study of predictor-corrector methods for monte carlo burnup...
TRANSCRIPT
Study of Predictor-corrector methods
for Monte Carlo Burnup Codes
Supervisor
By Dan Kotlyar
Dr. Eugene Shwageraus
Serpent International Users Group Meeting
Madrid, Spain, September 19-21, 2012
Introduction – High Conversion LWRs
1
0.90
0.95
1.00
1.05
1.10
1.15
0 200 400 600 800 1000 1200
Time, days
Eig
en
va
lue
2D (no T-H)
Introduction – High Conversion LWRs
1
0.90
0.95
1.00
1.05
1.10
1.15
0 200 400 600 800 1000 1200
Time, days
Eig
en
va
lue
3D (T-H)
2D (no T-H)
Introduction – Advisor’s response
2
What ?
What?
Why ?
Programming ?
Physical (model) ?
Numerical (coupling)?
Cause
How ?
Solution !
Is this problem common for all MC based codes ? …
0.00
0.05
0.10
0.15
0.20
0.25
0 50 100 150 200 250 300 350
Height, cmP
ow
er
sh
are
in
seed
T [i]
T[i+1]
Outline
3
Simplify the model
Generic PWR assembly in 3D
No T-H feedback, but non-uniform coolant density fixed in time
Study efficient schemes for coupling MC with multi-physics
feedbacks
Identify origins of the observed problem
Focus of this work:
Examine MC-burnup coupling schemes (& t size)
Was this problem identified earlier?
Recent burnup schemes sensitivity studies
Description of the test case (3D) & Results
Conclusions & future work
Introduction
4
Examples of Integrated MC-burnup codes:
MOCUP (Moore et al., 1995)
MCODE (Xu et al., 2002)
MONTEBURNS (Trellue, 2003)
SERPENT (Leppänen, 2007)
BGCore (Fridman et al., 2008)
MCNPX (Hendricks et al., 2008)
What about Integration of MC-burnup-TH
BGCore
SERPENT
Issues related to Integrated MC codes
Accuracy
Computation requirements
Objectives
5
Investigate stability and accuracy of current depletion calculations
Monte-Carlo/Burnup coupling scheme
Depletion timestep size
Commonly used methods
Explicit Euler predictor method
Euler predictor-corrector method
Propose an extension to the modified Euler predictor-corrector
Accuracy as the predictor-corrector method
Computation requirements as the predictor method
Possible multi-physics coupling schemes
6
Thermal feedback
Depletion module
Neutronics MC
Power & BU
Distribution
Temp & density
Distribution
(r, E)
Beginning of Step analysis (predictor)
Total exe time (Depletion/TH)≈ n·Tmcnp
New concentration for next step
(x 1)
(x n)
Integration scheme 1
Inner loop: TH-neutronic
Outer loop: Depletion
Integration scheme 2
Inner loop: Depletion
Outer loop: TH-neutronuc
Important issues
Combination of #1 and #2
Calculation time
Source convergence
Distribution of errors
Variable convergence tolerances
Recent sensitivity studies
7
Yamamoto et al., 2008 – Projected predictor corrector method
Linear correlation between the number density and the microscopic RR
Tested for the Gd-bearing 2D fuel assembly
Carpenter et al., 2010 (Bettis Atomic Power Laboratory)
Modified Log Linear correlation (Yamamoto)
Tested for the Gd-bearing 2D fuel assembly
Isotalo et al., 2011 – Higher order methods
Use more (previous) BU points
Tested on PWR pin cell and seed/blanket assembly
Saadi et al., 2012 – Burnup sensitivity analysis
Tested on 1D UO2 PWR unit cell
Dufek et al., 2009 – Numerical stability of MC-burnup codes
Instability demonstrated on infinitely reflected 3D unit cell
Burnup coupling scheme
methodology
Extended Predictor-Corrector
Euler Predictor-Corrector
Explicit Euler Predictor
Formulation of the burnup problem
Burnup calculations
8
Eigen-value
transport problem
Burnup equation
Matrix Exponential solution
trNdEEtEr
dt
trdN,,,
,
0
0,,,,1
,
tErtrNF
ktrNL
n n+1 n+2 ……… EOC
nnnn ttrNrN 11 exp
Burnup calculations
8
Eigen-value equation
Burnup equation
Mat. Exp. solution
Objective: coupled space-energy-time dependent solution:
Coupling scheme: Independent neutronics and depletion solvers
trNdEEtEr
dt
trdN,,,
,
0
0,,,,1
,
tErtrNF
ktrNL
n n+1 n+2 ……… EOC
nnnn ttrNrN 11 exp
Euler predictor-corrector methods
9
1. Explicit Euler Predictor method
Reaction rate (RR) calculation (neutronics - ) at BOT
Depletion with BOT RR’s Nn+1
2. Modified Euler Predictor-Corrector method
Reaction rate calculation ( ) with predicted Nn+1 values
Depletion with EOT (n+1) reaction rates to obtain corrected Nn+1 values
Finally, the predicted and corrected ND are averaged
Used as initial values for the following step
n n+1 Time/BU
Nn Nn+1
const RRn
const RRn+1
Nn Nn Nn Nn+1 Nn
Why is the P-C method is not enough ? t !
10
Quote: “Predictor-corrector methods are numerically explicit”
The stable size may not be much larger than of that of the predictor
Numerical methods for engineers and scientists, Joe D. Hoffman
0.00
0.01
0.02
0.03
0.04
0.05
0.06
0.07
0.08
0.09
0 50 100 150 200 250 300 350 400
Height, cm
No
rmalized
co
mm
ula
tive e
nerg
y (
50 d
ays)
5 d
10 d
25 d
50 d
0.000
0.020
0.040
0.060
0.080
0.100
0.120
0.140
0 50 100 150 200 250 300 350 400
Height, cm
Po
we
r S
ha
re
Predictor
Corrector
50,0t
i ttPEt
50 d 0
50 days
Suggested E-P-C method
11
3. Extended Predictor-Corrector method (E-P-C)
Assume (power, RR, σ, N) are known for the previous [tn-1,tn] interval
Deplete with previous average RR Nn+0.5 ( =Nnexp[-RRn-1·t0.5] )
Update transport solution (RR calculation ) at MOT (tn+0.5)
Deplete with MOT RR’s Nn+1
Reaction rate calculation ( ) with predicted Nn+1 values
Depletion with EOT (n+1) reaction rates to obtain corrected Nn+1 values
Finally, the predicted and corrected ND are averaged
n 100d
n+0.5 125d
Time/BU
Nn Nn+0.5
const RRn+0.5
const RRn+1 n+1 150d
Nn+1 Nn-1
n-1 75d
Different coupling
scheme results
3D assembly test case
12
+20,000 Radial layout
Cylindrical pins
With guide tubes
Axial layout / Coolant density profile
21.5 cm
0.6
9
0.7
0
0.7
1
0.7
2
0.7
3
0.7
4
0.7
5
0.7
6
Coolant Density, gr/cm3
BGCore vs. SERPENT comparison
13
1.10
1.15
1.20
1.25
1.30
1.35
1.40
0 50 100 150 200 250 300 350 400
Time, days
Eig
en
valu
e
BGCore
SERPENT
0.0E+00
1.0E-05
2.0E-05
3.0E-05
4.0E-05
5.0E-05
6.0E-05
7.0E-05
8.0E-05
9.0E-05
1.0E-04
0 50 100 150 200 250 300 350 400
Time, days
Co
ncen
trati
on
of
Pu
239, #/b
·cm
BGCore
SERPENT
Maximum difference,
SERPENT vs. BGCore Parameter
~70 pcm k-eff
~0.5 % Xe135
~0.4 % U235
~0.4 % Pu239
Method : predictor-corrector
Timestep : 5 days
Burnup coupling scheme method & T
14
Description Case
no. Designated T,
days Method
P-C-0 (25d) 25 Predictor 1
P-C-1 ( 5d) 5 Predictor-Corrector a 3
P-C-1 (25d) 25 Predictor-Corrector 3
P-C-1 (50d) 50 Predictor-Corrector 4
E-P-C (50d) 50 Extended-Predictor-Corrector 5
a. Chosen to be the reference case
Burnup coupling scheme method & T
14
-200
0
200
400
600
800
1000
1200
0 50 100 150 200 250 300 350 400
Time, days
Dif
fere
nce in
eig
en
valu
e, p
cm
P-C-0 (25d)
P-C-1 (25d)
P-C-1 (50d)
Description Case
no. Designated T,
days Method
P-C-0 (25d) 25 Predictor 1
P-C-1 ( 5d) 5 Predictor-Corrector a 3
P-C-1 (25d) 25 Predictor-Corrector 3
P-C-1 (50d) 50 Predictor-Corrector 4
E-P-C (50d) 50 Extended-Predictor-Corrector 5
a. Chosen to be the reference case
Burnup coupling scheme method & T
14
-200
0
200
400
600
800
1000
1200
0 50 100 150 200 250 300 350 400
Time, days
Dif
fere
nce in
eig
en
valu
e, p
cm
P-C-0 (25d)
P-C-1 (25d)
P-C-1 (50d)
0.0E+00
2.0E-09
4.0E-09
6.0E-09
8.0E-09
1.0E-08
1.2E-08
0 50 100 150 200 250 300 350 400
Time, days
Co
ncen
trati
on
of
Xe-1
35
P-C-0 (25d)
P-C-1 (25d)
P-C-1 (50d)
P-C-1 (5d)
Description Case
no. Designated T,
days Method
P-C-0 (25d) 25 Predictor 1
P-C-1 ( 5d) 5 Predictor-Corrector a 3
P-C-1 (25d) 25 Predictor-Corrector 3
P-C-1 (50d) 50 Predictor-Corrector 4
E-P-C (50d) 50 Extended-Predictor-Corrector 5
a. Chosen to be the reference case
Burnup coupling scheme method & T
14
-200
0
200
400
600
800
1000
1200
0 50 100 150 200 250 300 350 400
Time, days
Dif
fere
nce in
eig
en
valu
e, p
cm
P-C-0 (25d)
P-C-1 (25d)
P-C-1 (50d)
0.0E+00
2.0E-09
4.0E-09
6.0E-09
8.0E-09
1.0E-08
1.2E-08
0 50 100 150 200 250 300 350 400
Time, days
Co
nc
en
trati
on
of
Xe-1
35
P-C-0 (25d)
P-C-1 (25d)
P-C-1 (50d)
P-C-1 (5d)
0.0E+00
5.0E+13
1.0E+14
1.5E+14
2.0E+14
2.5E+14
3.0E+14
3.5E+14
4.0E+14
0 50 100 150 200 250 300 350 400
Height, cm
Flu
x d
istr
ibu
tio
n, n
/s·c
m2
T(i)
T(i+1)
Description Case
no. Designated T,
days Method
P-C-0 (25d) 25 Predictor 1
P-C-1 ( 5d) 5 Predictor-Corrector a 3
P-C-1 (25d) 25 Predictor-Corrector 3
P-C-1 (50d) 50 Predictor-Corrector 4
E-P-C (50d) 50 Extended-Predictor-Corrector 5
a. Chosen to be the reference case
The effect of non-uniform coolant density
15
0.80
0.90
1.00
1.10
1.20
1.30
1.40
0 200 400 600 800 1000 1200
Time, days
Eig
en
valu
e
P-C-0
P-C-1
Averaged coolant density along the z-axis
1.10
1.15
1.20
1.25
1.30
1.35
1.40
0 50 100 150 200 250 300 350 400
Time, days
Eig
en
valu
e
P-C-0 (25d)
P-C-1 (25d)
P-C-1 (50d)
Realistic coolant density along the z-axis
Comparison between P-C and E-P-C
16
1.10
1.15
1.20
1.25
1.30
1.35
1.40
0 50 100 150 200 250 300 350 400
Time, days
k-e
ff
-50
50
150
250
350
450
550
Dif
fere
nc
e [
pc
m]
to P
-C-1
-25
d
P-C-1 (25d)
P-C-1 (50d)
E-P-C (50d)
DIFF (P-C-1)
DIFF (E-P-C)
4.5E-04
5.0E-04
5.5E-04
6.0E-04
6.5E-04
7.0E-04
7.5E-04
8.0E-04
8.5E-04
0 50 100 150 200 250 300 350 400
Time, days
Co
ncen
trati
on
of
U-2
35
-3.0%
-2.5%
-2.0%
-1.5%
-1.0%
-0.5%
0.0%
0.5%
1.0%
Rela
tive d
iffe
ren
ce t
o P
-C-1
-25d
P-C-1 (25d)
P-C-1 (50d)
E-P-C (50d)
DIFF (P-C-1)
DIFF (E-P-C)
Maximum difference (compared to P-C-1) in different core parameters
Maximum difference,
E-P-C (50d) vs. P-C-1 (25 d) Maximum difference,
P-C-1 (50d) vs. P-C-1 (25 d) Parameter
~20 pcm ~500 pcm k-eff
~0.5 % ~2.0 % Xe135
~0.1 % ~1.9 % U235
~0.2 % ~2.5 % Pu239
Conclusions
&
future work
Summary
17
Sensitivity studies of different coupling schemes were performed
Explicit Euler predictor method may be unstable for some 3D problems
Predictor-corrector method with Δt ~50d does not resolve the problem
EPC method allows using larger timesteps
Maintains sufficient accuracy
Future work:
How/When is the Thermal-hydraulic feedback should be applied?
Thank You for your attention!