The neutron diffusion equation

Alain Hébert
alain.hebert@polymtl.ca

Institut de génie nucléaire

École Polytechnique de Montréal

ENE6103: Week 2 The neutron diffusion equation – 1/31

 Content (week 2) 1
Full core calculations
The steady-state diffusion equation
Continuity and boundary conditions
The finite homogeneous reactor
Cartesian coordinate system
Spherical coordinate system
Cylindrical coordinate system
The heterogeneous 1D slab reactor
Two region example

ENE6103: Week 2 The neutron diffusion equation – 2/31

 Full core calculations 1
The full-core calculation consists of solving a simplified transport equation, either the
diffusion equation or the simplified Pn equation.
This solution can be performed either in transient or steady-state conditions, using a
small number of energy groups (generally, G = 2 is sufficient).
A steady-state full-core calculation generally uses the effective multiplication factor
Keff as eigenvalue. Another possible choice of eigenvalue is to select a poison
concentration or the position of a reactivity device.
Full-core calculations offer the possibility of accurately representing the reactor
boundary. A correct representation of neutron leakage will be possible, as neutrons
effectively escape the reactor domain through its boundary.

ENE6103: Week 2 The neutron diffusion equation – 3/31

 Full core calculations 2
10 MeV

Energy
secondary fission neutrons Keff , B2

neutron absorption group 1

leakage group 1 group fission group 1
1

neutron absorption group 2

leakage group 2 group fission group 2
2

neutron absorption group 3

leakage group 3 group fission group 3
3

neutron absorption group G

leakage group G group
fission group G
G

0 ev

ENE6103: Week 2 The neutron diffusion equation – 4/31

 Full core calculations 3
We will first investigate the steady-state solution of the transport equation over the complete
reactor domain. The neutron balance over any control domain in energy group g is written

Leakage rate + Collision rate = Sources

or, in symbolic form,

(1) ∇ · J g (r) + Σg (r)φg (r) = Qg (r)

where J g (r) is the neutronic current. The scalar product of the neutronic current with N is
equal to the net number of neutrons crossing the arbitrary surface per unit surface and time.

ENE6103: Week 2 The neutron diffusion equation – 5/31

 Full core calculations 4
The neutronic sources Qg (r) represent the production of secondary neutrons from
scattering (including neutrons from (n,xn) reactions) and fission reactions:

G G
X χg (r) X
(2) Qg (r) = Σg←h (r) φh (r) + νΣfh (r) φh (r)
h=1
K eff h=1

where

G = total number of energy groups

Σg←h (r) = macroscopic scattering cross section from group h toward group g
χg (r) = fission spectrum in group g
νΣfh (r) = product of the macroscopic fission cross section by the average number of
neutrons emitted per fission in group h.

ENE6103: Week 2 The neutron diffusion equation – 6/31

 Full core calculations 5
In the particular case where G = 2, two 1 [ νΣ (r) φ (r) + νΣ (r) φ (r) ]
Keff
Keff f1 1 f2 2
approximations will be made:
a neutron cannot be accelerated from ∆ →
fast
• J1(r) Σa1(r) φ1(r)
group 2 (thermal) toward group 1 (fast) neutrons

all the secondary neutrons from fission

are produced in group 1. Σ2←1(r) φ1(r)

The approximations are written

∆ → thermal
• J2(r) neutrons Σa2(r) φ2(r)

Σ1←2 (r) = 0 , χ1 (r) = 1 and χ2 (r) = 0 .

Using these approximations, the sources Qg (r) simplify to

1
Q1 (r) = Σ1←1 (r)φ1 (r) + [νΣf1 (r)φ1 (r) + νΣf2 (r)φ2 (r)]
Keff
(3) Q2 (r) = Σ2←1 (r)φ1 (r) + Σ2←2 (r)φ2 (r) .

ENE6103: Week 2 The neutron diffusion equation – 7/31

 Full core calculations 6
We have obtained a balance equation for a steady-state reactor, corresponding to the
situation where the leakage and absorption rates are exactly equal to the production
rate of new neutrons, at all times, in each energy group.
In this case, the effective multiplication factor Keff is an artifact (or an eigenvalue) that
satisfies this equality. This factor is expected to be close to one for a reactor in a
nominal situation.
Later, we will show how an additional term can be added to the transport equation to
represent transient behavior of the reactor in cases where the equality is not met.
Equation (1) must be solved on the scale of the complete reactor, correctly taking into
account the position of its boundaries.
Solution of Eq. (1) requires additional information to relate the neutron flux and current.
Two approaches are possible to obtain this information:
use a spherical harmonics (Pn ) or discrete ordinate (SN ) approach. A legacy
alternative is to use a variant of the Pn method based on the simplified Pn
equation.
relate the neutron current to the gradient of the neutron flux using the Fick law.
This second choice will lead to the diffusion equation.

ENE6103: Week 2 The neutron diffusion equation – 8/31

 The steady-state diffusion equation 1
The Fick law is a heuristic relation between the neutron current and the gradient of the
neutron flux, translating the fact that neutrons have a tendency to migrate from regions
where they are more numerous to regions where they are less. This relation is known to be
acceptable on the scale of the complete reactor, but not at the level of lattice calculations
where it breaks down. It is written

(4) J g (r) = − Dg (r)∇φg (r)

where Dg (r) is a 3 × 3 diagonal tensor containing directional diffusion coefficients.

Non-directional diffusion coefficients are generally sufficient to represent streaming effects in

the lattice. In this case, the three diagonal components are simply set to the same value.
Directional diffusion coefficients are more closely related to the B1 heterogeneous streaming
effect occurring when long streaming channels or planes are open in the reactor. The unit of
the diffusion coefficient is the centimeter (cm).

Substituting Eq. (4) into Eq. (1), we get the neutron diffusion equation as

(5) − ∇ · Dg (r)∇φg (r) + Σg (r)φg (r) = Qg (r) .

ENE6103: Week 2 The neutron diffusion equation – 9/31

 The steady-state diffusion equation 2
The next step consists in subtracting the within-group scattering rate from both sides of
Eq. (5). We obtain the one-speed neutron diffusion equation as

(6) − ∇ · Dg (r)∇φg (r) + Σrg (r)φg (r) = Q⋄g (r)

where Σrg (r) = Σg (r) − Σg←g (r) is the removal cross section and where Q⋄g (r) is written

G G
X χg (r) X
(7) Q⋄g (r) = Σg←h (r) φh (r) + νΣfh (r) φh (r) .
h=1
Keff h=1
h6=g

In the two–energy group case (G = 2), Eq. (7) simplifies to

1
Q⋄1 (r) = [νΣf1 (r)φ1 (r) + νΣf2 (r)φ2 (r)]
Keff
(8) Q⋄2 (r) = Σ2←1 (r)φ1 (r) .

The neutron diffusion equation can be solved analytically in academic cases or using
standard numerical analysis techniques such as the finite difference or finite element
method.

ENE6103: Week 2 The neutron diffusion equation – 10/31

 The steady-state diffusion equation 3
Substituting the source term from Eq. (7) into Eq. (6), we get the multigroup form of the
steady-state neutron diffusion equation:

− ∇ · Dg (r)∇φg (r) + Σrg (r)φg (r)

G G
X χg (r) X
(9) = Σg←h (r) φh (r) + νΣfh (r) φh (r) .
h=1
λ h=1
h6=g

Equation (9) is an eigenproblem, whose solution behaves in a typical way:

A trivial solution of Eq. (9) is φg (r) = 0 , ∀ r. Many non-trivial solutions of Eq. (9) exist
for different eigenvalues λ. The largest eigenvalue in absolute value corresponds to
the fundamental solution of the eigenproblem and is equal to the effective
multiplication factor Keff of the reactor. Only the fundamental solution has a physical
meaning. The eigenspectrum of Eq. (9) is the set of all its eigenvalues (including Keff ).
Only the fundamental solution can lead to a positive neutron flux φg (r) over the
reactor domain. The other solutions are called neutron flux harmonics and lead to
oscillating values of the flux, sometime positive, sometime negative.

ENE6103: Week 2 The neutron diffusion equation – 11/31

 The steady-state diffusion equation 4
Every solution of Eq. (9) can be renormalized with an arbitrary normalization constant.
If φg (r) is a solution, then C φg (r) is also a solution for any value of constant C. The
normalization constant is generally computed from the knowledge of the reactor power:

G Z
X
(10) d3 r Hg (r) φg (r) = P
g=1 V

where V is the volume of the reactor, Hg (r) is the H–factor and P is the power of the
reactor. The H–factor permits the computation of the recoverable energy produced by
the reactor.

ENE6103: Week 2 The neutron diffusion equation – 12/31

 The steady-state diffusion equation 5
It is possible to find a mathematical adjoint to Eq. (9). This adjoint equation has the
same eigenspectrum as Eq. (9).

The mathematical adjoint of Eq. (9) is obtained by permuting primary and secondary group
indices. It is written

− ∇ · Dg (r)∇φ∗g (r) + Σrg (r)φ∗g (r)

G G
X νΣ fg (r) X
(11) = Σh←g (r) φ∗h (r) + χh (r) φ∗h (r)
h=1
λ h=1
h6=g

where φ∗g (r) is the adjoint flux or adjoint flux harmonics. The particular case with G = 1 is
said to be self-adjoint as it corresponds to the case where φ∗g (r) = φg (r). The adjoint flux
can also be renormalized with an arbitrary normalization constant. It is generally normalized
using the following arbitrary relation:

G Z
X
(12) d3 r χfg (r) φ∗g (r) = 1 .
g=1 V

ENE6103: Week 2 The neutron diffusion equation – 13/31

 Continuity and boundary conditions 1
The neutron flux is a continuous distribution of r and the neutron current must be continuous
across an infinite plane placed at abscissa x0 . The flux continuity condition at this point is

+
(13) φg (x−
0 , y, z) = φg (x0 , y, z) ∀ y and z .

The neutron current continuity condition is written after introducing the unit normal
N = (1, 0, 0), perpendicular to the infinite plane. We write

+
(14) J g (x−
0 , y, z) · N = J g (x0 , y, z) · N ∀ y and z .

Using the Fick law from Eq. (4), we obtain

+ +
(15) − Dg (x− −
0 , y, z)∇φg (x0 , y, z) · N = − Dg (x0 , y, z)∇φg (x0 , y, z) · N ∀ y and z

or

d ˛
+ d ˛
Dg (x−
0 , y, z) φg (x, y, z)˛ = D (x , y, z) φ (x, y, z) ∀ y et z .
˛ ˛
(16) g 0 g
x=x− x=x+
˛
dx 0 dx 0

Equation (16) indicates that the neutron flux gradient is discontinuous at each point of the
domain where the diffusion coefficient is discontinuous.
ENE6103: Week 2 The neutron diffusion equation – 14/31
 Continuity and boundary conditions
Boundary conditions must be applied at
each point of ∂V .
real boundary
2
incoming and outgoing net currents, using
J g (r) · N(r) = Jg+ (r) − Jg− (r) if r ∈ ∂Wi

where N(r) is a unit vector, normal to ∂Wi ,

reactor
and pointing in the outgoing direction.
CL
symmetry boundary N
r
The most straightforward way to represent a reactor Jg+(r)
Jg−(r)
real boundary is to set a zero-flux boundary
condition on it. In this case, we simply write
The albedo at r is defined by the relation
(17) φg (r) = 0 if r ∈ ∂Wi
Jg− (r)
(18) βg (r) = if r ∈ ∂Wi .
This condition is not satisfactory, as it Jg+ (r)
assumes the absence of neutrons on ∂Wi .
In general, we use the same albedo in all
It would be more exact to assume a zero energy groups. The most usual values are
incoming current condition. This condition is β(r) = 0 to represent a zero incoming
a particular case of the more general albedo current condition and β(r) = 1 to represent
boundary condition. We first identify the a symmetry condition.
ENE6103: Week 2 The neutron diffusion equation – 15/31
 Continuity and boundary conditions 3
The incoming and outgoing net currents can be obtained from the neutron current and flux
distributions provided the angular flux is represented by a limited P1 expansion

1 1 1 1
(19) Jg− (r) = φg (r) − J g (r) · N(r) and Jg+ (r) = φg (r) + J g (r) · N(r) .
4 2 4 2

Substituting Eqs. (19) into (18) and using the Fick law (4), we obtain the albedo boundary
condition as

1 1 − β(r)
(20) Dg (r)∇φg (r) · N(r) + φg (r) = 0 if r ∈ ∂Wi
2 1 + β(r)

where ∂Wi is the fraction of ∂V where the albedo boundary condition is applied.

The Fick law and the P1 approximation introduce an error in the zero incoming current
condition that can be reduced by using a value of β(r) slightly above zero. We recommend
to represent the zero incoming current condition using the value β(r) = 0.031758.

Another particular case is the symmetry condition obtained by setting the albedo to value
β(r) = 1. In this case, Eq. (20) reduces to

(21) ∇φg (r) · N(r) = 0 if r ∈ ∂Wi .

ENE6103: Week 2 The neutron diffusion equation – 16/31
 The finite homogeneous reactor 1
Consider an homogeneous and finite reactor surrounded by zero-flux or symmetry
boundary conditions. The nuclear properties of the reactor are independent of space
We use non-directional diffusion coefficients.

Equation (9) simplifies to

G G
2
X χg X
(22) − Dg ∇ φg (r) + Σrg φg (r) = Σg←h φh (r) + νΣfh φh (r) .
h=1
Keff h=1
h6=g

It is possible to factorize the flux according to

(23) φg (r) = ψ(r) ϕg .

Substituting Eq. (23) into Eq. (22), we obtain

8 9
G G
> >
∇2 ψ(r)
> >
Σrg 1 <X χg X =
(24) − =− + Σg←h ϕh + νΣfh ϕh .
ψ(r) Dg Dg ϕg > Keff h=1
>
: h=1
> >
;
h6=g

ENE6103: Week 2 The neutron diffusion equation – 17/31

 The finite homogeneous reactor 2
We note that the left side of Eq. (24) is independent of the neutron energy whereas its right
side is independent of the position in reactor. This fact is only possible if each side of
Eq. (24) is itself equal to the same constant. This constant was set equal to B 2 , the buckling
of the reactor. We therefore obtain two independent equations as

(25) ∇2 ψ(r) + B 2 ψ(r) = 0

and
G G
X χ g
X
Dg B 2 + Σrg ϕg =
ˆ ˜
(26) Σg←h ϕh + νΣfh ϕh .
h=1
Keff h=1
h6=g

Equation (25) is a Laplace equation, an eigenproblem whose eigenvalue is the buckling B 2 .

Its solution is a function of the shape and size of the reactor and of the boundary conditions:
Zero-flux boundary condition: ψ(r) = 0 if r ∈ ∂Wi
Symmetry boundary condition: ∇ψ(r) · N(r) = 0 if r ∈ ∂Wi .
Equation (25) has many non-trivial solutions, each of them corresponding to an element of
its eigenspectrum, but only the fundamental solution corresponds to a positive neutron flux
everywhere in the domain.

ENE6103: Week 2 The neutron diffusion equation – 18/31

 Cartesian coordinate system 1
The Cartesian coordinate system is the most usual choice for real application problems. In
this case, the Laplace operator is written

2 ∂2ψ ∂2ψ ∂2ψ

(27) ∇ ψ= + + .
∂x2 ∂y 2 ∂z 2

Let us consider a prismatic homogeneous reactor of dimension Lx × Ly × Lz . In this case,

the fundamental solution of Eq. (25) is

πx πy πz
(28) ψ(x, y, z) = C sin sin sin .
Lx Ly Lz

The Cartesian domain is defined over 0 ≤ x ≤ Lx , 0 ≤ y ≤ Ly and 0 < z < Lz . A zero-flux

boundary condition is imposed on the surface of the domain. The normalization constant C
is arbitrary as both Eqs. (25) and (26) are eigenproblems.
The corresponding critical buckling is
„ «2 „ «2 „ «2
π π π
(29) B2 = + + .
Lx Ly Lz

ENE6103: Week 2 The neutron diffusion equation – 19/31

 Spherical coordinate system 1
The Laplace operator is written

1
»
∂
„
∂ψ
«
∂
„
∂ψ
«
1 ∂ 2ψ –
(30) ∇2 ψ = 2 sinθ r2 + sinθ + 2
.
r sinθ ∂r ∂r ∂θ ∂θ sinθ ∂ǫ

Z The fundamental solution of Eq. (25)

represents the neutron flux in a spherical
reactor of radius R. A zero-flux boundary
r condition is imposed at r = R (i.e.,
θ
r ψ(R) = 0). The fundamental solution is
written
Y C πr
(31) ψ(r) = sin
r R
ε with the critical buckling equal to
X “ π ”2
2
(32) B = .
R

ENE6103: Week 2 The neutron diffusion equation – 20/31

 Cylindrical coordinate system 1
The Laplace operator is written

1 ∂2ψ ∂2ψ
» „ « –
1 ∂ ∂ψ
(33) ∇2 ψ = ρ + 2
+ρ 2 .
ρ ∂ρ ∂ρ ρ ∂ǫ ∂z

Z The fundamental solution of Eq. (25)

represents the neutron flux in a cylindrical
z reactor of radius R and height Lz . A
ρ zero-flux boundary condition is imposed on
the surface. The fundamental solution is
r
„ «
2.405ρ πz
(34) ψ(ρ, z) = C J0 sin
R Lz
Y
with the critical buckling equal to
ε
«2 «2
X
„ „
2.405 π
(35) B2 = +
R Lz

where J0 (x) is a zeroth order ordinary

Bessel function, such as J0 (2.405) = 0.
ENE6103: Week 2 The neutron diffusion equation – 21/31
 The heterogeneous 1D slab reactor 1
The one-dimensional (1D) heterogeneous reactor configurations correspond to the
case where the neutron flux is a function of a unique spatial variable.
These cases can be solved analytically whatever the type of conditions imposed at
boundaries.
The nuclear properties of the reactor are only a function of the independent variable x.

Equation (9) simplifies to

d dφg
− Dg (x) + Σrg (x) φg (x) = Q⋄g (x)
dx dx
G G
X χg (x) X
(36) = Σg←h (x) φh (x) + νΣfh (x) φh (x) .
h=1
K eff h=1
h6=g

The boundary conditions are either a zero-flux condition (φg (x) = 0) or an albedo condition:

dφg 1 1 − β(x)
(37) ∓ Dg (x) + φg (x) = 0
dx 2 1 + β(x)

where the sign “−" or “+" is used for a left (x = x1/2 ) or a right boundary (x = xI+1/2 ).
ENE6103: Week 2 The neutron diffusion equation – 22/31
 The heterogeneous 1D slab reactor 2
Each slab is assumed to be homogeneous, so that the corresponding nuclear properties
Dg (x), Σrg (x), Σg←h (x), χg (x) and νΣfh (x) are piecewise continuous. The reactor
domain is divided into I regions of indices 1 ≤ i ≤ I, in such a way that the nuclear
properties in region i are constant and equal to Dg,i , Σrg,i , Σg←h,i , χg,i and νΣfh,i .

region i-1 region i region i +1

X
x i -3/2 x i -1 x i-1/2 xi x i+1/2 x i+1 x i+ 3/2

∆xi -1 ∆xi ∆xi+1

Equation (36) can be written in such a way as to be valid in region i as

G G
d2 φg X χ g,i
X
− Dg,i 2
+ Σrg,i φg (x) = Q⋄g (x) = Σg←h,i φh (x) + νΣfh,i φh (x)
dx h=1
K eff h=1
h6=g

(38) if xi−1/2 < x < xi+1/2 .

ENE6103: Week 2 The neutron diffusion equation – 23/31

 The heterogeneous 1D slab reactor 3
At this point, we introduce the analytical solution approach for Eq. (38). It is based on a
linear transformation technique, valid only for multigroup 1D problems. Equation (38) is first
rewritten in matrix form as

d2
(39) Φ(x) + Fi Φ(x) = 0 if xi−1/2 < x < xi+1/2
dx2

with
f11,i f12,i ... f1G,i
0 1
0 1
φ1 (x) B f21,i
B . C B f22,i ... f2G,i C
C
(40) @ .. A
Φ(x) = B and Fi = B
B .. . .. . C
C
@ . . . . C
. . A
φG (x)
fG1,i fG2,i ... fGG,i
where the components fgh,i of this matrix are written as
» –
1 χg,i
(41) fgh,i = −Σrg,i δgh + Σg←h,i (1 − δgh ) + νΣfh,i .
Dg,i Keff

ENE6103: Week 2 The neutron diffusion equation – 24/31

 The heterogeneous 1D slab reactor 4
The next step consists in finding all eigenvectors tℓ,i of matrix Fi with the associated
eigenvalues λℓ,i . We build a matrix Ti whose columns are the eigenvectors of Fi :

(42) Ti = ( t1,i t2,i ... tG,i )

so that
(43) Fi Ti = Ti diag(λℓ,i ) .

The linear transformation technique used to solve Eq. (36) is based on the introduction of an
unknown vector Ψ(x) defined in such a way that

t11,i t12,i ... t1G,i ψ1 (x)

0 10 1

C B ψ2 (x) C
B t21,i t22,i ... t2G,i C B C
B
(44) Φ(x) = Ti Ψ(x) = B
B .. . .. . C B . C
. . . ABC . C
@ . . . @ . A
tG1,i tG2,i ... tGG,i ψG (x)

and to its substitution in Eq. (39). We obtain

d2
(45) Ti Ψ(x) + Fi Ti Ψ(x) = 0 if xi−1/2 < x < xi+1/2 .
dx2
ENE6103: Week 2 The neutron diffusion equation – 25/31
 The heterogeneous 1D slab reactor 5
We next left–multiply each side of Eq. (45) by [Ti ]−1 and use Eq. (43) to obtain

d2
(46)
2
Ψ(x) + diag(λℓ,i )Ψ(x) = 0 if xi−1/2 < x < xi+1/2 .
dx

Equation (46) is similar to Eq. (39) with the difference that all the energy groups are
uncoupled. Its resolution is reduced to the solution of G one-speed problems. In each
energy group g, we assume an analytical solution of the form
8 p p
< Ag,i cos( λg,i x) + Bg,i sin( λg,i x)
> if λg,i ≥ 0;
(47) ψg (x) =
> p p
g,i cosh( −λg,i x) + Eg,i sinh( −λg,i x) otherwise
:C

if xi−1/2 < x < xi+1/2 .

The analytical expression of the flux φg (x) is obtained after substitution of Eq. (47) into
Eq. (44) as
XG
(48) φg (x) = tgh,i ψh (x) if xi−1/2 < x < xi+1/2 .
h=1

ENE6103: Week 2 The neutron diffusion equation – 26/31

 The heterogeneous 1D slab reactor 6
The final step consists of coupling together the analytical solutions of each region i and
applying the boundary conditions. If region i − 1 exists (i.e., if the left side of region i is not a
boundary), the solution obeys the continuity relations (13) and (16), so that

+
(49) φg (x−
i−1/2
) = φg (xi−1/2
)

+
(50) and Dg,i−1 φ′g (x−
i−1/2
) = Dg,i φ′
g (xi−1/2
)

dφg ˛
˛
where we defined φ′g (x− ) = . Similarly, if region i + 1 exists, we write
i−1/2 dx ˛x=x−
i−1/2

+
(51) φg (x−
i+1/2
) = φg (xi+1/2
)

+
(52) and Dg,i φ′g (x−
i+1/2
) = Dg,i+1 φ′
g (xi+1/2
) .

The boundary conditions are imposed in the same way, by forcing the value of φg (x) or
φ′g (x) on a boundary abscissa.

An analytical solution of Eq. (38) with its continuity and boundary conditions can always be
found, with a computational effort increasing with the number of energy groups and regions.

ENE6103: Week 2 The neutron diffusion equation – 27/31

 Two-region example 1
We apply this technique to a one-speed, CL
two-region problem. Eq. (38) simplifies to

zero-flux
region 1 region 2
d2 φ 1
− Di 2 + Σr,i φ(x) = νΣf,i φ(x) 0 1
X
dx Keff 1/2

(53) if xi−1/2 < x < xi+1/2 .

We assume an analytical solution of the form

8
1
< A11 cos(κ1 x) + A21 sin(κ1 x)
> if 0 ≤ x ≤ 2
;
(54) φ(x) =
> 1
:A
12 cos(κ2 x) + A22 sin(κ2 x) if 2
≤x≤1

8
1
< −A11 κ1 sin(κ1 x) + A21 κ1 cos(κ1 x)
> if 0 ≤ x ≤ 2
;
(55) φ′ (x) =
> 1
: −A
12 κ2 sin(κ2 x) + A22 κ2 cos(κ2 x) if 2
≤x≤1
8
2 2 1
d2 φ < −A11 κ1 cos(κ1 x) − A21 κ1 sin(κ1 x)
> if 0 ≤ x ≤ 2
;
(56) =
dx2 1
κ22 cos(κ2 x) − A22 κ22 sin(κ2 x)
>
: −A
12 if 2
≤x≤1 .
ENE6103: Week 2 The neutron diffusion equation – 28/31
 Two-region example 2
The left zero-flux boundary condition is written

(57) φ(0) = A11 = 0 .

1
The continuity conditions at x = 2
are written
“κ ” “κ ” “κ ” “κ ”
1 1 2 2
(58) A11 cos + A21 sin = A12 cos + A22 sin
2 2 2 2

and
“κ ” “κ ” “κ ” “κ ”
1 1 2 2
−A11 κ1 D1 sin +A21 κ1 D1 cos = −A12 κ2 D2 sin +A22 κ2 D2 cos .
2 2 2 2
(59)
Finally, the right symmetry boundary condition is written

−A12 κ2 sin(κ2 ) + A22 κ2 cos(κ2 ) = 0

so that
(60) A22 = A12 tan(κ2 ) .

ENE6103: Week 2 The neutron diffusion equation – 29/31

 Two-region example 3
Substituting Eqs. (57) and (60) into Eqs. (58) and (59), we find the following matrix equation:
0 ` κ1 ´ 1 “κ ”1
2
sin 2
− cos „ « „ «
cos(κ2 ) 2 C A21 0
= .
B
(61) @ ` κ1 ´ κ2 D2 “ κ2 ” A
A 12 0
κ1 D1 cos 2
− sin
cos(κ2 ) 2

Equation (61) has a non-trivial solution only if its determinant is zero. We write
0 ` κ1 ´ 1 “κ ”1
2
sin 2
− cos
cos(κ2 ) 2 C
det @ κ2 ” A = 0
B
(62) ` κ1 ´ κ2 D2 “
κ1 D1 cos 2
− sin
cos(κ2 ) 2

so that the resulting characteristic equation is a non-linear relation in κ1 and κ2 , written

“κ ” “κ ” κ1 D1
1 2
(63) tan tan = .
2 2 κ2 D2

Substituting Eqs. (54) and (56) into Eq. (53), we find the expressions of κ1 and κ2 as
s s
νΣf,1 − Keff Σr,1 νΣf,2 − Keff Σr,2
(64) κ1 = and κ2 = .
Keff D1 Keff D2
ENE6103: Week 2 The neutron diffusion equation – 30/31
 Two-region example 4
Combining Eqs. (63) and (64), we obtain the non-linear expression of the effective
multiplication factor Keff as
s ! s ! s ` ´
1 νΣf,1 − Keff Σr,1 1 νΣf,2 − Keff Σr,2 D1 νΣf,1 − Keff Σr,1
tan tan = ` ´ .
2 Keff D1 2 Keff D2 D2 νΣf,2 − Keff Σr,2
(65)

The effective multiplication factor is obtained by solving Eq. (65) using a numerical technique
for the solution of a non-linear equation with one unknown. Knowledge of Keff permits the
computation of κ1 and κ2 , and the determination of the analytical expression of the neutron
flux φ(x), the eigenvector corresponding to Keff . Its value is
8 1
> A21 sin(κ1 x) if 0 ≤ x ≤ 2
;
>
>
>
< “κ ”
1
(66) φ(x) = sin
>
> A21 “ κ2 ” cos[κ2 (1 − x)] if 1
≤x≤1 .
> 2 2
cos
>
:
2

The neutron flux can be renormalized to the power of the reactor by adjusting the remaining
constant A21 .

ENE6103: Week 2 The neutron diffusion equation – 31/31