + χ(E) νΣf ϕ(r, E 0 , t)dE 0 + S(E)
0
E

50 Preview version © W.F.G. van Rooijen. Reproduction prohibited.


8.3. EXERCISE 8-3 (JAPANESE VERSION: 8-2)

Since J (r, E, t) = −D∇ϕ(r, E, t), we can also write:

1 ∂ϕ(r, E, t)
+ ∇ · J (r, E, t) + Σt ϕ(r, E, t) =
v ∂t
Z∞ Z ∞
Σs (E 0 → E)ϕ(r, E 0 , t)dE 0 + χ(E) νΣf ϕ(r, E 0 , t)dE 0 + S(E)
0
E

• Extending the definitions in the book:

Z∞ ZE
q(r, E, t) = Σs (E 0 → E 00 )ϕ(r, E 0 , t)dE 00 dE 0
E 0

• In the specific case of hydrogen, we can simplify q(E) as follows:

Z∞ ZE Z∞
Σs (E 0 )ϕ(E 0 ) 00 0 Σs (E 0 )ϕ(E 0 ) 0
q(E) = dE dE = E dE ⇒
E0 E0
E 0 E
Z∞
q(E) Σs (E 0 )ϕ(E 0 ) 0
= dE
E E0
E

The diffusion equation for hydrogen simplifies to (steady state, downscatter only):

Z∞
Σs (E 0 )ϕ(E 0 ) 0
∇ · J + Σt ϕ = dE + S(E)
E0
E

Thus it follows that:

q = E(Σt ϕ + ∇ · J − S)

Preview version © W.F.G. van Rooijen. Reproduction prohibited. 51


Chapter 9

Thermal spectrum calculations and


thermal group constants

9.1 Exercise 9-1 (Japanese version: 9-1)


Show that the scattering cross section σf r of a free nucleus is related to that characterizing a per-
fectly bound nucleus σb by σb = (1 + A−1 )2 σf r . (Hint: First demonstrate that σb = 4π dσ
dΩ |cos θ=1 =
b

dσf r
4π dΩ |cos θ=1 . Then use the expression for dσf r /dΩ for s-wave scattering.)

Solution
When a neutron scatters at low energy from a liquid or solid moderator, one must take into account
the fact that the neutron does not scatter from one individual, free atomic nucleus, but from an
atomic nucleus that is bound, as part of a molecule or as part of a crystal. At energies below
approximately 1 eV, the energy levels of molecular vibrations are comparable to the amount of
energy exchanged in a scattering collision. Any scattering event causes excitation of molecular
and/or crystal vibrations, and if the kinetic energy of scattering neutron is sufficiently low, the
energy transfered to the molecular or crystal vibrations is not negligible compared to the change of
the kinetic energy of the neutron. As a result, the scatter kinematics become more intricate than
in the case of scattering of a free atomic nucleus.
Consider that a neutron scatters of a free atomic nucleus; such a scattering collision is isotropic in
the Center-of-Mass (COM) system, and thus anisotropic in the Laboratory system (LAB system). If
a neutron scatters off an atomic nucleus, which is tightly bound into a molecule, then the scattering
is isotropic in the COM system, but in this case, the COM system is basically tied to the molecule
if the molecule is sufficiently heavy, and thus the COM and LAB systems (more or less) coincide.
Thus, the scattering reaction of a molecule is different from scattering of a free atomic nucleus.
 
Introduce µ = cos Ω̂ · Ω̂0 . Let µ0 occur in a scattering reaction from a free nucleus, and µ
in a scattering reaction from a bound nucleus. Measured in the respective COM systems, both
scattering reactions are isotropic and thus:

σf r
σf r µ0 =


σb
σb (µ) =

where σf r and σb are the respective total scattering cross sections. In general, the following
relation holds:

Preview version © W.F.G. van Rooijen. Reproduction prohibited. 53


CHAPTER 9. THERMAL SPECTRUM CALCULATIONS AND THERMAL GROUP
CONSTANTS

σf r µ0 dµ0 = σf r (µ) dµ


σb µ0 dµ0 = σb (µ) dµ


because the scattering kinematics dictate that the number of neutrons scattered into an angle
dµ is the same whether that angle is measured in the COM system (µ0 ) or in the LAB system (µ).
If the scattering reaction is such that µ = 1, no energy is transferred and thus the scattering cross
section is the same for bound and free nuclei:

σf r (µ = 1) = σb (µ = 1)
and thus:

σb = 4πσb (µ = 1) = 4πσf r (µ = 1)
 dµ0 dµ0
= 4πσf r µ0 |µ=µ0 =1 = σf r |µ=µ0 =1
dµ dµ

In Chapter 2 a relation was established between the scattering angles in the COM and LAB
systems:

1 + Aµ0
µ= q
1 + 2Aµ0 + A2
and thus:

1
 2
σb = 1 + σf r
A

9.2 Exercise 9-2 (Japanese version: 9-2)


Generalize the definition of the slowing down density q(r, E) given by Eq. (8-18) to account for
upscattering as well as downscattering.

Solution
Define q(E) as the net number of neutrons passing energy E. These neutrons may be scattering
down or scattering up. First, consider the number of neutrons scattering down from some energy
E 0 > E to some energy E 00 < E:

Z∞
Nd = Σds E 0 → E φ E 0 dE 0
 

where the notation Σds (E 0 → E) indicates that the scattering kernel for down-scatter events is
used. Then the total number of neutrons slowing down past E becomes:

ZE Z∞
qd (E) = Σds E 0 → E φ E 0 dE 0 dE 00
 

0 E

Similarly, we can define the number of neutrons speeding up from some energy E 0 < E to some
energy E 00 > E:

54 Preview version © W.F.G. van Rooijen. Reproduction prohibited.


9.3. EXERCISE 9-3

Z∞
Nu = Σus E 0 → E φ E 0 dE 0
 

so that the total number of neutrons speeding up past E is given by:

ZE Z∞
qu (E) = Σus E 0 → E φ E 0 dE 0 dE 00
 

0 E

and thus the net transfer of neutrons across energy E:

ZE Z∞ ZE Z∞
0 0 0 00
q(E) = qd (E) − qu (E) = Σds E → E φ E dE dE − Σus E 0 → E φ E 0 dE 0 dE 00
  

0 E 0 E

Notice the symmetry in this expression. In fact, if the scattering kernel is symmetric for up-
and down-scatter, then q(E) = 0, which is another way of stating the balance condition of Eq.
(9-9). The condition q(E) = 0 does not imply the principle of detailed balance.

9.3 Exercise 9-3


Using the definition of the slowing down density for energies E > Ec in the infinite medium
spectrum equation (9-1), derive the balance condition Eq. (9-4).

Solution
By definition,

Z∞ ZEc
0
q(Ec ) = dE dE 00 Σs (E 0 → E 00 )ϕ(E 0 )
Ec 0

In neutron thermalization problems, fission neutrons appear at high energy, and can only end
up in the thermal energy range by scattering and losing energy. Thus, the source S(E) of neutron
appearing directly in the thermal range is zero. Using this knowledge, we can write down the
balance equation:

h i Z∞
Σa (E) + Σs (E) ϕ(E) = Σs (E 0 → E)ϕ(E 0 )dE 0
0

Now we can integrate this expression over the entire thermal neutron energy range:

ZEc ZEc ZEcZ∞


Σa (E)ϕ(E)dE + Σs (E)ϕ(E)dE = Σs (E 0 → E)ϕ(E 0 )dE 0 dE
0 0 0 0

Now we can use the identity Equation (9-2):

Z∞
Σs (E) = Σs (E → E 00 )dE 00
0

to find:

Preview version © W.F.G. van Rooijen. Reproduction prohibited. 55


CHAPTER 9. THERMAL SPECTRUM CALCULATIONS AND THERMAL GROUP
CONSTANTS

ZEc ZEcZ∞ ZEcZ∞


00 00
Σa (E)ϕ(E)dE + Σs (E → E )ϕ(E)dE dE = Σs (E 0 → E)ϕ(E 0 )dE 0 dE
0 0 0 0 0

In the scatter integral on the left hand side, we can use the fact that there will not be upscatter
to above Ec to truncate the integral over the outgoing energies:

ZEc ZEcZEc ZEcZ∞


00 00
Σa (E)ϕ(E)dE + Σs (E → E )ϕ(E)dE dE = Σs (E 0 → E)ϕ(E 0 )dE 0 dE
0 0 0 0 0

On the right hand side, we can now break the integral over infinity into two integrals:

ZEc ZEcZEc
Σa (E)ϕ(E)dE + Σs (E → E 00 )ϕ(E)dE 00 dE =
0 0 0
ZEcZEc ZEcZ∞
0 0 0
Σs (E → E)ϕ(E )dE dE + Σs (E 0 → E)ϕ(E 0 )dE 0 dE
0 0 0 Ec

Now, in the scatter integral on the left hand side do a substitution of variables: E = E 0 , E 00 = E,
to find:

ZEc ZEcZEc
Σa (E)ϕ(E)dE + Σs (E 0 → E)ϕ(E 0 )dE 0 dE =
0 0 0
ZEcZEc ZEcZ∞
0 0 0
Σs (E → E)ϕ(E )dE dE + Σs (E 0 → E)ϕ(E 0 )dE 0 dE
0 0 0 Ec

and thus:

ZEc ZEcZ∞
Σa (E)ϕ(E)dE = Σs (E 0 → E)ϕ(E 0 )dE 0 dE = q(Ec )
0 0 Ec

56 Preview version © W.F.G. van Rooijen. Reproduction prohibited.


Chapter 10

Cell calculations for heterogeneous


core lattices

Note: the text book contains two exercises 10-24!

10.1 Exercise 10-1 (Japanese version: 10-1)


Determine the ratio of neutron mfp to lattice dimension (e.g., fuel-pin diameter or lattice pitch) in
LWR, HTGR, and LMFBR cores for both fast and thermal neutrons. Also compare the ratio of
core size to migration length in these reactors.

Solution
I have used the SCALE software to perform a lattice calculation of a typical PWR fuel assembly,
as well as one fuel pin of a metal-fueled fast reactor. I did not have access to a good model for
an HTGR so I will not cite numbers for that type of reactor. See Table 10.1. I used the diameter
of the fuel pellet as the typical dimension. For the LMFBR, I have not listed thermal values as
these are irrelevant. For the PWR, the MFP in the thermal range is less than a centimeter, and
the size of a fuel pellet is comparabel to the MFP. As far as the core size is concerned, the typical
dimension of a PWR core is on the order of H = D = ∼ 400 cm and thus the core size is hundreds
of MFP; in an LMFBR, the typical core dimension is H ∼ = 100 cm and D ∼ = 200 cm, and thus the
core size is tens of MFP. Thus an LMFBR core is “neutronically” much smaller than a PWR core.

Table 10.1: Table for Exercise 10-1.


Parameter PWR LMFBR HTGR
λ fast [cm] 1.9393 3.5452 -
λ thermal [cm] 0.812 92 - -
Dfuel [cm] 0.85 0.7 -
λ/Dfuel fast 2.2815 5.0646 -
λ/Dfuel thermal 0.953 68 - -

10.2 Exercise 10-2 (Japanese version: 10-2)


Derive an expression for the thermal utilization of a three-region lattice cell (including fuel, clad,
and moderator) in terms of thermal disadvantage factors. Also derive an expression for the cell-
averaged group constants characterizing this cell in terms of the appropriate disadvantage factors.

Preview version © W.F.G. van Rooijen. Reproduction prohibited. 57


CHAPTER 10. CELL CALCULATIONS FOR HETEROGENEOUS CORE LATTICES

Solution
Start out with a general definition of the thermal utilization factor:
Z
drΣFa (r)φ(r)
f=Z Z Z
drΣFa (r)φ(r) + drΣC
a (r)φ(r) + drΣM
a (r)φ(r)

Given that the fuel, clad, and moderator regions are piecewise homogeneous, this becomes:
Z
ΣFa drφ(r)
VF
f= Z Z Z
ΣFa drφ(r) + ΣC
a drφ(r) + ΣM
a drφ(r)
VF VC VM

Introducing VF φF = drφ(r) and similarly for the cladding and moderator, one has:
R
VF
ΣFa VF φF
f=
ΣFa VF φF + ΣC
a VC φC + Σa VM φM
M

Now introduce two disadvantage factors:

φC
ζC ≡
φF
φ
ζM ≡ M
φF

to make our final definition:

ΣFa
f=
ΣFa (r) + ΣC
a (VC /VF ) ζC + Σa (VM /VF ) ζM
M

For the group cross sections, use the general definition:


Z
drΣx,g (r)φ(r)
Σx,g = Z
drφ(r)

If the fuel, clad and moderator regions are piecewise homogeneous, one can write:

VF ΣFx,g φF + VC ΣC
x,g φC + VM Σx,g φM
M
Σx,g =
VF φF + VC φC + VM φM
Or, in terms of the disadvantage factors:

ΣFx,g + ΣCx,g (VC /VF ) ζC + Σx,g (VM /VF ) ζM


M
Σx,g =
1 + (VC /VF ) ζC + (VM /VF ) ζM

10.3 Exercise 10-3 (Japanese version: 10-3)


Show that the diffusion length characterizing a heterogeneous lattice cell can be written approxi-
mately as L2 ∝ L2M (1 − f ), where f is the thermal utilization for the cell. Does the nonleakage
probability PT N L characterizing a bare, uniform core increase or decrease when going from a ho-
mogeneous to a heterogeneous lattice?

58 Preview version © W.F.G. van Rooijen. Reproduction prohibited.


10.3. EXERCISE 10-3 (JAPANESE VERSION: 10-3)

Solution
The following definitions apply:

DF
L2F =
ΣFa
DM
L2M = M
Σa
D
L2 =
Σa

where the last equation concerns the homogenized cell. A straightforward definition of L2 could
be as follows:

DF DM
L2 = VF φF + M VM φM
Σa
F Σa
Now assume that ΣFa is large, and that VM  VF (which is often the case, especially for graphite
moderated reactors), so that one obtains:

DM
L2 ≈ V M φM
ΣM
a
We can improve this estimation by replacing the absorption cross section with the cell-averaged
absorption cross section:

DM
L2 ≈ V M φM
ΣFa VF φF + ΣM
a V M φM
This can be written as:

DM ΣM
a VM φM
L2 ≈ = L2M (1 − f )
Σa Σa V F φF + Σ M
M F
a VM φ M

Preview version © W.F.G. van Rooijen. Reproduction prohibited. 59


Part IV

An introduction to nuclear reactor


core design

Preview version © W.F.G. van Rooijen. Reproduction prohibited. 61


Chapter 11

General aspects of nuclear reactor


core design

No exercises in this chapter.

Preview version © W.F.G. van Rooijen. Reproduction prohibited. 63


Chapter 12

Thermal-hydraulic analysis of nuclear


reactor cores

12.1 Exercise 12-1 (Japanese version: 12-1)


Compare the average thermal power densities (q 0 , q 00 , q 000 ) of each of the major reactor types (e.g.,
PWR, BWR, HTGR, LMFBR, GCFR, ...) along with that of the boiler of a fossil-fuel-fired plant.
Use data from PSAR’s for actual plants, if possible.

Solution
In 2020, it is surprisingly difficult to find objective data about fossil fueled power plants in the public
domain1 . Therefore we will list the relevant data from Appendix H in Table 12.1 and provide some
discussions.

• The process of designing a machine, whether it is an automobile, aeroplane, or a nuclear reac-


tor, takes several steps. The first step is the statement of the (desired) specifications. In the
case of a nuclear power plant, these specifications include the desired electrical power output
(in most cases, as high as technically feasible), the desired fuel type (as of 2020 this choice
is limited to low-enriched UO2 with very few exceptions), the operating period (commonly
an economic lifetime of 40 years, although there are reactor designs for 60 years and even 80
years of economic lifetime; plant operation is usually on a 12-month cycle, although modern
nuclear power plants can be operated in 18-month cycles and there are proposals to go to
24-month cycles), the desired average availability factor (availability factors vary between
markets; 70% is common in Japan, whereas power plants in France often reach an availability
of 95%), and a maximum cost of a unit of electricity produced by the plant. Often the desired
reactor type (PWR, BWR, CANDU, etc) is also written into the specifications; the desire
for a specific reactor type can be due to existing experience, availability of the manufacturer,
local regulatory framework, etc.

• Once the specifications are available, the process of dimensioning begins. This process leads
to a first, rough image of the power plant. The desired electrical output is translated into the
required thermal power of the core, the desired operating time is translated into the minimal
amount of fuel, and from the amount of fuel and the average power density, the size of the
core is roughly determined.
1
For example, an internet source I use often is Wikipedia. Most of the contributions concerning fossil fueled
power plants on WikiPedia only give the name of the power plant, details about the operating company, the nominal
power output of the plant or individual units and (often grossly overstated) claims about impact on climate (e.g. due
to CO2 emissions) and/or the local environment. At the same time, WikiPedia contains many articles with detailed
discussions about the properties of “innovative” nuclear power plants, which only exist on paper.

Preview version © W.F.G. van Rooijen. Reproduction prohibited. 65


CHAPTER 12. THERMAL-HYDRAULIC ANALYSIS OF NUCLEAR REACTOR CORES

• Following the process of dimensioning, the detailed design work for the core starts. Most
manufacturers have only a limited number of basic core designs which they will then adapt
to the specifications given by the customer. Note that it is very common in industry that
a manufacturer only has a (very) limited product range. For example, a car manufacturer
only produces a few types of engines, which will then be fitted, with small changes, to many
different types of cars.

• In the process of dimensioning, one of the most important parameters is the average linear
0
power density qavg (sometimes also referred to as the average linear heat rate). From the
desired thermal power output P and the desired value of qavg
0 the total length of the fuel pins
is determined. Then the number of fuel subassemblies, the number of pins per assembly, and
the height of the fuel stack in the fuel pins is determined to give a core with a form factor
(H/D-ratio) of approximately 1.0. Once the number of fuel pins per assembly is determined,
the pin pitch is determined taking into account the desired fuel-to-moderator ratio, and then
the overall size of a fuel assembly is known. The next step is to determine the number of
fuel assemblies such that a symmetric subassembly loading pattern becomes possible. Now
the total core size is known, and the H/D-ratio can be compared to the optimal value.
If the dimensioned core deviates too much from the optimal H/D-ratio, the design of the
subassembly may be changed and a new iteration of the dimensioning exercise is performed.

• It is possible, in fact, in many cases it is more likely than not, that the desired specifications
cannot be met. For example, the desired specifications may lead to a reactor core which is too
expensive, or which has poor performance in some aspect. In such a case, the manufacturer
will talk to the customer and try to find an acceptable compromise.

From Table 12.1 we can see that both LWR types (PWR and BWR) do not differ very much
in their basic thermohydraulic design. The BWR has a lower power density than PWR, that is
to say, for the same power output, the reactor vessel of a BWR plant is larger than the reactor
vessel of a PWR. At the same time, the primary system of the BWR is simpler than the PWR. For
example, a PWR requires steam generators which are very large pieces of equipment, whereas in a
BWR plant the steam separators are integrated into the reactor vessel. It is due to the properties
of the coolant (light water) and the nuclear fuel (UO2 ) that the thermohydraulic design of PWR
and BWR are similar.
The power density of the HTGR is very low, a full order of magnitude lower than PWR / BWR.
That means that for the same power output, the HTGR reactor vessel is very large - in fact, most
large-scale HTGR designs proposed to use concrete reactor vessels (so-called Pre-stressed Concrete
Reactor Vessel or PCRV) rather than steel reactor vessels, simply because it is not possible or not
practical to manufacture steel vessels of the required size. Due to the low power density, the HTGR
has a very large safety margin for any kind of accident and is often considered to be the safest type
of nuclear power plant.
The LMFBR has a very high power density, that is to say, a very small core with very high
power. The high power density is accommodated by using many fuel pins of very small diameter,
so that the core is a very tight (triangular) lattice of small fuel pins. A liquid metal coolant is used
so that a large heat flux can be safely used. Liquid metals have excellent heat transfer properties
and acceptable properties from the point of view of nuclear reactor physics. The most common
choice is liquid sodium, although that material carries with it some weak points: liquid sodium
burns when in contact with oxygen (air), reacts explosively with water, and becomes radioactive
due to neutron capture in the core.
A note about the data in Table 12.1. The data for HTGR and LMFBR are hypothetical, in
the sense that the presented data for GTGR and LMFBR in Appendix H of the text book do not
represent any existing nuclear power plant. At the time of writing of these answers (Fall 2020),
there was only one LMFBR in operation in the world, the BN-800 reactor at the Beloyarsk Nuclear
Power Plant in the Russian Federation, with a rated thermal output of 2100 MW, a gross electrical

66 Preview version © W.F.G. van Rooijen. Reproduction prohibited.


12.2. EXERCISE 12-2

Table 12.1: Data for several nuclear power plants.

Parameter PWR BWR HTGR LMFBR


Thermal Power [MW] 3600 3579 3000 2410
0
qavg [W cm−1 ] 178 206 78.7 295
qmax [W cm−1 ]
0 483 440 229 492
00 [W cm−2 ]
qavg 64 50.3 20.4 105
qmax [W cm−2 ]
00 168 111.5 58.3 237
000 [MW m−3 ]
qavg 95.7 56.0 8.4 380

output of 885 MW and a net electrical output of 800 MW. Operating experience with LMFBRs
has remained very limited in the world, with most operational experience gained in 3 prototype
power plants: BN-600 in the Russian Federation, and Phénix and Super-Phénix in France2 . At the
time of writing, no HTGR power plants were in operation in the world. The largest HTGR ever
constructed was the THTR-300 reactor in Germany (thermal output 750 MW, electrical output
300 MW). The THTR-300 program was canceled prematurely in 1989 due to a combination of
technical problems, financial difficulties and political opposition.

12.2 Exercise 12-2


Derive the equation of heat conduction by considering an energy balance for an arbitrary volume
(in analogy to the derivation of the neutron diffusion equation).

Solution
Consider an arbitrary (but non-reentrant) body of volume V bounded by surface S. The total
amount of thermal energy in V is given by:

2
In 2020, looking back on the development of the LMFBR as a nuclear reactor for electricity production, it is
rather disappointing to see that all efforts have, for all events and purposes, been wasted. In 2020, there are no
commercial nuclear power plants using LMFBR technology. Certainly, this failure cannot be contributed to a lack
of effort. Enormous amounts of resources, both financial resources and human resources, have been dedicated to the
development of LMFBR since the 1960s in all industrialized economies. Alas, societal acceptance of nuclear power
(or rather, a lack thereof), the availability of (cheap) energy resources in the form of fossil fuels, and the availability
of larger than expected amounts of natural uranium have caused LMFBR development to be largely abandoned.
In the US, the proposed Clinch River Breeder Reactor (CRBR) was canceled in the design phase in the late 1970s
and no subsequent proposals for LMFBR development have come to fruition in the US. In the UK, the Prototype
Fast Reactor (PFR) was closed prematurely in 1994 due to a lack of funding and no projects have come to fruition
since. The SNR-300 project in Germany, also known as the Kalkar Nuclear Power Plant (named after the village
where the plant was located), was canceled prematurely in 1991 for political reasons. The reactor was completely
constructed, but fuel was never loaded, and the entire site was eventually redeveloped as an amusement park. The
Monju reactor in Japan reached first criticality in 1994, but was subject to a politically motivated safety review
following a sodium leak in 1995. Following the sodium leak, the project never really recovered. Monju was restarted
in 2010, but suffered another unexpected accident three months later, and was put on standby following the accident
at the Fukushima Dai-ichi Nuclear Power Plant in 2011. Finally, the entire Monju project was canceled in 2018
after several decades of technical difficulties and very limited operation time (only several hundred hours of reactor
operation in 25 years). The Super-Phénix reactor in France was closed prematurely in the 1990s following technical
difficulties, controversy about the economics of the project and a general decline in political support for nuclear
power. Strangely enough, Phénix, the predecessor to Super-Phénix, was never subject to large-scale protests or
political pressure; the Phénix project operated succesfully from 1974 until 2006, with some periods without operation
due to upgrades and renovations. The reactor was finally closed in 2006 when the irradiation limits were reached
for several core components which could not be replaced. Efforts in India with respect to the construction of the
Prototype Fast Breeder Reactor (PFBR) seem to have stalled. Criticality was expected in 2010, then later revised
to 2014, but as of 2020, the PFBR project has not reached criticality.

Preview version © W.F.G. van Rooijen. Reproduction prohibited. 67


CHAPTER 12. THERMAL-HYDRAULIC ANALYSIS OF NUCLEAR REACTOR CORES

Z
ρ (r) cp (r) T (r, t) dr
V
Here, ρ is the material density and cp is the heat capacity. A temperature difference leads to
a flow of thermal energy, the so-called heat flux, call it F (r, t). Furthermore, there may be a heat
source q 000 (r, t). Thus, the time rate of change of the amount of heat (thermal energy) in volume
V:


Z Z Z
ρ (r) cp (r) T (r, t) dr = F · n̂dS + q 000 (r, t) dr
∂t
V S V
Now we can use the Gauss theorem to replace the surface integral with a volume integral:


Z Z Z
ρ (r) cp (r) T (r, t) dr = − ∇ · F dr + q 000 (r, t) dr
∂t
V V V
Now we can introduce Fick’s Law, i.e., the assumption that the heat flux is proportional to
the derivative of the temperature and the “diffusion coefficient” for heat (more often called heat
conduction coefficient or thermal conductivity):

F (r, t) = −λ∇T (r, t)


Thus:


Z
ρcp T (r) − ∇ · λ∇T − q 000 dr = 0
 
∂t
V
The integral must hold, regardless of the size and location of the volume V , which implies that
the integrand must be zero. If we further assume that the material properties such as density and
thermal conductivity do not depend on the temperature (i.e. thermal expansion is negligible, as is
the case for most solids and many fluids), one has:

∂T (r, t)
ρcp = ∇ · λ∇T + q 000
∂t

12.3 Exercise 12-3


Determine the temperature profile in plate-type fuel elements composed of fuel of thickness 2rF
sandwiched between a clad of thickness tC . Assume a gap thickness tG .

Solution
Assuming a steady state and assuming that the material properties are constant throughout each
material region, the equation to solve in the fuel is:

d dT
λ + q 000 = 0
dx dx
Integrate once to find:

dT
+ q 000 x = 0 λ
dx
Assuming that the thermal conductivity does not depend on x, one obtains:
ZrF ZrF
dT
λ dx + q 000 x = 0
dx
0 0

68 Preview version © W.F.G. van Rooijen. Reproduction prohibited.


12.3. EXERCISE 12-3

Thus:

1 r2
λ [T (0) − T (rF )] = q 000 x2 |r0F ⇒ λ [T (0) − T (rF )] = q 000 F
2 2
In the gap, the equation to solve is:

d dT dT
λG = 0 ⇒ λG =C
dx dx dx
Use Fick’s law, i.e. heat flux is proportional to the derivative of the temperature:

dT
= q 00
−λG
dx
This should be integrated from rF to x, keeping in mind that the heat flux is constant, i.e.:
Zx
dT
λG dξ = −(x − rF )q 00 ⇒ λG (T − T (rF )) = −q 00 (x − rF )

rF

Thus, the temperature profile in the gap is found as:

q 00
T (x) = T (rF ) + (rF − x)
ΛG
Thus, at the inside of the cladding x = rF + tG :

q 00 tG
λG (T (rF + tG ) − T (rF )) = −q 00 (x − rF ) ⇒ T (rF + tG ) = T (rF ) +
λG
In the clad, the situation is exactly the same as in the gap, thus:

q 00 tC
T (rF + tG + tC ) = T (rF + tG ) +
λC
Thus: in the “fuel meat”, the temperature difference is a quadratic function of the thickness
of the fuel meat, in the gap and cladding the temperature difference is a linear function of the
thickness.

Preview version © W.F.G. van Rooijen. Reproduction prohibited. 69


Chapter 13

The calculation of core power


distributions

13.1 Exercise 13-1 (Japanese version: 13-1)


Numerous factors in a nuclear reactor core design depend on the moderator-to-fuel volume ratio
VM /VF in a fuel cell. Suppose we imagine decreasing this ratio in a LWR, by decreasing lattice
pitch while keeping fuel rod radius fixed. Indicate qualitatively how and why each of the parameters
below might be affected by such a change in the reactor core if the core is initially undermoderated.

(a) Nuclear factors: η, f , p, , PF N L , PT N L , and k∞ .

(b) Thermal-hydraulic factors: Core power density, coolant flow rate, core pressure drop, fuel
and clad temperature, and DNBR.

(c) Safety factors: Temperature coefficient of reactivity and response to a loss of coolant accident.

(d) Economic factors: Fuel inventory, core size, and conversion ratio.

Solution
The assumption is that one decreases the lattice pitch, maintaining the original radius of the fuel
pins. There is one assumption to be made: one can either maintain the original core volume,
or one can maintain the original number of fuel pins. In the first case, the number of fuel pins
increases, while the core volume stays the same; in the second case, the number of fuel pins stays
the same, and as a result the core volume decreases. I have answered this exercises taking the
second assumption, i.e. the number of fuel pins stays the same and the equivalent core diameter
becomes smaller.

(a) Nuclear factors:

η this factor depends on the fuel composition. The core with reduced moderation requires a
(somewhat) higher enrichment, but η is relatively insensitive to enrichment if enrichment
is larger than 3% (see Exercise 2-22), thus η will not change very much.
f thermal utilization increases, because the spectrum will be somewhat harder, and since
most parasitic absorbers have 1/v cross sections, parasitic absorption will be reduced
and f then increases.
p resonance escape probability decreases, because the smaller moderator regions means
that a neutron is more likely to encounter a fuel lump while slowing down, and this
increases the probability of resonance absorption while slowing down.

Preview version © W.F.G. van Rooijen. Reproduction prohibited. 71


CHAPTER 13. THE CALCULATION OF CORE POWER DISTRIBUTIONS

 fast fission factor increases because the smaller moderator regions make it more likely
for a high-energy neutron to cross the moderator without collision, hence ending up in
a fuel lump at high energy, and that increases fast fission.
PF N L decreases because the neutron spectrum will be harder, which increases fast leakage, and
the physical size of the core is smaller, which increases both fast and thermal leakage.
PT N L decreases because the physical size of the core is smaller, which increases both fast and
thermal leakage.
k∞ decreases. The original core was already undermoderated, and decreasing the fuel pin
pitch will only make the core even more undermoderated than it already was.

(b) Thermal-hydraulic factors: Total core volume is decreased and power density is thus in-
creased. The coolant flow rate is determined by the total power and that parameter does
not change, hence the coolant flow rate does not change. The cross sectional area of the core
will become smaller, leading to a smaller cross sectional surface for coolant flow; the surface
area of the cladding stays the same. Since core pressure drop is dictated by the square of the
coolant velocity (increased) and the surface area for friction (stays the same), overall core
pressure drop will increase. Fuel volume stays the same and thus the power density in the
coolant stays the same, and since the coolant temperature does not change, the temperatures
of fuel pellet and clad do not change. DBNR will be somewhat reduced since the heat flux
stays the same but coolant flow velocity (and thus heat transfer coefficient) are higher.

(c) Safety factors: as the core becomes more undermoderated, the Doppler effect is reduced since
there are less neutrons in the low end of the slowing down energy range; the reduction of the
amount of moderator also means that any temperature effects due to moderator absorption
are reduced. A loss of coolant accident will not be affected in any meaningful way.

(d) Economic factors: the number of fuel pins is conserved, thus the overall fuel inventory does
not change while enrichment will be somewhat higher. Conversion will be somewhat reduced
due to higher enrichment, but at the same time the probability of resonance capture increases;
thus it is difficult to provide a solid answer to this question.

13.2 Exercise 13-2 (Japanese version: 13-2)


Using first-order perturbation theory, estimate the sign of the change in the thermal group constant
f due to the presence of Pu. For convenience, model the thermal resonance of 239Pu as a Dirac
239
σ 25
δ-function such that
s 
ET
Σ49
a = Σa (kT )ψ
 + αδ (E − E0 )
E

and, in particular, examine the cases of E0 > ET = kT and E0 < ET = kT .

Solution
The text book does not provide any theory about the requested calculation. Chapter 13 does not
discuss this topic. Chapter 9 does not have information, and while perturbation theory is discussed
in Chapter 5 for the core multiplication factor and the influence of control rods, perturbations
involving cross section resonances are not discussed, as far as I can tell. Therefore, I have decided
not to answer this question.

72 Preview version © W.F.G. van Rooijen. Reproduction prohibited.


13.3. EXERCISE 13-3 (JAPANESE VERSION: 13-3)

13.3 Exercise 13-3 (Japanese version: 13-3)


Demonstrate that for small ψ, the thermal group constants σ depend linearly on ψ.

Solution
I guess that this exercise is about the ψ of Eq. (13-3). There is barely any theory about what
ψ is and how it is calculated, so I cannot really answer this exercise. Suffice it to say that in
general, a sufficiently small perturbation in any independent variable leads to a linear change of
the dependent variable, i.e. if the independent variable (for example, coolant temperature) changes
by x%, the dependent variable (for example, an effective group cross section) changes by α × x%.
The fact that most perturbations produce more or less “first order” changes is reflected in the
common use of linearization to analyze the effects of small and localized perturbations on complex
physical phenomena.

Preview version © W.F.G. van Rooijen. Reproduction prohibited. 73


Chapter 14

Reactivity control

14.1 Exercise 14-1 (Japanese version: 14-1)


Calculate the effective control cross section Σeff
c characterizing a circular control cell geometry in
which the control rod is of radius a and the cell is of radius R. The transport parameter α for this
geometry is given in ANL-5800 as

1 0.2524 0.0949
λ= = 0.7104 + C +
3α Σa a (ΣC
a a)
2

Solution
Start this solution with the assumption that Eq. (14-9) of the text book applies, i.e.:

Pc φc
   
Σeff
c = α
Acell φ
The perimeter Pc and the cell surface Acell can be calculated from basic geometry, the value
of α is given in the text, so the only remaining work is to determine the flux solution in the
cell. Consider the following configuration: the control rod extends from r = 0 to r = b, and the
moderator region extends from r = b to r = c. There exists a uniform source of neutrons of
strength S0 in the moderator region, and at r = c a zero-current boundary condition exists. Write
the diffusion equation in cylindrical coordinates for the moderator region:

d2 ϕ 1 dϕ S0
+ − κ2 ϕ = −
dr 2 r dr D
The solution of this equation consists of a homogeneous solution ϕh and a particular solution
ϕp , where ϕh is the solution of

d2 ϕh 1 dϕh
+ − κ2 ϕh = 0
dr2 r dr
Assert that ϕp is a constant, then one obtains:

S0 S0
−κ2 ϕp = − ⇒ ϕp = 2
D κ D
The homogeneous equation becomes:

d2 ϕ dϕ
r2 +r − κ2 r 2 ϕ = 0
dr2 dr
Change variables to (κr), so that d(κr) = κdr, and define ρ ≡ κr:

Preview version © W.F.G. van Rooijen. Reproduction prohibited. 75


CHAPTER 14. REACTIVITY CONTROL

d2 ϕ dϕ
+ρ ρ2 − ρ2 ϕ = 0
dρ 2 dρ
The boundaries at r = b and r = c now become boundaries at ρ = β and ρ = γ, thus the
boundary conditions are:

J(β)

ϕ(β)
J(γ) = 0

Note: the current into the absorber is oriented in the negative ρ-direction, and since the current
is a vector, α < 0 in the present case (!). The equation for ϕh is a modified Bessel equation of order
zero, with solutions I0 (ρ) and K0 (ρ) (see Appendix D in the text book). Thus the flux solution is
found as:

S0
ϕ(ρ) = AI0 (ρ) + BK0 (ρ) +
κ2 D
where the constants A and B are to be determined from the boundary conditions. The effective
cross section is defined as:

Pc ϕc
Σeff
c = α
Ac ϕ
where Pc = 2πβ is the perimeter of the control rod, Ac = π γ 2 − β 2 is the surface area of


the moderator region, α is the blackness parameter as defined in this exercise, ϕc = ϕ(β) is the
flux in the moderator region just outside the control rod surface, and ϕ is the average flux in the
moderator region, i.e.:
Zγ    Zγ
2πρ AI0 + BK0 + S0/κ D 2
dρ 2πρϕ(ρ)dρ
β β
ϕ≡ =

 
π γ2 − β2
2πρdρ
β

Thus, the term Ac cancels with the denominator of ϕ and one finds for the effective cross
section:
 
AI0 (β) + BK0 (β) + S0 /κ2 D
Σeff
c = (2πβ) α Zγ   
2πρ AI0 (ρ) + BK0 (ρ) + S0 /κ2 D dρ
β

The integral over the Bessel functions can be simplified using the recurrence relations as given
in Appendix D of the text book:

 
AI0 (β) + BK0 (β) + S0 /κ2 D
Σeff
c = (2πβ) α    
2πA (γI1 (γ) − βI1 (β)) + 2πB (βK1 (β) − γK1 (γ)) + πS0 γ 2 − β 2 /κ2 D

The values of the constants A and B are determined as follows. From the boundary condition
J(γ) = 0, one obtains:


|ρ=γ = 0 ⇒ AI1 (γ) = BK1 (γ)

76 Preview version © W.F.G. van Rooijen. Reproduction prohibited.


14.2. EXERCISE 14-2

and from the boudary condition at ρ = β:

S0
 
−D [AI1 (β) − BK1 (β)] = α AI0 (β) + BK0 (β) +
κ2 D
This leads to:

I1 (γ)
B=A
K1 (γ)
and

S0
A [−DI1 (β)K1 (γ) + DI1 (γ)K1 (β) − αI0 (β)K1 (γ) − αI1 (γ)K0 (β)] = K1 (γ)α
κ2 D
This last equation can be used to determine the constant A. In Fig. 14.1 a plot is given of the
flux in the moderator region to given an impression of the shape of the neutron flux. For this plot
the following settings were used: β = 1.5 cm, γ = 4.0 cm, S0 = 1, D = 9.21 cm, κ2 = 0.04 cm2 , and
α = −0.47 (i.e. the simplest approximation for α).

Exercise 14-1
2.70

2.68

2.66
[-]

2.64

2.62

1.5 2.0 2.5 3.0 3.5 4.0


[cm]

Figure 14.1: The flux for Exercise 14-1.

14.2 Exercise 14-2


Demonstrate that for a perfectly black absorbing slab, the transport correction parameter is α =
0.47.

Solution
Start out by making some definitions. Assume two half spaces with a planar interface; set a
coordinate system such that the interface occurs at x = 0. Both half spaces contain homogeneous
materials. The half space x < 0 contains a diffusion medium with a homogeneously distributed
source of neutrons. The other half space at x > 0 contains a perfect absorber. The interface
condition on the boundary is that the current of neutrons emerging from the absorber is zero, i.e.:

j − (0) = 0

Preview version © W.F.G. van Rooijen. Reproduction prohibited. 77


CHAPTER 14. REACTIVITY CONTROL

This system is equivalent to a perfect vacuum for x > 0 and thus the boundary condition from
diffusion theory for a perfect vacuum applies. The blackness coefficient α in this case is defined as:

J(0)
α≡
ϕ(0)
In diffusion theory, this translates to:

D dϕ
α≡− |x=0
ϕ(0) dx
Now introduce the concept of extrapolation distance d from diffusion theory, which is defined
as:

ϕ(0)
d≡
dϕ/dx|x=0
and thus the blackness coefficient is given as:

D
α=
d
For the extrapolation distance, consider the following. The so-called Mark boundary conditions
for diffusion theory are:


j − (0) = 0 ⇒ ϕ(0) + 2D |x=0 = 0
dx
which leads to:

ϕ(0) 2
dMark = = 2D = = 0.6667λtr
dϕ/dx|x=0 3Σtr
The other well-known value for the extrapolation distance is 0.7104λtr . Thus we find for the
blackness coefficient:

D 1/3Σtr
αMark = = = 0.5
d 2/3Σtr
D 1/3Σtr
α= = = 0.46922
d 0.7104/Σtr

The real question is then how the extrapolation distance d = 0.7104λtr is determined. This
extrapolation length is determined from the solution of the so-called Milne Problem: assume a half
space x < 0 filled with a diffusing medium. The diffusion medium has isotropic scattering, capture,
but no fission. The half space x > 0 is vacuum. There is an isotropic source of mono-energetic
particles (neutrons) of infinite strength for x → −∞. Scattering reactions occur without energy
loss. This problem is a famous problem of mono-energetic transport theory which can be solved
analytically, see for instance [4, 1, 5]. The scalar flux in the Milne problem consists of an asymptotic
flux and a particular flux. The analytical solution of the Milne problem shows that the asymptotic
flux would go to zero at a distance from the interface [5, Page 111]:

1 (1 − c)2 (1 − c)3
" !#
d = 0.710446 − 0.0199 +O
c c c
where c is the number of secondary particles per collision. For perfect scattering, i.e. no
absorption, c = 1. The Milne Problem is one of the few problems in transport theory which
can be solved analytically. Results of the analysis of the Milne Problem are used throughout
the development of diffusion theory. The Milne Problem is also used to define the blackness

78 Preview version © W.F.G. van Rooijen. Reproduction prohibited.


14.3. EXERCISE 14-3 (JAPANESE VERSION: 14-2)

coefficients of (small and large) spheres, (small and large) cylinders, grey absorbers, etc. For
example, the extrapolation distance given in Exercise 14-1 for a cylindrical rod is calculated based
on a formulation of the Milne Problem in cylindrical geometry.
The Milne Problem was discovered in the analysis of the transport of photons through the
gaseous atmosphere of a star, for which the assumptions of weak absorption and an infinite source
strength are reasonable. The analytical solution of the Milne Problem has been used often to test
approximative solution methods, such as the solutions based on spherical harmonics expansions [1].
Even in modern times the Milne Problem remains relevant as a benchmark of numerical solution
methods [6].

14.3 Exercise 14-3 (Japanese version: 14-2)


Determine the effective control cross section Σeff
c for the slab geometry control cell described in
Section 14-II-B if the control rod is sufficiently weakly absorbing that diffusion theory is valid.
(That is, calculate α using diffusion theory in the control rod.)

Solution
The derivation on page 546 of the text book does not change if the absorber area is (also) treated
in diffusion theory. The value of the parameter α changes. Suppose the moderator region is at
0 ≤ x ≤ a and the absorber region is at a ≤ x ≤ b. Start by writing the diffusion equations in the
moderator region (subscript M ) and the absorber region (subscript A):

d 2 ϕM 1 S0
− 2 ϕM = −
dx 2 LM DM
d ϕA
2 1
− 2 ϕA = 0
dx 2 LA

with boundary conditions:

JM (0) = 0
JA (b) = 0
ϕM (a) = ϕA (a)
JM (a) = JA (a)

Pose a solution as follows:

x x SL2M
   
ϕM = A1 cosh + A2 sinh +
L L D
 M  M  M
b−x b−x
ϕA = A3 cosh + A4 sinh
LA LA

The boundary conditions x = 0 and x = b give:

DM x x
    
JM = − A1 sinh + A2 cosh
LM LM LM
DA b−x b−x
    
JA = A3 sinh + A4 cosh
LA LA LA

Preview version © W.F.G. van Rooijen. Reproduction prohibited. 79


CHAPTER 14. REACTIVITY CONTROL

Use JM (0) = 0 and JA (b) = 0 to find A2 = 0 and A4 = 0. The continuity conditions at x = a


then give:

a SL2M b−a
   
A1 cosh + = A3 cosh
LM DM LA
DM a DA b−a
     
− A1 sinh = A3 sinh
LM LM LA LA

The parameter α is defined as:

J(a)
α≡
ϕ(a)
We can use the moderator region or the absorber region to define α; if we use the absorber
region, the continuity condition at x = a yields:

DA b−a
 
α= tanh
LA LA
One can define the thickness of the absorber region as t = b−a to find α = (DA /LA ) tanh (t/La ).
It is interesting to determine the values of A1 and A3 to find the actual neutron flux in the system.
From the continuity conditions one finds:

DM LA sinh (a/LM )
A3 = −A1
DA LM sinh (t/LA )
and

SL2M
A1 = − DA LM [DA LM cosh (a/LM ) + DM LA sinh (a/LM ) coth (t/LA )]−1
DM
After some arithmetic, the solution is finally found as:

x S
 
ϕM = A1 cosh +
L ΣaM
 M 
b−x
ϕA = A3 cosh
LA
S D A LM
A1 = −
ΣaM sinh (a/LM ) [DA LM coth (a/LM ) + DM LA coth (t/LA )]
S DM LA
A3 =
ΣaM [DA LM coth (a/LM ) sinh (t/LA ) + DM LA cosh (t/LA )]

We can improve these expressions somewhat, as follows. For ϕM (compare Eq. (14-10) in the
text book):
 
α cosh LM
 
x
S 
ϕM (x) = 1−
ΣaM
   
α cosh LM + ΣaM LM sinh LM
a a

with (as before):

DA b−a
 
α≡ tanh
LA LA
For ϕA one finds:

80 Preview version © W.F.G. van Rooijen. Reproduction prohibited.


14.3. EXERCISE 14-3 (JAPANESE VERSION: 14-2)

 
β cosh
 
b−x
S  LA
ϕA (x) =
ΣaM ΣaA LA sinh b−a + β cosh b−a
   
LA LA

with:

DM a
 
β≡ tanh
LM LM
In Fig. 14.2 an illustration is given of the flux in the absorber cell. For this figure, the numerical
values of Table 14.1 were used. In this case, the absorption cross section in the absorber region is
probably too high to allow treatment with diffusion theory but it allows to understand the physical
phenomena. In this case, α = 0.27 applies. Under the assumption of a black absorber the flux
expression of Eq. (14-10) in the text book applies (note that Σa and L apply to the moderator
region in this case):

α cosh Lx
" #
S0
ϕ(x) = 1−
Σa α cosh La + Σa L sinh La
Under the assumption of a black absorber, the flux in the cell is lower than if diffusion theory
is valid; this is to be expected, because if diffusion theory is assumed to be valid, the absorber
must be relatively weak, and thus a “black” absorber is much stronger causing more neutrons to
be absorbed, leading to a decrease of the flux.

Exercise 14-3
M, diffusion theory
14 A, diffusion theory
M, black absorber
12

10

8
[-]

0
0 1 2 3 4 5 6
x [cm]

Figure 14.2: Flux solutions of Exercise 14-3.

Preview version © W.F.G. van Rooijen. Reproduction prohibited. 81


CHAPTER 14. REACTIVITY CONTROL

Table 14.1: Numerical data used in Exercise 14.3.


Quantity Value
Moderator thickness [cm] 5.0
Absorber thickness [cm] 1.0
ΣaM [cm−1 ] 0.061 040
ΣaA [cm−1 ] 0.764 690
DM [cm] 0.170 099
LM [cm−1 ] 1.669 337
DA [cm] 0.096 217
LA [cm−1 ] 0.354 718
α [-] 0.269 325

82 Preview version © W.F.G. van Rooijen. Reproduction prohibited.


Chapter 15

Analysis of core composition changes

15.1 Exercise 15-1


A thermal power reactor is shut down after an extended period of high-power operation. What
happens to: (a) the power output of the reactor (you may ignore the short-term transients if you
wish), (b) the average reactor temperature, and (c) the ability of the reactor to be restarted.

Solution
In the following, it is assumed that the reactor undergoes a scheduled shutdown, i.e. the reactor
does not suffer from a SCRAM or other emergency situation.

• As discussed in Chapter 6, the power due to nuclear fissions decreases quickly once the
control rods are inserted. In a scheduled shutdown, the reactor power is reduced gradually,
until eventually the power due to fissions becomes negligible. However, the decay heat, which
is due to the radioactive decay of fission products, remains. During normal operation of a
nuclear reactor, the decay heat amounts to some 7% of the total heat generated in the core.
After shutdown of the fission chain reaction, no new fission product nuclei are created but the
decay of the existing fission product nuclei continues. Since decay heat is due to radioactive
decay, the decay heat decreases as a function of time. In the case of a scheduled shutdown,
decay heat amounts to a few percent of the steady state power output. After an hour or so,
the decay heat amounts to about 1% of the steady state power, and after that decay heat
power decreases slowly.

• The most obvious effect is that the temperature profile inside the fuel pins changes. During
steady state operation, the radial tempeature profile in the fuel pins is a parabolic (quadratic)
function of the fuel pellet radius. Once the fission power is no longer present and the heat
release in the fuel pellet reaches decay heat level (1% to 2% of steady state power), the
temperature profile in the pellets is (nearly) flat. The temperature of the cladding and fuel
pellet thus becomes the same as the tempeture of the coolant, and since there is almost no
power production in the core, the outlet temperature of the coolant becomes the same as
the inlet temperature1 , that is to say, the reactor becomes isothermal. In a commercial size
nuclear reactor, the pumps in the primary system release a significant amount of energy to
the coolant, so that the primary system can be maintained at the desired (high) temperature
by the correct setting of the primary pump flow rate. This situation is commonly denoted as
Hot Zero Power or HZP.
1
Specialists of thermohydraulics might add here that the friction of the coolant along the cladding surface and
grid spacers is, in general, non-negligible and that the energy released due to friction is sufficient to cause the coolant
temperature to increase slightly as the coolant passes through the core.

Preview version © W.F.G. van Rooijen. Reproduction prohibited. 83


CHAPTER 15. ANALYSIS OF CORE COMPOSITION CHANGES

• Assuming that this is a power reactor, that the operation time has been sufficient to reach
saturation levels of 135Xe and 149Sm, and that the flux level is high enough as to have an
increase of the xenon concentration after shutdown, the reactor will be more difficult, if
not impossible, to restart during the period over which the xenon-effect is considerable (the
so-called reactor dead time, which is on the order of several tens of hours, or a few days).
After shutdown, the 149Sm concentration will steadily increase to a maximum, but at a
time-scale much slower than the Xe-transient, and with a smaller effect on the reactivity. It
should be noted, though, that the reduction of the average temperature of the fuel pellets
also increases the reactivity in a non-negligible manner due to the Doppler effect. Details
depend on the operation history of the reactor, the fuel composition, etc, but suffice it to say
that a commercial size nuclear reactor with a thermal neutron spectrum cannot be restarted
immediately after shutdown2 . Nuclear reactors for military applications, such as the nuclear
reactors used for propulsion of submarines or aircraft carriers, are often so-called epi-thermal
or intermediate spectrum reactors. These are not fast reactors in the true sense of the word,
but they are not thermal reactors either. The neutron spectrum in such reactors is deliberately
kept “above” the thermal range to avoid reactivity transients due to xenon and samarium
poisoning during operation at variable power.

15.2 Exercise 15-2 (Japanese version: 15-1)


Solve the isotopic rate equations to determine the buildup in xenon concentration following the
startup of a clean core to a steady-state flux level ϕ0 .

Solution
The equations describing the time dependent concentrations of 135
I, I(t), and Xe, X(t), are:
135

dI
= yI Σf ϕ(t) − λI I(t)
dt
dX
= YX Σf ϕ(t) + λI I(t) − λX X(t) − σaX ϕ(t)X(t)
dt
The general solutions are given in the book as:

Zt
 

I(t) = exp(−λI t) I(0) + yI dt0 Σf (t0 )ϕ(t0 ) exp(λI t0 )


0

and

Zt
 

X(t) = exp − λX + σax ϕ(t00 )dt00  ×


0
  0  
Zt Zt
X(0) + λI I(t0 ) + yX Σf (t0 )ϕ(t0 ) exp  λX + σaX ϕ(t00 )dt00  dt0 
    

0 0

Now introduce some assumptions: clean core at startup, i.e. I(0) = X(0) = 0, constant flux
level ϕ(t) = ϕ0 , and constant fission cross section, Σf (t) = Σf . Then, we find for I(t):
2
The well-known accident at unit 4 of the Chernobyl Nuclear Power Plant (April 26 1986) was partly caused
by the presence of a large amount of xenon in the core, due to the fact that the reactor was (accidentally) nearly
shutdown, about 80 minutes prior to the accident.

84 Preview version © W.F.G. van Rooijen. Reproduction prohibited.


15.3. EXERCISE 15-3

Zt
yI Σf ϕ0
I(t) = yI Σf ϕ0 exp(−λI t) dt0 exp(λI t0 ) = [1 − exp(−λI t)]
λI
0

This equation can now be substituted into the equation for X(t). First, we note that:

Zt
 
 
exp − (λX + σaX ϕ0 )dt0  = exp −(λX + σaX ϕ0 )t
0

and
 
Zt0  
exp − (λX + σaX ϕ0 )dt00  = exp −(λX + σaX ϕ0 )t0
 

Then, we find for X(t):

 
X(t) = exp −(λX + σaX ϕ0 )t ×
" Zt 

yI Σf ϕ0 exp (λX + σaX ϕ0 )t0 − exp (λX − λI + σaX ϕ0 )t0 dt0 +
  

0
Zt #
yX Σf ϕ0 exp (λX + σaX ϕ0 )t0 dt0
 

Performing all the integrals and multiplications then yields:

 #
1 exp −(λX + σaX ϕ0 )t
"
X(t) = yI Σf ϕ0 − −
λX + σaX ϕ0 λX + σaX ϕ0
 #
exp (−λI t) exp −(λX + σaX ϕ0 )t
"
yI Σf ϕ0 − +
λX − λI + σaX ϕ0 λX − λI + σaX ϕ0
 #
1 exp −(λX + σaX ϕ0 )t
"
yX Σf ϕ0 −
λX + σaX ϕ0 λX + σaX ϕ0

And this can be simplified to:

(yI + yX )Σf ϕ0 h  i
X(t) = 1 − exp −(λX + σ X
ϕ0 )t +
λX + σaX ϕ0 a

yI Σf ϕ0 h   i
exp −(λ X + σ X
ϕ 0 )t − exp (−λI t)
λX − λI + σaX ϕ0 a

15.3 Exercise 15-3


Determine the ratio of atomic number densities of equilibrium 135
Xe and 235
U as a function of the
steady thermal flux level of a reactor.

Preview version © W.F.G. van Rooijen. Reproduction prohibited. 85


CHAPTER 15. ANALYSIS OF CORE COMPOSITION CHANGES

Solution
Assuming that U is the only fissile isotope in the system, we know from the equilibrium value
235

of X(t):

∞ (yI + yX )Σf ϕ0 (yI + yX )N 5 σf5 ϕ0


lim X(t) = X = =
t→∞ λX + σaX ϕ0 λX + σaX ϕ0
and thus:

X∞ (yI + yX )σf5 ϕ0
=
N5 λX + σaX ϕ0

86 Preview version © W.F.G. van Rooijen. Reproduction prohibited.


Bibliography

[1] Alvin M. Weinberg and Eugene P. Wigner. The physical theory of neutron chain reactors. The
University of Chicago Press, 1958.

[2] George I. Bell and Samuel Glasstone. Nuclear Reactor Theory. Robert E. Krieger Publishing
Company, 1970. ISBN 0-88275-730-3.

[3] Andreas Pautz and Adolf Birkhofer. DORT-TD: A Transient Neutron Transport Code with
Fully Implicit Time Integration. Nuclear Science and Engineering, 145(3):299–319, 2003. doi:
10.13182/NSE03-A2385. URL https://doi.org/10.13182/NSE03-A2385.

[4] K.M. Case, F. de Hoffmann, and G. Placzek. Introduction to the theory of neutron diffusion.
Los Alamos Scientific Laboratory, 6 1953.

[5] James J. Duderstadt and William R. Martin. Transport Theory. John Wiley and Sons, 1979.

[6] S.K. Loyolka and S. Naz. Milne’s half-space problem: A numerical solution of the
related integral equation. Annals of Nuclear Energy, 35:1900 – 1902, 2008. URL
doi:10.1016/j.anucene.2008.04.002.

Preview version © W.F.G. van Rooijen. Reproduction prohibited. 87


Copyright © Amasaka Publishing and W.F.G. van Rooijen

©January 2022
Amasaka Publishing
919-1501
Fukui-ken Mikatakaminaka-gun
Wakasa-chō Amasaka 18-12-1
JAPAN

ISBN 978-4-600009-55-7 9 784600 009557

C3042 U8909 E 1 923042 089095


No part of this publication may be redistributed, multiplied, or transmitted, in any
form, electronically or in print, without prior written permission from the publisher.

View publication stats

You might also like