ArticlePDF Available

Abstract and Figures

Breeding blankets are designed to ensure tritium self-sufficiency in deuterium-tritium fusion power plants. In addition to this, breeder blankets play a vital role in shielding key components of the reactor, and provide the main source of heat which will ultimately be used to generate electricity. Blanket design is critical to the success of fusion reactors and integral to the design process. Neutronic simulations of breeder blankets are regularly performed to ascertain the performance of a particular design. An iterative process of design improvements and parametric studies are required to optimize the design and meet performance targets. Within the EU DEMO program the breeding blanket design cycle is repeated for each new baseline design. One of the key steps is to create three-dimensional models suitable primarily for use in neutronics, but could be used in other computer-aided design (CAD)-based physics and engineering analyses. This article presents a novel blanket design tool which automates the process of producing heterogeneous 3D CAD-based geometries of the helium-cooled pebble bed, water-cooled lithium lead, helium-cooled lithium lead and dual-coolant lithium lead blanket types. The paper shows a method of integrating neutronics, thermal analysis and mechanical analysis with parametric CAD to facilitate the design process. The blanket design tool described in this paper provides parametric geometry for use in neutronics and engineering simulations. This paper explains the methodology of the design tool and demonstrates use of the design tool by generating all four EU blanket designs using the EU DEMO baseline. Neutronics and heat transfer simulations using the models have been carried out. The approach described has the potential to considerably speed up the design cycle and greatly facilitate the integration of multiphysics studies.
Content may be subject to copyright.
PAPER • OPEN ACCESS
Multiphysics analysis with CAD-based parametric breeding blanket
creation for rapid design iteration
To cite this article: Jonathan Shimwell et al 2019 Nucl. Fusion 59 046019
View the article online for updates and enhancements.
This content was downloaded from IP address 89.185.78.173 on 09/03/2019 at 00:47
1 © EURATOM 2019 Printed in the UK
Nuclear Fusion
Multiphysics analysis with CAD-based
parametric breeding blanket creation
for rapid design iteration
JonathanShimwell1, RémiDelaporte-Mathurin2, Jean-CharlesJaboulay3,
JulienAubert3, ChrisRichardson4, ChrisBowman5, AndrewDavis1,
AndrewLahiff1, JamesBernardi6, SikanderYasin7,8 and XiaoyingTang7,9
1 Culham Centre for Fusion Energy (CCFE), Culham Science Centre, Abingdon, Oxfordshire OX14 3DB,
United Kingdom of Great Britain and Northern Ireland
2 Département Thermique-Énergétique, Polytech Nantes, Université de Nantes, Rue Christian Pauc,
CS50609, 44306 Nantes Cedex 3, France
3 Den-Département de Modélisation des Systèmes et Structures (DM2S), CEA, Université Paris-Saclay,
F-91191 Gif-sur-Yvette, France
4 BP Institute, Bullard Laboratories, Madingley Road, Cambridge CB3 0EZ, United Kingdom of Great
Britain and Northern Ireland
5 York Plasma Institute, University of York, Heslington YO10 5DD, United Kingdom of Great Britain
and Northern Ireland
6 University of Cambridge, The Old Schools, Trinity Lane, Cambridge CB2 1TN,
United Kingdom ofGreat Britain and Northern Ireland
7 University of Manchester, Oxford Road, Manchester M13 9PL, United Kingdom of Great Britain
and Northern Ireland
8 Blackpool and The Fylde College, Asheld Rd, Blackpool FY2 0HB, United Kingdom of Great Britain
and Northern Ireland
9 School of Mechanical Engineering, Shanghai Jiao Tong University, Shanghai 200240, China
E-mail: mail@jshimwell.com
Received 27 September 2018, revised 21 December 2018
Accepted for publication 18 January 2019
Published 8 March 2019
Abstract
Breeding blankets are designed to ensure tritium self-sufciency in deuteriumtritium
fusion power plants. In addition to this, breeder blankets play a vital role in shielding key
components of the reactor, and provide the main source of heat which will ultimately be used
to generate electricity. Blanket design is critical to the success of fusion reactors and integral
to the design process. Neutronic simulations of breeder blankets are regularly performed to
ascertain the performance of a particular design. An iterative process of design improvements
and parametric studies are required to optimize the design and meet performance targets.
Within the EU DEMO program the breeding blanket design cycle is repeated for each new
baseline design. One of the key steps is to create three-dimensional models suitable primarily
for use in neutronics, but could be used in other computer-aided design (CAD)-based physics
and engineering analyses. This article presents a novel blanket design tool which automates
the process of producing heterogeneous 3D CAD-based geometries of the helium-cooled
pebble bed, water-cooled lithium lead, helium-cooled lithium lead and dual-coolant lithium
lead blanket types. The paper shows a method of integrating neutronics, thermal analysis
J. Shimwell etal
Multiphysics analysis with CAD-based parametric breeding blanket creation for rapid design iteration
Printed in the UK
046019
NUFUAU
© EURATOM 2019
59
Nucl. Fusion
NF
10.1088/1741-4326/ab0016
Paper
4
Nuclear Fusion
IOP
Original content from this work may be used under the terms
of the Creative Commons Attribution 3.0 licence. Any further
distribution of this work must maintain attribution to the author(s) and the title
of the work, journal citation and DOI.
International Atomic Energy Agency
2019
1741-4326
1741-4326/19/046019+12$33.00
https://doi.org/10.1088/1741-4326/ab0016
Nucl. Fusion 59 (2019) 046019 (12pp)
J. Shimwell etal
2
and mechanical analysis with parametric CAD to facilitate the design process. The blanket
design tool described in this paper provides parametric geometry for use in neutronics
and engineering simulations. This paper explains the methodology of the design tool and
demonstrates use of the design tool by generating all four EU blanket designs using the EU
DEMO baseline. Neutronics and heat transfer simulations using the models have been carried
out. The approach described has the potential to considerably speed up the design cycle and
greatly facilitate the integration of multiphysics studies.
Keywords: fusion, parametric, CAD, neutronics, 3D model, breeder blanket
(Some guresmay appear in colour only in the online journal)
1. Introduction
Breeding blankets are designed to full several high-level plant
requirements, including tritium self-sufciency, shielding
non-sacricial components from the intense neutron ux and
producing heat which is ultimately used to generate elec-
tricity. Designing and engineering components for use within
fusion reactors is challenging due to the high radiation uxes
and signicant heat loads that they experience. Maintaining
an operational and safe component within the inner vessel of a
fusion reactor presents a range of difculties; however, adding
functional requirements such as tritium breeding, heat genera-
tion and heat removal further complicates the task.
Methods of design optimization such as parameter studies
and a designing by analysis approach are possible avenues
for designing fusion reactor components that could provide
solutions to this challenge. Such methods rely on human intui-
tion and iterative analysis of models to close in on an optimal
solution. Performing analysis in an isolated discipline will
only nd the optimal solution for performance metrics that
are obtainable within that discipline. For instance neutronics
optimizations may nd the tritium breeding ratio (TBR) but
may nd unacceptable temperatures. Multiphysics analysis
is required to optimize component design. To maintain data
provenance it would be preferable to have a single model basis
when sharing data between analysis techniques.
Traditionally, models are generated for neutronics using
constructive solid geometry (CSG) and the models are suit-
able for use in parametric studies. Engineering analysis tends
to require computer-aided design (CAD) models and CSG
models are typically not compatible with engineering pro-
grams. Models for use in engineering analysis are often cre-
ated via graphical user interfaces. The process of creating new
engineering and neutronics models can be a time-consuming
exercise. This is compounded since the models must be gen-
erated with the release of a new EU DEMO baseline design,
and there are also four EU blanket designs for each iteration.
To analyze the performance of different designs within
the parameter space it would be desirable to be able to pro-
duce an accurate three-dimensional (3D) geometry rapidly.
Adopting a common geometry format would allow geometry
to be used in multiple domains. Allowing ne details (such as
cooling pipes) to be included or excluded during model gen-
eration can facilitate specic requirements of the particular
analysis. Use of open source geometry-producing software
such as FreeCAD [1] (used for this project), Salome [2] or
PythonOCC [3] can be used to quickly generate parametric
CAD geometry which can be exported into a variety of for-
mats. CAD les in STEP format [4] are an open le standard
compatible with engineering simulation software. STEP les
can be easily converted into surface faceted geometry (e.g.
h5m or STL les) for use in neutronics codes such as Serpent
2 [5] (used for this project) and DAG-MCNP5/6 [6]. Ideally
any solution to making component models would be exible
enough to work with new DEMO baseline models and also to
produce different blanket designs.
2. Method
2.1. Geometry creation
It is clear that in order for any advanced method to generate
any detailed geometry it must require the minimum human
intervention, so it was determined that a software library
should be created that allows arbitrary geometric operations
to be performed, with the end goal of creating parametrically
built blanket modules for the EU Demo program. This soft-
ware is called the Breeder Blanket Model Maker and can be
found here [7]. Routines for the generation, modication and
serialization of blanket envelopes were created that ultimately
automatically produce detailed heterogeneous blankets for
use in DEMO modelling. Demonstration neutronic and heat
diffusion simulations were performed to illustrate the ease of
carrying out parameter studies. Parametric models of all four
EU blanket designs were generated to demonstrate the design
tool. These include the helium-cooled pebble bed blanket
(HCPB) [810], helium-cooled lithium lead blanket (HCLL)
[11, 12], water-cooled lithium lead blanket (WCLL) [13] and
dual-coolant lithium lead blanket (DCLL) [14]. The process
has been broken down into two parts, the rst owchart (see
gure1) summarizes the construction of all the non-breeder
zone components, this includes the rst wall, armour and rear
plates. The second owchart (see gure2) summarizes the
construction of the breeder zone structure.
The breeding blanket is segmented into different mod-
ules which have different shapes, orientations and positions
depending upon the positioning within the reactor (see stage 1
in gure1). The blanket designs share some common features
Nucl. Fusion 59 (2019) 0 46019
J. Shimwell etal
3
such as: lleted corners on the toroidal HCPB, HCLL and
WCLL designs or poloidal DCLL edges.
2.1.1. Structural components. The procedure for the creation
of the structural part of the blanket module is performed rst.
A number of key parameters dene the structure of the blan-
ket: rst wall armour thickness, rst wall thickness, rear plate
count and thickness, and end cap thickness. An automated
procedure regarding the automatic construction of a full
detailed blanket structure was dened, and the full implemen-
tation details can be found in [7]. Additional ne detail is also
included at the users discretion including the introduction of
llets and cooling channels. The overall programmatic ow
is shown in gure2, where differing blankets follow different
logical routes.
2.1.2. Breeder zone components. In order to generate the
full heterogeneous blanket description the internal detail
of the breeder zone must be generated. This is a two-stage
process: rst the cooling structure is generated starting from
the breeder zone envelope and generating the cooling struc-
ture from it then the cooling structure is subtracted from the
breeder zone envelope, leaving the non-structural breeding
material (lithium lead/lithium ceramic/neutron multiplier).
2.1.3. Cooling structure generation. The segmentation of the
cooling structure varies for each of the four breeder blanket
modules, but is almost entirely a combination of poloidal-,
toroidal- and radial-based segmentations. The cooling structure
of the HCLL advanced plus module [12] can be represented
by a series of poloidal segmentations with alternating layers of
1. Selection of example blanket
moduleenvelope from the EU
DEMO baseline [15]
2. Example blanketenvelopeshowing the
front face (green), poloidal edges (blue)
and toroidal edges (red)
3. Example blanket envelope with filleted
poloidal edges (right) and toroidal edges
(left) the fillet radius has been increased
to clearlyshow the operation
4. Example blanket envelope with
first wall armour shownin green
5. Example blanket envelope with
first wall shown in green
6.
First wall cooling
c
hannels (optional)
6. Firstwall co
oling
channels (optional)
7. Example blanket envelope with
end caps shown in green
8. Example breeder zoneenvelope
with backwall components
shown in blue and green
Figure 1. Automated workow for generating rst wall, end caps and back plates from EU DEMO baseline [15] blanket envelopes.
Nucl. Fusion 59 (2019) 0 46019
J. Shimwell etal
4
stiffening plates. This can be reproduced using alternate poloi-
dal segmentation with alternating poloidal extrusion lengths, as
shown in gure2 section 2. In the case of the HCPB module
there are alternating poloidal layers of lithium ceramic and neu-
tron multiplier between the stiffening plates. The poloidal seg-
mentation functions have been designed to allow any number
of layer repetitions; this is demonstrated in stage 3 of gure2).
The wedge-shaped regions at the upper and lower extremities
of the HCPB module are lled with neutron multiplier and are
therefore not considered part of the cooling structure. The soft-
ware is able to identify these wedge-shaped regions and group
them with the other neutron multiplier regions.
Radial cuts, and thereby radial segmentation, are also
implemented; these are required for both the WCLL and the
DCLL blanket designs. The WCLL cooling structure can be
generated with a combination of poloidal and toroidal seg-
mentation (see stage 6 in gure 2). Both the toroidal and
poloidal directions have alternating thicknesses for the struc-
tural plates and the lithium lead regions. Every other layer of
the poloidal structural plate has an offset from the rst wall
that allows lithium lead to ow between plates. The poloidal
segmentation for such a model can be carried out in a similar
way to the HCLL, but the WCLL has an additional complica-
tion which requires radial segmentation. The WCLL model
requires toroidal segmentation and additionally requires that
the upper and lower wedge volumes should be considered to
be entirely lithium lead. The resulting product of the toroidal
segmentation can be seen in stage 6 in gure2.
The DCLL cooling structure can be formed from a com-
bination of radial and toroidal segmentation plus some detail
to guide the ow of lithium lead. The procedure used was
to rst radially segment the blankets into three or ve parts
(depending upon the radial depth of the blanket). In general
most of the inboard blankets accommodate three radial layers
and the outboard blankets accommodate ve radial layers. The
addition of toroidal segmentation to the previously radial seg-
mented breeder zone forms the rst stage of the DCLL model
(see stage 8 of gure2). The DCLL blanket design allows the
lithium lead to ow around the structure. An additional struc-
tural component at the upper end of each blanket module is
also required by the DCLL design, the only additional compli-
cation is that a Boolean subtraction with the rst radial layer is
also required to obtain the desired structural plate shape (see
stage 10 of gure2).
Breeder zone
1. Poloidal
segmentation
4. Toroidal
segmentation
7. Radial
segmentation
6. Alternate poloidal
layers cut with
radiallayer
HCLL
HCPB
WCLL
DCLL
9. Radial plate
cut with upper
andlower offset
8. Combination of
radial and toroidal
segmentation
10.Addition o
f
upper channel
guide
3. Poloidal segmentation
with pair of repeating
layers andnoplates in
theend regions
2. Poloidal segmentation
with single repeating
layerand platesin
theend regions
5. Combination of
toroidal andpoloidal
segmentation
Figure 2. Creation of internal breeder zone structure using a combination of toroidal, poloidal and radial segmentations.
Nucl. Fusion 59 (2019) 0 46019
J. Shimwell etal
5
2.1.4. Breeding material. The complete description of the
breeder zone comprises the description of the cooling struc-
ture and the description of the breeding material. The nal
stage is take the original breeder zone envelope and subtract
the newly created cooling plates/stiffening plates from it,
thus dening the description of the complete breeding zone.
HCLL, WCLL and DCLL all require that the end regions are
lithium lead and HCPB requires the end regions to be neutron
multiplier.
2.1.5. Slice geometry generation. The slice geometry used in
the thermal simulation can also be generated automatically.
The procedure for generating the slice geometry of the HCLL
is outlined in gure3. The user species the blanket module
from which to extract a slice. A slice envelope is created with
a poloidal height equal to the poloidal height of a stiffening
plate plus the poloidal height of the breeder zone. The enve-
lope poloidal position is centred around the stiffening plate
so that the slice contains half a breeder zone above and half a
breeder zone bellow the stiffening plate. The cooling channel
positions were found by offsetting the rst wall towards the
rear of the blanket, with the size of the offset progressively
increasing. Surface identication for cooling surfaces was
carried out by merging the coolant volumes with the structure
volumes and searching for merged surfaces in the resulting
geometry. The identication was manually checked after this
stage to ensure a robust procedure.
2.2. Parametric geometries
As a result of the method previously described there is now an
automated procedure for obtaining semi-detailed CAD geom-
etry for the HCPB, HCLL, WCLL and DCLL blankets. The
process relies on a library of common functions which can
be mixed and matched to create particular blanket designs.
The breeder blanket design tool is released as an open source
project under the Apache 2.0 license and distributed via the
UKAEA Github repository [7]. The software is subject to a
test suite and the build status is updated automatically with
every commit. Continuous integration practices are employed
using Circle CI and Docker.
The model construction process is parametric, which
allows models required for parameter studies to be generated
rapidly. Currently the parameters that a user can input are:
lename of blanket envelope required for segmentation
blanket type (HCPB, HCLL, WCLL, DCLL) which also
denes the geometry layout as shown in gures1 and 2
poloidal llet radius for the rst wall and rst wall armour
toroidal llet radius for the rst wall and rst wall armour
rst wall armour thickness
rst wall thickness
end cap thickness
thickness of each rear plate
thickness of each poloidal segmentation
thickness of each toroidal segmentation
Complete HCLL
blanketmodule
Envelope reserved
for slice
Geometry commonto the blanket
module and the slice envelope
Cen
tral stiffening plate with
added cooling channels
Stiffening plate dividedinto three
sections (upper, lower and central)
Homogenized stiffening plate
extracted for further
detailing
Figure 3. Creation of the slice geometry structure used in thermal analysis.
Nucl. Fusion 59 (2019) 0 46019
J. Shimwell etal
6
thickness of each radial segmentation
rst wall coolant channel poloidal height
rst wall coolant channel radial height
rst wall coolant channel pitch
rst wall coolant channel offset from the front face
output le format (STEP or STL) and tolerance.
Not all parameters are needed for each design as some are not
applicable, for instance the breeder zone in the HCPB blanket
has no radial segmentation option and does not require
this input. Figure 4 shows each of the four blanket designs
formed from a particular module from the baseline DEMO
model [15]. Currently the tool requires that the rst wall is a
at plane to determine the location of the internal structure.
The tool would therefore need some alterations to work with
blanket envelopes with curved front surfaces (i.e. full banana-
shaped segments).
The process of building a blanket module from an enve-
lope typically takes a few seconds on a single core. Build time
depends on the input parameters, as many very small layers
would necessitate more Boolean operations than for the case
of a few large layers. The process is parallelizable, and there-
fore a model such as the EU DEMO with 26 blanket modules
typically takes less than 5 min on a quad core Intel i5 7600
CPU.
3. Results
3.1. Neutronics model creation
Once the parametric CAD models have been created, one
potential use is in neutronics simulations. There are several
routes from CAD to neutronics models, such as conversion
to CSG using conversion software such as McCad [16] or
SuperMC [17]). Alternatively the use of faceted geometry is
also possible. Previously, parameter studies for fusion blanket
optimization have converted parametric CAD models to CSG
(a) HCPB blanket (b) HCLL blanket (c) WCLL blanket (d) DCLL blanket
Figure 4. Example parametric blanket modules; parameter values have been enlarged in some cases to increase the visibility of
components.
(a) Slice of EU Demo viewed from above (b) Slice of EU Demo side view.
Figure 5. A neutronics model of EU DEMO using faceted geometry (STL) with detailed HCLL blankets, showing plasma (purple),
lithium lead (green), magnets (red), and structural steels (grey).
Nucl. Fusion 59 (2019) 0 46019
J. Shimwell etal
7
models using McCad and performed the simulation using
MCNP [18]. This study opted to simulate using faceted geom-
etry in the STL le format and perform the neutronics simula-
tion using Serpent 2, which natively supports STL geometry.
The process of converting from STEP to STL is quicker than
from STEP to CSG and the results are easier to visually verify.
To demonstrate practical use of the parametric geometry, a
series of tritium breeding simulations were obtained for the
HCLL. The poloidal height of the lithium lead sectionsand
the 6Li enrichment of the lithium lead were varied indepen-
dently. The thickness of the rst wall and the stiffening plates
were varied simultaneously with the poloidal height of the
lithium lead sectionsusing equations(1) and (2).
=CD+
Cp×LL2
p
=0.01 +9.552
102
LL
(1)
CPT =
P
s
×LL
p
1.1 ×SmD
=3.332
×
LLp
.
(2)
Here FWT is the rstwall thickness (m), CPT is the stiffening
plate thickness (m),
CD
is the radial width of the rst wall
cooling channel (0.01 m),
Cp
is the specic heat capacity
(J K–1),
Ps
is the coolant pressure (10 MPa for the helium
coolant), SmD is the stress limit criterion for EUROfer
(274 MPa) [19] in case of accident and
LLp
(m) is the poloidal
height of the lithium lead sections. These equations are
described in more detail in [11] and [20]. This HCLL study
is a demonstration of the model-making tool developed and
thermal-mechanical constraints are not taken into account.
Halton sampling [21] was used as the sampling technique to
select points within the parameter space. The parameter space
encompassed blankets with a poloidal height of lithium lead
between 0.01 m and 0.12 m and 6Li enrichment between 0%
and 100%. The requested simulation points found using the
Halton sampling method are shown as red crosses in gure6.
These requested input parameters were uploaded to a online
cloud-based database (Mongo Atlas [22]). Entries within
the database were agged with in progress, completed or
requested to indicate their simulation status. The central-
ized accessible database allowed independent simulations to
be carried out in parallel and the results to be coordinated. A
containerized workow was implemented using Docker [23]
which held the breeder blanket model-maker software along
with all the dependences required, such as neutron interaction
data, FreeCAD, Python and MongoDB. Containers were then
launched on EGI FedCloud resources [24] and connected to
the cloud database to retrieve their simulation input param-
eters (6Li enrichment and lithium lead poloidal height) and
set the database entry for the simulation to in progress.
Once the simulation input parameters were received the CAD
models were automatically built and combined with material
Figure 6. Showing interpolated TBR values with a
5σ
condence for a range of different 6Li enrichments and poloidal lithium lead heights.
The Gaussian process software used [30] was able to t the TBR values along with their statistical errors and nd the condence values.
The reference design HCLL has 90% 6Li enrichment, 34.5 mm of poloidal lithium lead and achieves a TBR of 1.235.
Nucl. Fusion 59 (2019) 0 46019
J. Shimwell etal
8
properties to create the neutronics model. Upon completion
of the neutronics simulation the results (TBR and heating tal-
lies) were uploaded to the cloud-based database and the entry
ag was set to completed. At this point the container would
either continue with the next requested simulation in the data-
base or terminate and free up resources if all the requested
simulations were complete.
FENDL 3.1b [25] cross sectionswere used for neutron trans-
port. Neutronics materials denitions from the EUROfusion
material composition [26] were used in the model and a para-
metric plasma source based on [27] with plasma parameters
from [28] was used. The number of starting particles run for
each TBR simulation was
1×107
.
3.2. Neutronics simulations
The models generated are suitable for neutronics simulations
and gure 5 shows the blanket models within a Serpent 2
geometry. Figure6 shows the resulting TBR values from a
neutronics parameter study. The TBR was found to change
with the poloidal height of lithium lead. Models with a
small lithium lead poloidal height contain a relatively large
EUROfer fraction compared with models with a large lithium
lead poloidal height. This is due to the large number of stiff-
ening plates and it appears to have reduced tritium produc-
tion. However, models with a large poloidal height can also
have large quantities of EUROfer in the breeder blanket. As
the poloidal height of lithium lead increases the thickness of
the EUROfer rst wall (see equation(1)) and the thickness of
the EUROfer stiffening plates (see equation(2)) also increase.
There appears to be an optimal poloidal height which
becomes more pronounced with 6Li enrichment. Figure 5
shows the variation of TBR with 6Li atom fraction within the
lithium lead. Increasing 6Li enrichment shows an increase in
TBR as conrmed by previous studies [29]. Figure5 shows the
variation of TBR with the poloidal height of the lithium lead
regions within the breeder zone of the blanket. The poloidal
height has less of an effect on TBR but can be optimized.
The achievable increase in TBR depends upon the 6Li enrich-
ment. Figure 6 shows that each different enrichment of 6Li
has a different optimal height for the poloidal lithium lead. At
90% 6Li enrichment a TBR increase of nearly 0.1 is possible
(see gure 6). The size of the
5σ
condence regions varies
depending on the proximity and statistical error of nearby sim-
ulations. This is most noticeable towards the extremities of the
search space where there are fewer simulations and the size
of the
5σ
condence regions is larger. The variation in height
of the poloidal lithium lead also has mechanical considera-
tions as the rst wall thickness and stiffening plate thickness
are also increased when the lithium lead poloidal height is
increased (see equations(1) and (2)). This helps explain why
we observe an optimal lithium lead poloidal height for TBR
values from the neutronics model.
A maximum TBR of
1.278 ±0.010
(
5σ
condence) was
found using Gaussian process software [30] to t the simu-
lation data and statistical error (see gure 6). The highest
TBR value was found for a blanket design with a lithium lead
poloidal height of 0.061 m and a 6Li enrichment of 100%.
Additional constraints such as the capability of the thermal-
hydraulic design to cool the structure with reasonable pressure
drops must also be considered, and the maximum TBR design
may not meet such requirements. Thermal modes are devel-
oped in the next section.
3.3. Creation of the heat diffusion model
A slice of the HCLL blanket geometry was used to create
a simplied model of the temperature eld in the tungsten,
EUROfer and lithium lead. Dimensions for the rst wall
thickness and stiffening plate thickness were calculated using
equations(1) and (2) with the poloidal height of lithium lead
set at 34.5 mm. The model contains a single stiffening plate
with cooling channels and lithium lead either side, encased
with a rst wall; the material layout is shown in gure8. A
mesh with 2.3 million tetrahedra was created using Trelis [31]
complete with boundary conditions and volumetric source
terms. Heat diffusion simulations were carried out using
FEniCS [32] which was able to apply boundary conditions
to surfaces and temperature-dependent materials proper-
ties to different volumes. The heat diffusion equationused is
described by equation(3).
2
(λ
T
)+
Q
=
0
(3)
where T is the temperature in K,
λ
is the thermal conductivity
of the given material expressed in W m1 K1 and Q is the
volumetric source term in W m3. As this equation is solved
using the nite elements method, equation (3) needs to be
brought to its weak formulation (or variational for mulation)
as follows:
λ
(
δ
δT
δn
vdS
v
Tdx)=
vQdx
R
3
(4)
where
is the domain on which equation(4) is solved, n is the
normal direction on the external surface and
v
is a test func-
tion. The integration term on the boundary of
is determined
by the boundary conditions.
3.3.1. The Robin boundary condition. The Robin boundary
condition allows the assignment of a convective heat ux on
a boundary. In equations(5) and (6), it is shown that the heat
ux depends on the temperature of the uid and the convec-
tive coefcient h (W m2 K1). This coefcient depends on
the type of convection (natural, forced, laminar or turbulent)
and the uid in contact with the surface:
λ
dT
dn
=hFWCC(T
TFWCC)on Γ
FWCC
(5)
λ
dT
dn
=hHSPCC(T
THSPCC)on ΓHSPCC
.
(6)
ΓFWCC
and
ΓHSPCC
are surface domains and are shown
in blue and purple, respectively, in gure 7. This condi-
tion is used on the surfaces of the helium cooling channels.
The coefcients h have been calculated for the rst wall
cooling channels (FWCC) and horizontal stiffening plate
Nucl. Fusion 59 (2019) 0 46019
J. Shimwell etal
9
cooling channels (HSPCC). They are determined using
Gnielinski correlation (equation (7)) with the parameters in
table1 in accordance with [33].
Nu
D=
hD
h
λHe
=(ξ/
8
)(
Re
D
1000
)
Pr
1+12.7(ξ/8)
1/2
(Pr
2/3
1)
(7)
where NuD is the Nusselt number,
Dh
is the hydraulic diam-
eter in m,
λHe
is the thermal conductivity of the uid in W
m1 K1, ReD is the Reynolds number and Pr is the Prandtl
number. We consider a smooth surface, and thereby the
DarcyWeisbach friction factor
ξ
is then given by:
ξ=(
0.790
ln
ReD1.64
)
2.
3.3.2. Neumann boundary conditions. By using Neumann
boundary conditions, xed heat ux can be assigned to the
front wall armour surface shown in red in gure7, as described
in equation(8):
λ
δT
δn
=Jon ΓFW
.
(8)
Here, J is a xed ux density in W m2 on a boundary. This
condition is used on the surface
ΓFW
which corresponds to the
front wall (shown in red in gure7) with a ux of J = 0.5 MW
m2 which corresponds to the heat ux emitted by the plasma.
The rest of the external surfaces are considered as insu-
lated. This assumption is valid as long as these surfaces are
part of a vacuum and are not exposed to an intense heat ux.
The values of the thermal conductivity
λ
in W m1 K1 in
equation(4) were found in [34].
Finally, the distribution of the volumetric source term Q in
equations(3) and (4) was taken from [35] and is inputted into
the nite element model using the following equations:
Q=7.53e8.98rMW m3in EUROfer
(9)
Q=23.2e71.74rMW m3in tungsten
(10)
Q=9.46e6.20rMW m3in LiPb
(11)
where r is the radial distance from the front face of the breeder
blanket in m. The resulting spatial distribution of volumetric
heating (Q) in MW m3 is shown in gure9.
3.4. Heat diffusion simulations
Using the same CAD-generated models as in section3.2, heat
diffusion simulations have been performed. The steady state
solution of the temperature eld is shown in gure10. Thanks
to these simulations, we are able to determine the maximum
temperature reached by each material and determine if the
design allows the materials to stay within their maximum oper-
ating temperature limits (550 °C for EUROfer and 1300 °C
for tungsten [36]). Although the design limit of 550 °C is
reached in part of the stiffening plates, this design could be
rened with additional cooling channels to reduce the temper-
ature. Additional assessments involving computational uid
dynamics could be performed to check the accuracy of the
temperatures predicted and the conservatism of assumptions
made in this simulation. The maximum temperatures are
shown in table2.
Figure 7. Cut of a HCLL module slice showing the surfaces used for boundary conditions:
ΓFW
(red),
ΓFWCC
(blue) and
ΓHSPCC
(purple).
Table 1. Parameters used for the determination of the convective coefcients
hFWCC
and
hHSPCC
.
Symbol Description Value of FWCC Value of HSPCC
ReDReynolds number
1.310 ×105
3.018 ×104
Pr
Prandtl number
6.599 ×101
6.599 ×101
ξ
DarcyWeisbach friction factor
3.000 ×102
2.000 ×102
Dh
Hydraulic diameter (m)
1.500 ×102
3.000 ×103
hConvective coefcient (W m2 K1)
4.531 ×103
4.848 ×103
Tcoolant
Average temperature of coolant (K)
6.230 ×102
7.160 ×102
Nucl. Fusion 59 (2019) 0 46019
J. Shimwell etal
10
Figure 8. HCLL module slice showing the materials: tungsten (brown), lithium lead (green) and EUROfer (grey).
Volumetricheating (×106Wm
3)
0.21 246810 12 14 16 18 20 22 23
(a) toroidalview(b) poloidalview
Figure 9. Volumetric heating source term applied to a slice from middle of the equatorial outboard blanket module in W m3.
(a) isometriccut
Temperature (K)
622 660 700 740 780820
863
(b) toroidal view
(d)poloidalview (c) clipped poloidalview
Figure 10. Resulting temperature eld of a slice from the middle of the equatorial outboard blanket module in K.
Nucl. Fusion 59 (2019) 0 46019
J. Shimwell etal
11
3.5. Discussion
The simulation results show that the materials stay within
their maximum operating temperature limits. However, one
must be aware that some assumptions have been made that
could introduce inaccuracy. The convective boundary con-
dition on
ΓHSPCC
and
ΓFWCC
are set with constant average
coolant temperatures along the channel. This assumption is
reliable in term of ux balance in the module although it may
introduce errors in the region near the inlets and the outlets of
the channels where the coolant should be, respectively, colder
and hotter than the average. One way to solve this issue would
be to allow
Tcoolant
to vary along the channel. Actual cooling
channel designs for HCLL might also differ from the one used
in this paper, leading to variation in the temperature eld. One
can also notice that the simulation has been run in steady state.
Running transient simulation could allow new hot spots to be
spotted before the module reaches steady state. The volu-
metric heat source used in this paper is a tted exponential
function mapped on a mesh using discontinuous Galerkin ele-
ments of order 0. In other words, the volumetric heat source
has been discretized. Again, this might lead to inaccuracy in
some big cells. This can be solved by rening the mesh, as
currently the model contains 2.3 million tetrahedral elements.
Finally, the lithium lead ow is not modelled in this paper for
simplication purposes. Lithium lead has been considered as
a solid. However, considering the very low velocity of lithium
lead in HCLL modules, this assumption can be made.
4. Conclusion
A design tool capable of generating parametric designs for
fusion breeder blankets has been demonstrated on single-
module blanket envelopes for HCLL, HCPB, WCLL and
DCLL. A wide range of design parameters can be changed
to generate CAD geometry for use in parameter studies. The
geometry generated is available in CAD format (STEP) and
faceted geometry (STL and h5m). Conversion to CSG for
neutronics simulation is achievable via existing software such
as McCad or MCAM. The option of faceted geometry allows
CSG geometry to be avoided in favour of more CAD-based
neutronic simulation techniques such as DAGMC or Serpent
2. The provision of CAD geometry also enables manipulation
to be performed with standard CAD software as opposed to
CSG geometry where manipulation of the shapes is less con-
venient. A demonstration neutronics parameter study was per-
formed, where poloidal lithium lead height and 6Li enrichment
were varied. This was done in order to optimize TBR for the
HCLL module. A maximum TBR of
1.278 ±0.010
(
5σ
con-
dence) was found using Gaussian processing to t the data.
The highest TBR value was found for a blanket design with a
lithium lead poloidal height of 0.061 m and a 6Li enrichment
of 100%. Currently the tool allows for a large range of specic
blanket module geometries to be made for use in simulations.
Thermal-mechanical and other constraints are not taken into
account when constructing the geometries and future research
will be required to identify allowable design parameters that
satisfy thermal-mechanical and thermal-hydraulic require-
ments. The geometry created can also be used in nite element
and nite volume software to simulate heat diffusion, tritium
diffusion and stress. A simplied heat diffusion problem was
demonstrated in this paper. The maximum temperature within
the different materials present in the midplane geometry slice
of the HCLL outboard equatorial blanket module was found to
be similar to previous research [33]. Maximum temper atures
were 774 K for tungsten, 863 K for lithium lead and 824 K
for EUROfer. The work performed here shows the value of
in silico design processes, multiphysics workows and, criti-
cally, integration with automated systems for the generation,
submission and analysis of calculations. The workow was
built using modern scalable techniques to run the simulations
(containerized cloud computing), store the output of the simu-
lations (centralised cloud databases) and analyze the results
(machine learning). The tool is open for extension and addi-
tional analysis such as thermal stress, mechanical stress and
tritium diffusion should be the next steps for development.
Acknowledgments
The authors would like to acknowledge the nancial support
of EUROfusion and EPSRC. This work has been carried out
within the framework of the EUROfusion Consortium and
has received funding from the Euratom research and training
programme 20142018 under grant agreement no. 633053.
The views and opinions expressed herein do not neces-
sarily reect those of the European Commission. This work
has also been part-funded by the RCUK Energy Programme
(grant number EP/I501045/1). This work beneted from ser-
vices and resources provided by the fedcloud.egi.eu Virtual
Organization, supported by the national resource providers
of the EGI Federation. The authors would also like to thank
the HCLL, HCPB, DCLL and WCLL breeder blanket design
teams, and Dr Lidija Shimwell Pasuljevic and Helen Gale.
ORCID iDs
Jonathan Shimwell https://orcid.org/0000-0001-6909-0946
Rémi Delaporte-Mathurin https://orcid.org/0000-0003-
1064-8882
Chris Richardson https://orcid.org/0000-0003-3137-1392
Andrew Davis https://orcid.org/0000-0003-4397-0712
James Bernardi https://orcid.org/0000-0001-9229-4613
Table 2. Maximum temperatures of each material.
Material
Maximum temperature
(K) (°C)
Tungsten 774 501
Lithium lead 863 590
EUROfer stiffening plates 824 550
EUROfer rst wall 767 494
Nucl. Fusion 59 (2019) 0 46019
J. Shimwell etal
12
Sikander Yasin https://orcid.org/0000-0001-8701-6205
Xiaoying Tang https://orcid.org/0000-0002-7894-6774
References
[1] RiegelJ., MayerW. and van HavreY. 20012018 Freecad
version 0.17 (http://freecadweb.org/)
[2] RibesA. and CaremoliC. 2007 Salome platform component
model for numerical simulation Proc. of the 31st Annual
Int. Computer Software and Applications Conf. (24–27
July 2007, Beijing, China) pp 55364 (http://conferences.
computer.org/compsac/2007/)
[3] PaviotT. 20152018 PythonOCC3D CAD for python
(www.pythonocc.org/)
[4] ISO 10303-21 2016 Industrial automation systems and
integrationProduct data representation and exchange
Part 21: implementation methods: clear text encoding of
the exchange structure (International Organization for
Standardization)
[5] LeppänenJ. etal 2015 The Serpent Monte Carlo code: status,
development and applications in 2013 Ann. Nucl. Energy
8214250
[6] WilsonP. etal 2010 Acceleration techniques for the direct
use of CAD-based geometry in fusion neutronics analysis
Fusion Eng. Des. 85 175965
[7] ShimwellJ. 2018 Breeder blanket model maker (https://doi.
org/10.5281/zenodo.1421059)
[8] HernàndezF. etal Overview of the HCPB research activities
in EUROfusion IEEE Trans. Plasma Sci. 46224761
[9] HernàndezF. etal A new HCPB breeding blanket for the EU
DEMO: evolution, rationale and preliminary performances
Fusion Eng. Des. 1248826
[10] PereslavtsevP. etal 2017 Neutronic analyses for the
optimization of the advanced HCPB breeder blanket design
for DEMO Fusion Eng. Des. 1249104
[11] AubertJ. etal 2018 Status of the EU DEMO HCLL breeding
blanket design development Fusion Eng. Des. B 136142832
[12] JaboulayJ. etal 2017 Nuclear analysis of the HCLL blanket
for the European DEMO Fusion Eng. Des. 124896900
[13] Del NevoA. etal 2017 WCLL breeding blanket design and
integration for DEMO 2015: status and perspectives Fusion
Eng. Des. 1246826
[14] PalermoI. etal 2017 Neutronic design studies of a conceptual
DCLL fusion reactor for a DEMO and a commercial power
plant Nucl. Fusion 56016001
[15] PereslavstevP. etal 2013 Generation of the MCNP model that
serves as a common basis for the integration of the different
blanket concepts EFDA D 2M7GA5 V.1.0
[16] LuL. etal 2017 Improved solid decomposition algorithms for
the CAD-to-MC conversion tool McCad Fusion Eng. Des.
124126972
[17] WuY. etal 2015 CAD-based Monte Carlo program for
integrated simulation of nuclear system SuperMC Ann.
Nucl. Energy 821618
[18] ShimwellJ. etal 2017 Automated parametric neutronics
analysis of the helium cooled pebble bed breeder blanket
with Be12Ti Fusion Eng. Des. 1249403
[19] AFCEN 2018 RCC-MRx, design and construction rules for
mechanical components of nuclear installations (http://
afcen.com/en/publications/rcc-mrx/170/rcc-mrx-2018)
[20] PeyraudJ. 2017 Optimization of the horizontal stiffening
plates channels layout design for the advanced-plus concept
of the DEMO HCLL breeding blanket DEN/DANS/DM2S/
SEMT/BCCR/NT/17-013/A
[21] HaltonJ. 1964 Radical-inverse quasi-random point sequence
Commun. ACM 77012
[22] MongoDB 2018 Mongo Atlas automated cloud MongoDB
service (www.mongodb.com/cloud/atlas)
[23] Docker 2018 Container platform. Build manage and secure
your apps anywhere (https://docs.docker.com/)
[24] EGI FedCloud 2018 Cloud platform. Build Manage and
Secure your apps anywhere (www.egi.eu/services/
cloud-compute/)
[25] IAEA 2018 FENDL-3.1d: fusion evaluated nuclear data
library ver.3.1d (www-nds.iaea.org/fendl/)
[26] FischerU. 2017 EFDA D 2MM3A6 v1.0material
compositions for PPPT neutronics and activation analyses
(https://idm.euro-fusion.org/?uid=2MM3A6)
[27] FausserC. etal 2012 Tokamak D-T neutron source models for
diferent plasma physics connement modes Fusion Eng.
Des. 8778792
[28] FischerU. 2017 EFDA D 2L8TR9 v1.6PMI-3.3-T014-
D001guidelines for neutronic analyses (https://idm.euro-
fusion.org/?uid=2L8TR9)
[29] JordanovaJ. etal 2006 Parametric neutronic analysis of HCLL
blanket for DEMO fusion reactor utilizing vacuum vessel
ITER FDR design Fusion Eng. Des. 81221320
[30] BowmanC. 2018 Inference tools (https://doi.org/10.5281/
zenodo.1423552)
[31] CSimSoft 2018 Trelis version 16.5 (www.csimsoft.com/)
[32] RichardsonC. and WellsG.N. 2016 High performance
multi-physics simulations with FEniCS/DOLFIN 5 Report
eCSE03-10 (www.archer.ac.uk/community/eCSE/eCSE03-
10/eCSE03-10-TechnicalReport.pdf)
[33] AubertJ. etal 2017 Design description document (DDD)
2016 for HCLL DEN/DANS/DM2S/SEMT/BCCR/
RT/17-027/A
[34] AubertJ. etal 2016 Design description document (DDD) 2016
for HCLL CEA rapport technique DEN
[35] AubertJ. etal 2016 Design Engineering HCLL Design Report
2016 EUROfusion IDM EFDAD2N8YZS
[36] ZinkleS.J. and GhoniemN.M. 2000 Operating temperature
windows for fusion reactor structural materials Fusion Eng.
Des. 5125571
Nucl. Fusion 59 (2019) 0 46019
... Extensive work has been completed on the modelling of BBs for DEMO in the literature. Typically focused on neutronics [14,15,16], magneto-hydrodynamics (MHD) [17,18], mechanical fields [19] and tritium transport [20,21]. However, previous work that evaluated values of tritium inventories in the structure and permeation fluxes does not account for trapping, though trapping mechanisms significantly affect inventory values within tungsten (W) [22] and EUROFER [23,8]. ...
... Where r is the distance from the plasma facing wall expressed in [m]. The heat source term for W was taken from [14]. ...
Article
Full-text available
Tritium transport simulations in Water-Cooled Lithium Lead (WCLL) breeding blankets have been performed withFESTIM. Fluid dynamics and heat transfer were coupled to evaluate how temperature and LiPb velocity fields affecttritium transport. The 2D WCLL model consisted of tungsten, EUROFER and liquid LiPb. Tritium inventories andpermeation fluxes have been computed and the influence of trapping effects in solid domains was highlighted. It wasevaluated that inclusion of trapping mechanisms increases inventory values by 15% and delays permeation to the coolingchannels. A parametric study was conducted to investigate the impact of the uncertainty in the literature (2 orders ofmagnitude) regarding hydrogen solubility in LiPb on these quantities. Varying the lithium lead solubility over the rangefound in the literature was found to vary the EUROFER inventory by a factor of 25. Permeation fluxes to the coolantchannels was found to vary over a factor of 3.
... Following the conclusion of analysis, feedback can be provided and the design cycle can be iterated to refine and optimise the design. Automating the analysis can help to rapidly develop a design as shown in [1]. While some automated analysis remains a challenge in certain disciplines, progress is being made on the automation of a number of domains used in the analysis of fusion reactors. ...
... The use of these models in automated workflows has yet to be demonstrated in a publication but this would be the next logical stage in the process and the authors plan to publish a range of use cases for the parametric geometry in the future. Future work will, amongst other improvements, incorporate detailed parametric blanket models which have previously been created [1]. rotation_angle offset_from_plasma thickness plasma +ve +ve s t o p _ a n g l e s t a r t _ a n g l e a ll _ r a d ia l_ th ic k n e s s in n e r _ p la s m a _ g a p _ r a d ia l_ th ic k n e s s p la s m a _ r a d ia l_ th ic k n e s s o u te r _ p la s m a _ g a p _ r a d ia l_ th ic k n e s s fi r s tw a ll _ r a d ia l_ th ic k n e s s o u tb o a r d _ b la n k e t_ r a d ia l_ th ic k n e s s b la n k e t_ r e a r _ w a ll _ r a d ia l_ th ic k n e s s tf _ c o il _ to _ r e a r _ b la n k e t_ r a d ia l_ g a p o u tb o a r d _ b la n k e t_ r a d ia l_ th ic k n e s s p f_ c o il _ to _ tf _ c o il _ r a d ia l_ g a p p f_ c o il _ r a d ia l_ th ic k n e s s e s ...
Article
During the conceptual design process of fusion reactors it is useful to rapidly prototype different design concepts and assess their suitability against a range of high level requirements. Rapid prototyping allows the 'fail early' mantra of other fields to be applied to engineering design. Furthermore, the rapid generation of low fidelity analysis allows fast exploration of design space, which enables better decisions to be made during concept selection and the detailed design phase. The Paramak is an open-source tool that aims to provide automated parameter driven 3D CAD models for fusion reactor components and magnetic fusion reactors. The geometry produced is compatible with several analysis workflows and this allows iterative automated model building and analysis to help steer the design concept optimisation process. The Paramak uses CadQuery 2 to create the 3D CAD model. The Paramak framework is used to create a few example reactor configurations including: a spherical reactor, a regular large radius tokamak and a compact submersion tank reactor. Input parameters for the various reactors that the Paramak can generate generally fall into three categories: continuous ranges such as blanket thickness, integer ranges such as number of toroidal field coils and categorical parameters such as type of divertor. The Paramak facilitates parameter studies where users can investigate the impact of input design parameters on the reactor performance. The use of modern software practices allows the geometry to be continuously tested in analysis workflows to ensure it is fit for purpose. The generation of output metrics from input parameters lends itself to the use of data science and machine learning approaches in order to steer the design. The Paramak provides rapid construction of analysis ready CAD in a manner that allows the designer to save time when exploring the design space for design studies and facilitate automated generative design.
... There are presently a few classes of computational tools integrating the physics, technology and engineering: systems codes, a new class of engineering scoping tools and early-stage flight simulators. These are supplemented by other more localised tools, for example integrated plasma models such as JIN-TRAC (used in [54]), integrated turbulence models [95], parametric engineering models for optimising breeding blankets including neutronics [96] and materials models that translate the theory of radiation effects into their engineering impact, e.g. [97]. ...
Article
Full-text available
A fusion power plant can only exist with physics and technology acting in synchrony, over space (angstroms to tens of metres) and time (femtoseconds to decades). Recent experience with the European DEMO programme has shown how important it is to start integration early, yet go deep enough to uncover the integration impact, favourable and unfavourable, of the detailed physical and technological characteristics. There are some initially surprising interactions, for example, the fusion power density links the properties of materials in the components to the approaches to waste and remote maintenance in the context of a rigorous safety and environment regime. In this brief tour of a power plant based on a tokamak we outline the major interfaces between plasma physics and technology and engineering considering examples from the European DEMO (exhaust power handling, tritium management and plasma scenarios) with an eye on other concepts. We see how attempting integrated solutions can lead to discoveries and ways to ease interfaces despite the deep coupling of the many aspects of a tokamak plant. A power plant’s plasma, materials and components will be in new parameter spaces with new mechanisms and combinations; the design will therefore be based to a significant extent on sophisticated physics and engineering models making substantial extrapolations. There are however gaps in understanding as well as data – together these are termed “uncertainties”. Early integration in depth therefore represents a conceptual, intellectual and practical challenge, a challenge sharpened by the time pressure imposed by the global need for low carbon energy supplies such as fusion. There is an opportunity (and need) to use emerging transformational advances in computational algorithms and hardware to integrate and advance, despite the “uncertainties” and limited experimental data. We use examples to explore how an integrated approach has the potential to lead to consistent designs that could also be resilient to the residual uncertainties. The paper may stimulate some new thinking as fusion moves to the design of complete power plants alongside an evolving and maturing research programme.
... These components are often made of several materials joined together. This is the case for the monoblock geometry used for the ITER divertor which is made of a W armour, a CuCrZr cooling pipe jointed by a Cu inter-layer but also for DEMO breeding blankets since current designs involve W and EU-ROFER but Lithium, Beryllium, and other materials can also be found in some designs [8]. This study focuses on the interfaces between these materials, the physics at stake and their approximations. ...
Article
Full-text available
This work investigates the influence of hydrogen chemical potential continuity across solid material interfaces. The implementation of the mathematical model in FESTIM is verified using the Method of Exact Solutions (MES) and the Method of Manufactured Solutions (MMS) in 1D, 2D, with complex material properties and inhomogeneous temperature fields. A comparison test between FESTIM, TMAP7 and Abaqus codes is also performed and the codes show good agreement. The chemical potential continuity condition has an impact up to 40% on the outgassing particle flux on 4 mm composite slabs (W/Cu and Cu/EUROFER) compared to mobile concentration continuity. A method for rapid identification of materials properties from outgassing flux measurements is given. The influence of chemical potential conservation on monoblock inventory is then studied. It is shown that, for the 1D and 2D ITER divertor monobolocks cases, discrepancies only start to appear after approximately 5E6s of full power.
... Finally, a robust way of loosely coupling the global reactor design level with the system-specific design level and the associated high-fidelity, multi-physics modelling (such as for the breeding blankets -described in the previous section, and e.g. [111,125]) must be developed. ...
Article
In the framework of tokamak design studies, the magnet system dimensioning stands as a crucial point, magnets being strong drivers for reactor performances. In this aim superconducting magnets design path to manufacture and integration should undergo several stages of development, including the detailed analysis for defined operation cases. Those studies are linked to several physic domains (electromagnetics, mechanics, thermohydraulics, thermal) that all show interdependances between each other. In this objective, the platform OLYMPE was developed at CEA, aiming at addressing those complex multi-physic problems in an integrated simulation tool, expected to improve the efficiency of the design workflow involving macroscopic and detailed analyses. The different OLYMPE submodules will be described and their dedicated simulation domain explained, together with the logic that led to the choice of specific modules interconnection. In fact several combinations of those modules are built in coupled mode so to ensure converging design loops. Their major role is to assess the refinement of magnet design at detailed scale in the objective to increase magnet design accuracy. Namely illustrating examples will be given with following loops: - TF system design coupled with thermohydraulics: it ensures the compliance of TF design with temperature margin criterion. - TF system design coupled with electromagnetics: it ensures the compliance of TF design with topologically-induced magnetic field map. - Overall TF design coupled with thermohydraulics and cryogenics: it ensures the optimization of TF design and its operation conditions with overall system merits (e.g. cost factor). - TF system design coupled with mechanics: it ensures the compliance of TF design with local stress criterion. The outcomes and methodological merits of these loops will be illustrated by applications on DEMO configurations, and performances cross-checks or comparisons with simpler models will also be provided to quantify the accuracy enhancement. Furthermore the management of the overall input database and the GUI interface will be presented in the aim of improving tool operability. Finally discussions on this emerging OLYMPE tool qualities will also be exposed, including calculation performances merits and further development perspectives.
Article
Once the high-level requirements for a fusion power plant are set, the expected performance of plant systems, including the plasma, must be defined and then the available design space for the plant can be explored in order to converge on a final overall plant operating point, setting the major plant parameters such as number of toroidal field (TF) coils, tokamak major radius, plant power flows etc. Such design space explorations are conducted using systems codes which contain simplified models for plant systems, and attempt to capture the interactions between them in order to balance performance trade-offs and enforce overall consistency. This paper summarizes the work carried out to identify the EU-DEMO baseline operating point and the underpinning assumptions about technology and physics performance. The major design choices are described and the principle drivers for the direction of conceptual development, resulting in the operating space for EU-DEMO, are identified. The final output of the systems code forms the basis for more detailed engineering and physics evaluation and design work.
Article
The creation of 3-D CAD models is one of the key steps in the fusion reactor design cycle, and is very much a recurring process. A common division of labour sees engineers, draughtspeople, and analysts iterate ideas, requirements and constraints, 3-D CAD model(s), and analyses, respectively, to iteratively converge upon a working design. Whilst this traditional approach may work well during detailed design (e.g. for ITER), we contend that it is sub-optimal during the conceptual design, due to the large overheads associated with dealing with relatively mercurial geometries, load cases, and boundary conditions. In this work, we demonstrate that the automatic generation of 3-D CAD models for conceptual DEMO-class tokamak reactors is possible. The models, which comprise all major components ranging from the in-vessel components to the concrete radiation shield, are automatically produced for a given design point in approximately 30 s. This represents a significant acceleration of the traditional manual 3-D CAD model generation process for a typical reactor concept (∼1 month). The CAD data are generated within a reactor design framework and can be used for a range of purposes. We highlight the use case for neutronics, by demonstrating that a simplified 3-D 360° global neutronics model can be automatically built from the ground up and run, using faceted geometry and smeared material properties, all in under 1 h.
Article
Full-text available
The Fusion Science Demonstrator in the European Open Science Cloud for Research Pilot Project aimed to demonstrate that the fusion community can make use of distributed cloud resources. We developed a platform, Prominence, which enables users to transparently exploit idle cloud resources for running scientific workloads. In addition to standard HTC jobs, HPC jobs such as multi-node MPI are supported. All jobs are run in containers to ensure they will reliably run anywhere and are reproduceable. Cloud infrastructure is invisible to users, as all provisioning, including extensive failure handling, is completely automated. On-premises cloud resources can be utilised and at times of peak demand burst onto external clouds. In addition to the traditional “cloud-bursting” onto a single cloud, Prominence allows for bursting across many clouds in a hierarchical manner. Job requirements are taken into account, so jobs with special requirements, e.g. high memory or access to GPUs, are sent only to appropriate clouds. Here we describe Prominence, its architecture, the challenges of using many clouds opportunistically and report on our experiences with several fusion use cases.
Article
Full-text available
A parametric study is performed with the 2D FESTIM code for the ITER monoblock geometry. The influence of the monoblock surface temperature, the incident ion energy and particle flux on the monoblock hydrogen inventory is investigated. The simulated data is analysed with a Gaussian regression process and an inventory map as a function of ion energy and incident flux is given. Using this inventory map, the hydrogen inventory in the divertor is easily derived for any type of scenario. Here, the case of a detached ITER scenario with inputs from the SOLPS code is presented. For this scenario, the hydrogen inventory per monoblock is highly dependent of surface temperature and ranges from 1E18 to 6E19 H after a 1E7 s exposure. The inventory evolves as a power law of time and is lower at strike points where the surface temperature is high. Hydrogen inventory in the whole divertor after a 1E7 s exposure is estimated at approximately 8 g.
Article
Full-text available
In the framework of the European “HORIZON 2020” innovation and research programme, the EUROfusion Consortium develops a design of a fusion power demonstrator (DEMO). One of the key components in the fusion reactor is the Breeding Blanket (BB) surrounding the plasma, ensuring tritium self-sufficiency, heat removal for conversion into electricity, and neutron shielding. CEA-Saclay, with the support of Wigner-RCP and Centrum výzkumu Řež, is in charge of the development of one of the four BB concepts investigated in Europe for DEMO: the Helium Cooled Lithium Lead (HCLL) BB. The rationales of the HCLL are the use of Eurofer as structural material, eutectic liquid lithium-lead (PbLi) as tritium breeder and neutron multiplier, and helium gas as coolant. This paper shows the basic description of the DEMO HCLL BB concept and its design evolution during the past years, from a design based on the ITER Test Blanket Module (TBM) concept to a more advanced design called “Advanced-Plus” concept. This new HCLL BB concept that has been designed in order to improve Tritium Breeding Ratio (TBR) and shielding performances is presented. This new reference HCLL BB design has been analyzed and show very promising nuclear performances. Nevertheless, the “Optimized Conservative” concept, based on ITER TBM, is still considered as a robust back-up solution since structural improvements are still necessary on the “Advanced-Plus” concept. Moreover, a new Back Supporting Structure (BSS) is presented in this paper, designed to support the BB modules, with the aim to reduce pressure drops and thermal stresses.
Article
Full-text available
In the framework of the EUROfusion’s Power Plant Physics and Technology, the Working Package Breeding Blanket aims at investigating 4 different Breeding Blanket (BB) concepts for a EU demonstration fusion reactor (DEMO). One of these concepts is the Helium Cooled Pebble Bed (HCPB) BB, which is based on the use of pebble beds of lithiated ternary compounds and Be or beryllides as tritium breeder and multiplier materials respectively, EUROFER97 as structural steel and He as coolant. This paper aims at giving an overview of the EU HCPB BB R&D being developed at KIT, in collaboration with Wigner-RCP, BUTE-INT and CIEMAT. The paper gives an outline of the HCPB BB design evolution, state-of-the-art, basic functionalities, requirements and performances, and the associated R&D activities in the areas of design, functional materials, manufacturing and testing. Additionally, attention is given also to the activities dedicated to the development of heat transfer augmentation techniques for the FW and the corresponding testing. Due to their nature as design drivers, a brief overview in the R&D of key HCPB interfacing areas is given as well, namely the Tritium Extraction and Recovery system, the Primary Heat Transfer & Power Conversion Systems and Safety topics, as well as some specific activities regarding the integration of in-vessel systems through the BB. As concluding remarks, an outline of the standing challenges and future R&D plans are summarized. https://doi.org/10.1109/TPS.2018.2830813
Article
Full-text available
This paper presents the nuclear analysis of the European DEMO baseline 2015 with HCLL blanket carried out with the TRIPOLI-4® Monte Carlo code and the JEFF-3.2 nuclear data library. The TRIPOLI-4® model was imported from CAD using the McCad tool. A procedure that generates the detailed 3D model describing all the HCLL blanket internal structures was developed. This procedure allows parametrization of the blanket internal structures such as the number of cooling plates, manifolds, etc. and the thickness of the stiffening grid for instance. Different design variants were studied to improve the tritium production. From this previous study a complete nuclear analysis was carried out on a promising design which is a compromise between tritium production and mechanical robustness. All criteria (TBR, nuclear heating in coils and displacement damage in vacuum vessel) are met using this new reference design.
Article
Full-text available
The Helium Cooled Pebble Bed (HCPB) breeder blanket is being developed as part of the European Fusion Programme. Part of the programme is to investigate blanket designs relevant for future demonstration fusion power plants. This paper presents neutronics analyses of the HCPB with an alternative neutron multiplier, Be12Ti which is incorporated into the design, replacing the current Be multiplier. A parameter study was performed for a range of geometries to identify the optimal heights of the lithium ceramic and neutron multiplier pebble beds. Automated creation of CAD models followed by conversion to constructive solid geometry (CSG) and unstructured mesh (UM) geometry allows the models to be useful for both neutronics simulations and engineering simulations. In this neutronics study simulations were performed using MCNP 6.1 to find the tritium breeding ratio, energy multiplication and the volumetric heat loads of different blanket designs. Combinations of geometry parameters and material choices that resulted in adequate TBR values were identified and will be further investigated with automated engineering simulations. This paper provides insight, supported by neutronics analysis, on the validity of the design and comments on some of the potential advantages and disadvantages of using Be12Ti in the Helium Cooled Pebble Bed (HCPB) breeder blanket. Blankets with Be12Ti neutron multiplier were found to produce less tritium but higher energy multiplication when compared to blankets with Be neutron multiplier.
Article
Full-text available
The Helium Cooled Pebble Bed (HCPB) Breeding Blanket (BB) is one of the 4 BB concepts being investigated in the EU for their possible implementation in DEMO. During 2014 the former “beer-box” BB concept based on the ITER’s HCPB Test Blanket Module suffered several design changes so as to meet the different counteracting nuclear, thermohydraulic and thermomechanical requirements. These studies evidenced that the concept is too rigid to meet the tight TBR requirements imposed for the EU DEMO (i.e. TBR ≥ 1.10). Additionally, the complex manifold system with unbalanced helium mass flow in each of the 2 parallel cooling loops made the concept thermohydraulically complex. However, parametric studies during 2015 revealed that the HCPB concept have potential for a better nuclear performance, as well as margin for a significant simplification of the cooling internals by redefining the cooling plates and the architecture of the blanket, building a symmetric flow scheme. This paper describes the new HCPB concept based on an integrated FW with the breeding zone thermohydraulics and helium manifold systems. The former complex manifold backplates have been compacted and integrated in the cooling plates, releasing ≈300 mm of radial space that can be used now to increase breeder zone, the neutron shielding, to reinforce the Back Supporting Structure (BSS) or basically to reduce the reactor size. Detailed neutronic analyses have yielded a TBR of ∼1.20 for the baseline design. Initial analyses show a correct thermohydraulic behavior. Preliminary thermomechanical analyses also indicate that the design can potentially withstand an in-box LOCA at 9 MPa at a level C according to the RCC-MRx code. Future consolidation activities are described, which shall lead to a concept meeting the BB requirements.
Article
Water-cooled lithium-lead breeding blanket is considered a candidate option for European DEMO nuclear fusion reactor. ENEA and the linked third parties have proposed and are developing a multi-module blanket segment concept based on DEMO 2015 specifications. The layout of the module is based on horizontal (i.e. radial-toroidal) water-cooling tubes in the breeding zone, and on lithium lead flowing in radial-poloidal direction. This design choice is driven by the rationale to have a modular design, where a basic geometry is repeated along the poloidal direction. The modules are connected with a back supporting structure, designed to withstand thermal and mechanical loads due to normal operation and selected postulated accidents. Water and lithium lead manifolds are designed and integrated with a consistent primary heat transport system, based on a reliable pressurized water reactor operating experience, and the lithium lead system. Rationale and features of current status of water-cooled lithium-lead breeding blanket design are discussed and supported by thermo-mechanics, thermo-hydraulics and neutronics analyses. Open issues and areas of research and development needs are finally pointed out.
Article
McCad is a geometry conversion tool developed at the Karlsruhe Institute of Technology (KIT) for the automatic conversion of CAD models into the semi-algebraic geometry representations which can then be used in Monte Carlo (MC) particle transport simulations applied in design analyses of fusion and fission reactors. New algorithms for the decomposition of complex geometry models have been designed and implemented in the McCad code. With these improvements, a weakness of the original McCad was mitigated which is due to the instability of the applied graphic kernel and the original decomposition algorithm. The improvements have been verified with some representative test models and a generic model of a DEMO fusion power reactor. The results show that the advanced McCad version with the new decomposition algorithms is more robust and provides more accurate and less complex conversion results. The decomposition process is efficient and produces correct results which are consistent with the original model.
Article
This work gives an overview of the neutronic analyses to support and optimize the advanced HCPB blanket concept for DEMO. Full scale 3D Monte Carlo particle transport simulations were performed to this end with the MCNP5 code employing a very detailed HCPB DEMO torus sector model. The HCPB blanket was optimized to fulfil the design requirements on tritium breeding, thermal stress stability and shielding performance.
Article
Embedded CSE (eCSE) report on block preconditioning and parallel discontinuous Galerkin support in FEniCS (http://fenicsproject.org).
Article
Neutronic analyses or, more widely, nuclear analyses have been performed for the development of a dual-coolant He/LiPb (DCLL) conceptual design reactor. A detailed three-dimensional (3D) model has been examined and optimized. The design is based on the plasma parameters and functional materials of the power plant conceptual studies (PPCS) model C. The initial radial-build for the detailed model has been determined according to the dimensions established in a previous work on an equivalent simplified homogenized reactor model. For optimization purposes, the initial specifications established over the simplified model have been refined on the detailed 3D design, modifying material and dimension of breeding blanket, shield and vacuum vessel in order to fulfil the priority requirements of a fusion reactor in terms of the fundamental neutronic responses. Tritium breeding ratio, energy multiplication factor, radiation limits in the TF coils, helium production and displacements per atom (dpa) have been calculated in order to demonstrate the functionality and viability of the reactor design in guaranteeing tritium self-sufficiency, power efficiency, plasma confinement, and re-weldability and structural integrity of the components. The paper describes the neutronic design improvements of the DCLL reactor, obtaining results for both DEMO and power plant operational scenarios.