Analysis and Solution of Bulk Power Oscillations in 540 MWe PHWR TAPS4

1

• Analysis and Solution of Bulk Power Oscillations
  in 540 MWe PHWR TAPS-4

May 2007
Presented at IAEA by BB Biswas, Head, Reactor
Control Division, Bhabha Atomic Research Centre,
Mumbai, INDIA
2
Introduction to 540 MWe PHWR

• India has two 540 MWe PHWR units TAPS 3 4, at
  Tarapur near Mumbai.
• Neutronically, 540 MWe PHWR is a large reactor.
• It is characterized by loose neutronic coupling
  among different regions of the core, i.e.,
• The total power output of the reactor may be
  constant
• But, the core power distribution may be
  nonuniform.

3
Introduction to 540 MWe PHWR

• The Reactor Regulating System (RRS) has been
  designed to control
• Power Distribution/ Flux Tilt
• Bulk Power
• Normally, by means of the following Reactivity
  Devices
• Control rods Coarse control
• Adjuster rods Coarse Control
• Liquid Zone Control System fine Control

• The Control Loop of RRS that maintains the Bulk
  Power close to the demand is called the Bulk
  Power Control Loop (BPCL).
• And, the Control Loop of RRS that maintains the
  Core Power Distribution close to the desired
  distribution is called the Zonal Power Control
  Loop (ZPCL).

4
Reactor Regulating System

• RRS of 540 MWe PHWR is required to handle huge
  number of inputs.
• So, multi-nodal architecture was selected.
• 3 Input nodes, 2 Main Processor nodes, 2 Output
  nodes catered by a high speed data highway.
• Each node functions independently of other nodes,
  so as to yield a overall response time of 300 ms
  (min.) to 900 ms (max.)

5
Bulk Power Oscillations Problem

• It is known that in Large PHWRs, the core power
  distribution may exhibit slow oscillations,
  caused by non-uniform xenon production and decay
  consumption rates.
• But, in TAPS 4, a strange problem of bulk power
  oscillations was reported in April 2006.
• This bulk power oscillation problem is the theme
  of this talk
• Site observation,
• Root Cause Analysis
• Solution.

6
Site Report of Bulk Power Oscillations Problem

• TAPP 4 unit operated satisfactorily during the
  initial stages. However, as the operation without
  refueling continued, it started experiencing bulk
  power oscillations.
• Frequency of Oscillations 3 6 cycles per
  minute (much higher than that for xenon induced
  oscillations).

7
Early Observations

• The initial core of the 540 MWe PHWR TAPP4 unit
  consisted of

• Natual Uranium Oxide Bundles, as well as
• Depleted Uranium Bundles, and
• Online refueling halted for some days due to
  fueling machine problems.

• Normally,
• The core loading should be maintained near a
  equilibrium configuration by refueling.

• Hence, it was felt that the oscillations could
  have been caused due to instability of the Bulk
  Power Control Loop.

8
Stability Analysis of BPCL

• Mathematical Modeling of BPCL
• Representation into Standard State Space form
• Stabilty assessment based on Eigenvalues of
  Closed Loop System

9
Model Development Differential Equations

• Liquid Zone Control System (LZCS)
• Water Inflow CV
• 2nd order model identified from frequency
  response test data collected at suppliers works.
• Damping coefficient0.49, undamped natural freq
  14.6 rad/s
• ZCC Water Level
• 1st order model
• Reactor
• 7th order model for Core Neutronics
• 2nd order model for fuel coolant temperature
  variations
• 1st order model for RTD for coolant temperature
  measurement
• Xe-I dynamics and Moderator Temperature
  Variations Ignored
• Neutronic Instrumentation
• 2nd order model for Log Amplifier
• 1st order model for SPND Amplifier
• 2nd order model for Rate Amplifier

10
Model Development Difference Equations

• Instrumented Coolant Channel Monitoring System
  (ICMS)
• 1st order model.

• RRS
• 5 First order filters in IPN
• 2nd order model for PMCR for log. power
• 2nd order model for PMCR for linear power
• 1st order model for OPN

11
State Space Model

• Plant
• dx1/dt A11 x1A12 x2
• x1n1 ?11 x1n ?12 x2n
• x2n1 ?21 x1n ?22 x2n ?2 un
• x1 is 18 dimensional vector x2 is 11 dimensional
  vector.
• Output
• Yn C xn
• ? ?11 ?12 ? 0
• ?21 ?22 ?2

12
Overall System
Bulk Power Control

• Control
• un G1 ? Effective Power Error
• G2 ? (1-L1) ? Zonal Power Error
• G3 ? L1 ? Level Error
• G4 ? Level Error
• (L1 0 above 0.25 FFP)
• represented as
• un H xn
• Closed Loop System
• xn1 (? ?HC) xn
• Stability Criterion
• ? (? ?HC) lt 1
• i.e. Roots of Characteristic Equation are within
  the unit circle.

Zonal Power Control
ZCC Water Level Control
13
Stability of Closed Loop System

• Stability Analysis was carried out at different
  power levels in the range 10-6 FP to 1 FP because
• Time constant of Log Amp Varies with Power
• Internal Reactivity feedback due to fuel
  coolant vary with power
• Stability analysis was also to be carried out for
  initial and other different stages of core,
  because
• Reactivity Feedback due to coolant is negative
  for initial core while positive for equilibrium
  core.
• Reactivity Feedback due to fuel is large negative
  for initial core and small negative for
  equilibrium core.
• Delayed neutron fraction for initial core is
  larger than that for equilibrium core.
• Stability analysis was also to be carried out for
  different Cycle Time of RRS, as it could vary in
  the range 300 ms to 900 ms.

14
Control Gains in RRS

• Values of Control Gains (when oscillations were
  reported)
• Bulk Power Control Gain, G1 12.5
• Level Control Gain, G3 0.05G1
• Permanent Level Control Gain, G4 0.025G1

15
Stability Analysis Results

• Largest Eigenvalue for Initial Core

Initial Core Control Loop might become unstable
for operation in the range 20 to 80 FP if RRS
cycle time approached 900 ms.
16
Stability Analysis Results

• Largest Eigenvalue after 35 FPD operation

After 35 FPD Operation Control Loop might become
unstable for operation above 15 FP if RRS cycle
time approached 750 ms.
17
Stability Analysis Results

• Largest Eigenvalue for the core when oscillations
  were observed

Control Loop might become unstable for operation
above 5 FP if RRS cycle time approached 750 ms
and above 25 FP if RRS Cycle time approached 450
ms.
18
Stability Analysis Results

• Largest Eigenvalue for Equilibrium Core

For Equilibrium Core Control Loop might become
unstable for operation above 5 FP if RRS cycle
time approached 600 ms.
19
Stability in (G1, GH) Plane Initial Core at 25
FP
Unstable Region
Stable Region
Initial Core Control Loop might become unstable
for operation at 25 FP if RRS cycle time
approached 900 ms.
20
Stability in (G1, GH) Plane 35 FPD Core at 25 FP
35 FPD Core Control Loop might become unstable
for operation at 25 FP if RRS cycle time
approached 750 ms.
21
Stability in (G1, GH) Plane Equilibrium Core at
25 FP
Equilibrium Core Control Loop might become
unstable for operation at 25 FP if RRS cycle
time approached 600 ms.
22
Summary of Stability Analysis Results

• Initial core had relatively better degree of
  stability than the equilibrium core
• Poor degree of stability at higher power levels
• Poor stability characteristics for larger values
  of RRS cycle time
• Imaginary part of eigenvalues 0.360.46 rad/s
• ?frequency of oscillation 3.54.5 cycles per
  minute.

23
Solution to the Bulk Power Oscillation Problem

• Controller Redesign
• The Controller equations should not be changed,
  So that
• Coding is not required afresh
• Reverification validation is not required
• Assessment of Transient Response from Simulation
  of Non-linear System
• Site Tests
• Final Implementation

24
Damping of the closed loop system at 25 FP
With GH 0
For G1 5 The system has good damping for wide
range of variation in RRS Cycle time.
Though the damping reduces for GH gt 0, the system
can still accommodate large range of variation in
RRS Cycle time.
With G1 5.0
25
Suggested Solution

• New Control Gains
• G1 5.0
• G3 0.2?G1 1.0
• G4 0.05?G1 0.25
• RRS Cycle time should be reduced to obtain good
  degree of damping, at all power levels

26
Transient Response SimulationDemand Change 25
to 30 FP Ts 0.45 s
27
Transient Response SimulationDemand Change 25
to 30 FP Ts 0.6 s
28
Transient Response SimulationEffect of Change in
Controller Gains
Variation of Reactor Neutronic Power and EPE when
the value of G1 is changed
29
Implementation at site

• To obtain RRS cycle time lt 600 ms
• Faster CPU boards put in Laboratory System for
  test and VV.
• Software in IPN OPN nodes modified to reduce
  overheads and render reduced overall cycle time.
• Simulated behaviour for the New Control Gains was
  thoroughly reviewed by Regulatory Board.
• Site Experiments conducted within limited scope
  to validate the results of analysis.
• Gain Changes incorporated in a phased manner
  while taking fast record of important variables.

30
Conclusion

• The Bulk Power Oscillations Problem
• Root Cause of the Probmem identified solved in
  a systematic manner using Control Theory
  Approach.
• The system has now good transient response and
  stability Characteristics.

However, there is a need to arrive at a design of
RRS so as to address all the possible variations
in the reactivity characteristics of the core
that may possibly occur in varied OM scenario.
Also, the design should be robust for unintended
cycle time variations, that could happen in
multi-nodal systems.