Folie 1

1
BEST ESTIMATE TRANSIENT ANALYSES BY THE COUPLED
SYSTEM CODE ATHLETQUABOX/CUBBOX FOR KURSK 1
CORE STATES AND COMPARISON TO RELAP
CALCULATIONS M. Clemente Gesellschaft für
Anlagen- und Reaktorsicherheit (GRSmbH) Germany
INTERNATIONAL CONFERENCE ON PRESSURE TUBE
REACTORS PROBLEMS AND SOLUTIONS N.A. Dollezhal
Research and Development Institute of Power
Engineering (NIKIET) Moscow, October 19. 20.
2004
2
The work performed was a contribution to the IAEA
Extrabudgetary Project Accident Analysis and
its Associated Training Programme for the RBMK
1000 Kursk NPP (Phase I and II) and supported
by Langenbuch, S., Kusnetzov P.B., Sakharova
T.Yu., Fyodorov V.L. , Savvatimskiy G.
Gesellschaft für Anlagen- und
Reaktorsicherheit (GRS)mbH, Germany Research
and Development Institute of Power Engineering,
RDIPE, Russia Russian Research Centre
Kurchatov Institute, RRC KI, Russia
3

• Contents
• Model Description
• Analysis of an ATWS Transient Loss of Feedwater
  for a
• Kursk 1 Core State before Reconstruction
• Analyses of Single Rod Withdrawal Results for the
  Reference Core State of Kursk 1 after
  Reconstruction
• Conclusions

4
Model Description I
3D-core model QUABOX/CUBBOX Solution of 2-group
diffusion equation coarse mesh method,
polynomial approximation Applied to study LWR
cores. Model is adapted to RBMK conditions
Thermal-hydraulic system code ATHLET Modular
structure with models for fluid dynamics, heat
transfer and heat conduction, neutron kinetics
and control systems. Applied to study
anticipated and abnormal plant transients, small
and intermediate leaks as well as large breaks
5
Model Description II

• Coupled Model ATHLET QUABOX/CUBBOX
• Both codes keep their independent capabilities
  and are exchanging data via an appropriate
  interface.
• Result of neutronic calculation
• 3D power density distribution
• transferred to ATHLET
• Result of thermal-hydraulic calculation
• Fuel-, graphite temperatures and coolant density
• transferred to QUABOX/CUBBOX
• Different time step sizes in both codes, which
  are synchronized appropriately

6
Model Description III
Coupled Model ATHLET QUABOX/CUBBOX Description
of reactor core In QUABOX/CUBBOX Each channel
of the core is modeled separately In ATHLET the
core is described by a reduced number of
thermal-hydraulic channels, describing core
sub-regions of different size, depending on the
problem to be solved. Loss of Feedwater 16
core channels Single rod withdrawal 77 79
core channels (to describe the local power change
near the withdrawn rod)
7
Example of Definition of core sub-regions for the
representation by 16 channels in ATHLET
8
Central Rod Withdrawal with Definition of Local
Channels in ATHLET around the withdrawn rod
9
Kursk 1 Nodalisation Scheme in ATHLET
10
Loss of Feedwater, ATWS
For the testing of the coupled model a Loss of
Feedwater transient assuming ATWS conditions was
chosen, which puts special demands on the code
capabilities. The transient is determined by a
strong feedback behaviour Basis of
calculation Kursk 1 core state of March 1998
before reconstruction
11
Mass Flow Rate at Core Inlet for Channel K-09 and
K-10
12
Void Fraction at Core Inlet and Outlet for
Different ATHLET Channels
13
Axial Fuel Temperatures for Lower Half of Core
14
Axial Fuel Temperatures for Upper Half of Core
15
ATWS Loss of Feedwater Kursk 1 Comparison of
point kinetics and 3D kinetics solution
16
Kursk 1 Reactivity behaviour as function of
coolant density Mean feedback parameters assumed
17
Total Power for an ATWS event with Loss of
Feedwater at INPP2
Comparison of 3D kinetic and point kinetic
results (no LAC activation) and a 3D kinetic
result with active LAC (L.Kuriene)
18
Loss of Feedwater ATHLET and RELAP5/mod.3.2
Results Point kinetics model
19
Application of the Coupled Code ATHLET
QUABOX/CUBBOX for Control Rod Withdrawal
Analyses Kursk 1 Reconstructed Reference Core
State

• Withdrawal of single rods
• Central rod 4235 without local automatic power
  control
• Central rod 4235 with local automatic power
  control
• Peripheral rod 1641 without local automatic
  power control
• Peripheral rod 1641 with active CPS. First 2
  Detector signals
• neglected

20
Local Power in channels around the withdrawn
central rod 4235
without local power control
with local power control
21
Local Power in channels around the withdrawn
peripheral rod 1642
without local power control
with shut down, the first

2 signals are neglected
22
Cladding Temperatures, peripheral rod withdrawal
without CPS
23
Total Power Comparison with RELAP3D
24
Conclusions Loss of Feedwater Analyses

• Replacing the point kinetic model by coupling
  the 3D core model QUABOX/CUBBOX with the system
  code ATHLET an improved tool for analyses of
  thermal-hydraulic and reactivity initiated
  transients is provided.
• Sudden reactivity changes are less pronounced
  and the void reactivity feedback is less positive
  in comparison to the point kinetic results.
• The comparison of ATHLET and RELAP5/mod3.2
  results (point kinetics) shows a good agreement,
  if the same boundary conditions are used.
• The ATWS event Loss of Feedwater is a beyond
  design accident for RBMK and limiting
  temperatures (cladding, pressure tube) can be
  exceeded. This shows the need of a 2nd diverse
  shut down system, which is now available after
  the reconstruction of Kursk 1.

25
Conclusions Control Rod Withdrawal Analyses

• The withdrawal of the peripheral rod shows a
  higher power increase than the withdrawal of the
  central rod.
• During the withdrawal of the peripheral rod
  signals for reactor shut down are generated
• No limiting cladding temperatures are exceeded
  for all cases investigated.
• With activated automatic local power control
  (AC-I rods) the total power is constant. Locally
  a maximum power increase to 3.8 MW is observed
  (central rod).
• The complex logic of the CPS (detector signals,
  rod movements) can be modeled explicitly for
  thermal-hydraulic transient analyses only by the
  coupled code system. This provides an essential
  improvement for the simulation of transients,
  where reactivity changes and thermal-hydraulic
  feedbacks are linked.