By
KENGYEN CHIANG
June 2012
Benoit Forget
Assistant Profe of Dep e cience and Engineering
Thesis CoSupervisor
Tom Newton
Associate Director of MIT Nuclear Reactor Laboratory
Thesis Reader
Accepted by: 
Mujid S. Kazimi
TEPCO P ofess of Nuclear Engineering
Chair, Department C ittee on Graduate Students
Thermal Hydraulic Limits Analysis for the MIT Research Reactor Low
Enrichment Uranium Core Conversion Using Statistical Propagation of
Parametric Uncertainties
by
KENGYEN CHIANG
Abstract
The MIT Research Reactor (MITR) is evaluating the conversion from highly enriched
uranium (HEU) to low enrichment uranium (LEU) fuel. In addition to the fuel
element redesign from 15 to 18 plates per element, a reactor power upgraded from 6
MW to 7 MW is proposed in order to maintain the same reactor performance of the
HEU core. Previous approaches in analyzing the impact of engineering uncertainties
on thermal hydraulic limits via the use of engineering hot channel factors (EHCFs)
were unable to explicitly quantify the uncertainty and confidence level in reactor
parameters. The objective of this study is to develop a methodology for MITR
thermal hydraulic limits analysis by statistically combining engineering uncertainties
in order to eliminate unnecessary conservatism inherent in traditional analyses. This
methodology was employed to analyze the Limiting Safety System Settings (LSSS)
for the MITR LEU core, based on the criterion of onset of nucleate boiling (ONB).
Key parameters, such as coolant channel tolerances and heat transfer coefficients,
were considered as normal distributions using Oracle Crystal Ball for the LSSS
evaluation. The LSSS power is determined with 99.7% confidence level. The LSSS
power calculated using this new methodology is 9.1 MW, based on core outlet coolant
temperature of 60 'C, and primary coolant flow rate of 1800 gpm, compared to 8.3
MW obtained from the analytical method using the EHCFs with same operating
conditions. The same methodology was also used to calculate the safety limit (SL) to
ensure that adequate safety margin exists between LSSS and SL. The criterion used
to calculate SL is the onset of flow instability. The calculated SL is 10.6 MW, which
is 1.5 MW higher than LSSS, permitting sufficient margin between LSSS and SL.
2
Acknowledgements
First and foremost I would like to express my deepest gratitude to my supervisor, Dr.
Linwen Hu, who has supported and guided me throughout my thesis with her
patience and knowledge, as well as offering me fantastic opportunity to work with
senior researchers in Argonne National Laboratory. I am also indebted to the many
previous and current contributors to the MITR LEU project for providing the
preliminary analyses, system measurement data and relevant document, guiding me
on the right track all over my research. Dr. Sung Joong Kim, Dr. Erik Wilson, Dr.
Floyd Dunn, Dr. Thomas Newton, and Prof. Benoit Forget have fascinated me with
not only their broad knowledge in nuclear science and engineering, but also their
sense of humor and blissful smiles. In many ways I have learnt how to make life
easier and happier in MIT; thanks to their encouragement.
Thanks also go to the MIT Chapel and Zesiger Sports&Fitness Center, the places
where I always get a warm feeling, making me smile. I experienced peace and
serenity of mind whenever having silence meditation with the Venerable Tenzin
Priyadarshi in the Chapel. The feeling of happiness also comes to me every time
when running and laughing on the basketball court with friends. Special thanks to the
staff (a cool black guy) at ZCenter: I appreciate your friendly smile and a "Yo,
what's up brother" every time we met. Every of these brightened my life at MIT,
thank you.
Finally, thanks to everything, either making me laugh or mad, that I have in MIT. It's
an unforgettable journey that I would cherish for the rest of my life.
3
Table of Contents
Chapter 1 Introduction
1.1 The Reduced Enrichment for Research and Test Reactors Program.................11
1.2 Core Conversion Safety Analyses..................................................................... 13
1.3 Description of the MIT Reactor........................................................................ 14
1.4 The Proposed Low Enrichment Uranium Fuel Design of the MITR................21
1.5 Thesis Objectives.............................................................................................. 24
2.1 Introduction..................................................................................................... 27
2.2 Historical Review on Engineering Uncertainty Treatment.............................. 29
2.3 Introduction to Engineering Hot Channel Factors (EHCFs)............................. 30
2.3.1 EHCFs....................................................................................................... 30
2.3.2 Common subfactors involved..................................................................... 32
2.4 Engineering Hot channel Factors (EHCFs) Used in MITRII......................... 35
4
3.7.1 Best estimate ONB using hot stripe technique............................................ 73
3.7.2 H ot stripe LSSS using EHCFs..................................................................... 75
3.8 Sum m ary..............................................................................................................77
4.1 Introduction..................................................................................................... 82
4.2 M onte Carlo Simulation................................................................................... 84
4.2.1 Oracle Crystal Ball....................................................................................... 84
4.2.2 Validation of Sim ple M odel on Crystal Ball................................................ 86
4.3 U ncertainty of input param eters for the M ITR................................................. 88
4.3.1 Prim ary Coolant Flow Rate....................................................................... 88
4.3.2 Heat Transfer Coefficient............................................................................ 91
4.3.3 Hot Channel M ass Flow Rate....................................................................... 93
4.3.4 Power...............................................................................................................97
4.4 Results...................................................................................................................99
4.5 Summ ary..............................................................................................................100
6.1 Introduction.........................................................................................................120
6.2 Onset of Flow Instability....................................................................................123
6.2.1 Introduction....................................................................................................123
6.2.2 Calculation of OFI for the MITR (Analytical Approach)..............................125
6.2.3 Calculation of OF for the MITR (Uncertainty Propagation Technique).......128
6.3 Critical Heat Flux................................................................................................130
6.3.1 Introduction....................................................................................................130
6.3.2 CHF Correlations (Including Natural Convection and Forced Convection).. 133
5
6.3.3 Calculation of CHF for the M1TR (Analytical Approach).............................135
6.3.4 Calculation of CHF for the MITR (Uncertainty Propagation Technique).....137
6.4 Comparison between OFI and CHF...................................................................140
6.5 Summary.............................................................................................................142
7.1 Introduction........................................................................................................144
7.2 RELAP5/M od3.3................................................................................................145
7.3 RELAP5 Input Deck for the M ITR....................................................................146
7.4 Natural Convection LSSS Calculation...............................................................149
7.5 Summary............................................................................................................153
8.1 Conclusions.........................................................................................................156
8.2 Recommendations for This Study......................................................................157
Appendix
Appendix A RELAP5 Input File for Natural Circulation LSSS of MITR (Steady
State).....................................................................................................159
Appendix B RELAP5 Input File for Natural Circulation LSSS of MITR (Restart
File)......................................................................................................169
6
List of Figures
7
Figure 65 Comparison between Sudo CHF prediction and experiment result.........134
Figure 71 RELAP5 Nodalization of M ITR..............................................................148
Figure 72 Decay power changes with time (initial power is set as 1MW)...............151
8
List of Tables
Table 11 Research Reactors that were converted or shut down since May 2004.......12
Table 12 HEU and LEU fuel plate and fullchannel (interior channel) dimensions...22
Table 13 HEU and LEU Fuel plate and sidechannel (outside channel) dimensions.23
Table 14 Composition and thermophysical properties of HEU and LEU fuel..........23
Table 21 Subfactors and EHCFs used in MITRII SAR.......................................37
Table 22 Description for subfactors used in MITRII SAR..................................38
Table 31 Parameters used in LSSS calculation..................................................... 43
Table 32 Parameters used for analytical LSSS calculation.....................................51
Table 33 Summary for the MITR pressure loss calculation...................................55
Table 34 Pressure calculated for each node of the fueled region............................55
Table 35 Axial power distribution for the MITR LEU fuel...................................57
Table 36 Hot channel coolant temperature calculated for at each node.................58
Table 37 Derived Geometry Parameters for MITR.............................................. 60
Table 38 The LEU geometry parameters in Carnavos correlation and their
counterpart in M1TR................................................................................. 62
Table 39 HTC calculated using DB Correlation and Carnavos Correlation..........64
Table 310 T/H conditions used in hot stripe LSSS calculation for each node............72
Table 311 Summary for LSSS powers calculated using different approaches...........78
Table 41 The input description for the model used to validate Oracle Crystal Ball...87
Table 42 Comparison between analytical solution and results using Crystal Ball.....87
Table 43 Input parametric distributions used in uncertainty propagation
methodology...............................................................................................101
Table 44 Summary for LSSS power obtained using different methodology............103
Table 51 Changes in LSSS due to the change in flow disparity factor....................106
Table 52 The resulting change in LSSS power when outlet temperature is fixed at
60 0 C............................................................................................................ 10 8
Table 53 The change in LSSS power with respect to the change in HTC when outlet
temperature is fixed at 60 0C........................................................................110
Table 54 Summary for correlations typically used to compute HTC in research
Reactors......................................................................................................112
Table 55 HTCs and outlet clad temperatures computed using different
correlations.................................................................................................114
Table 56 The change in LSSS power with respect to the change in local temperature
(viscosity) when outlet temperature is fixed at 60*C..................................117
Table 61 Parameters used to calculate OFI...............................................................127
9
Table 62 Parameters set in a form of distribution for OFI/CHF calculation.............129
Table 63 Parameters used in CHF calculation for the MITR....................................136
Table 64 Parameters that were set in a form of distribution for CHF calculation.... 139
Table 65 Comparison between OF and CHF...........................................................141
Table 71 Cladding temperature and temperature inducing ONB at each node (both
the NCVs and ASVs are open)...................................................................151
Table 72 Cladding temperature and temperature inducing ONB at each node (Only
N CV s are open)..........................................................................................152
Table 73 Calculated Coolant Temperature Rise and Film Temperature Rise for
Natural Convection Operation....................................................................154
10
Chapter 1 Introduction
The MIT Research Reactor (MITR) is evaluating the conversion from highly enriched
uranium (HEU) to low enrichment uranium (LEU) fuel. In addition to the fuel
element redesign, a reactor power upgraded from 6 MW to 7 MW is proposed in
order to maintain the same reactor performance of the HEU core.
1.1 The Reduced Enrichment for Research and Test Reactors Program
As introduced in the fact sheet from the National Nuclear Security Administration
(NNSA) [1], "The National Nuclear Security Administration established the Global
Threat Reduction Initiative (GTRI) in the Office of Defense Nuclear Nonproliferation
to, as quickly as possible, identify, secure, remove and/or facilitate the disposition of
high risk vulnerable nuclear and radiological materials around the world that pose a
threat to the United States and the international community."
The Reduced Enrichment for Research and Test Reactors Program (RERTR) was
placed under NNSA in 2004 as part of GTRI [2, 3] with an aim to better coordinate
several nonproliferation programs jointly under GTRI. To reduce the potential global
nuclear threat, these institutes are converting the existing high enriched uranium
(HEU) fuels with low enrichment uranium fuels (LEU, U235 enrichment < 19.99%)
in research facilities worldwide, as well as detecting, securing, safeguarding,
disposing, and controlling nuclear materials.
GTRI has made significant contribution in nuclear material threat reduction in recent
years. Twentytwo research reactors have been converted to LEU, including the
HIFAR in Australia in 2004 to the Kyoto University research reactor in Japan in 2010.
Moreover, twelve HEU research reactors were shut down without converting,
including the ZPPR reactor's decommission at Idaho National Laboratories started in
September 2008. These research reactors are summarized in Table 11, including the
period they were converted or shut down [1]. Note that these converted research
reactors used dispersion fuel, the fuel design allowing the fuel material to be broken
up into very small pieces that are dispersed into and encapsulated by a matrix material,
making the matrix material stable under irradiation. The dispersion fuel design is
sufficient for these conversions while high performance reactors require high density
fuel. An alternate fuel design, monolithic fuel was selected for the MITR conversion
project, which is discussed in section 1.4.
11
Table 11 Research Reactors that were converted or shut down since May 2004 [1]
12
1.2 Core Conversion Safety Analyses
In general, the preferred approach for converting from HEU to LEU fuel is direct
conversion without making modifications in fuel element dimensions or core
configurations, therefore minimizing requirements of altering the safetyrelated
parameters of the facility. Since the volume of HEU cores and fuel elements remain
the same, the way to achieve the same number of uranium235 atoms as in the HEU
cores is to increase fuel density. As a matter of fact, additional U235 is actually
required to offset the resonance absorption in U238.
While doing core conversion, in the beginning what have to be verified are the
feasibility of fabrication process and the LEU fuels performance under irradiation,
such as the capability to accommodate released fission gases. Moreover, a series of
conversion safety analyses have to be performed. These analyses [4] normally
include neutronics analyses, steadystate thermalhydraulic analyses, and transient
analyses.
Transient analyses generally include rapid reactivity insertion, runaway rod transient
(e.g. control rods move out from the core at their maximum withdrawal rate), lossof
flow transient and natural convection operation. Natural convection operation is
covered in this study.
13
1.3 Description of the MITR
The MIT Research Reactor (MITR) is a research nuclear reactor that is owned and
operated by the Massachusetts Institute of Technology. Figure 11[1] shows the cutaway
schematic of the MITR. The MITR has two tanks. The inner one is for the light water
coolant/moderator while the outer one is for the heavy water reflector, which is
surrounded by a graphite reflector. Reactor control is provided by six boronimpregnated
stainlesssteel shim blades and one cadmium regulating rod.
Currently, the MITR is licensed for 6MW. As can be seen in Figure 12 [5], the close
packed hexagonal reactor core is designed to be loaded with up to twentyseven
rhomboidal fuel elements. In general, twentyfour fuel elements are loaded during
normal operations. The remaining three positions are filled with either a solid
aluminum "dummy" element or an incore experimental facility. A rhomboidshaped
HEU fuel element consists of fifteen fuel plates and each of them is in the form of
uraniumaluminum matrix and cladded with finned 6061 aluminum alloy to increase heat
transfer area, as depicted in Figure 13 [6].
The MITR operates at atmospheric pressure with nominal primary coolant flow rate 2000
gpm. Primary coolant enters the bottom of the core tank through the core shroud, flows
upward through the fuel elements and then exits at the outlet piping at about 2 m above
the top of the core, as shown in Figure 14[7]. The compact core has an average power
density of about 80 kW/l, with fast, thermal, and gamma fluxes similar to those of a
commercial light water power reactor (LWR). The primary coolant core inlet temperature
of the MITR is approximately 42 *C and outlet temperature is about 50 *C. The
hexagonal core structure is about 38 cm across and the length of an active fuel length is
about 56 cm.
The MITR is designed passively safe that natural circulation and antisiphon valves
(NCVs and ASVs) provide natural circulation path for decay heat removal when forced
convection flow is not sufficient to keep these valves closed during transients. Figure 15
[7] illustrates the flow path for natural circulation. Four NCVs were located at the bottom
of the core tank while two ASVs were installed inside the core tank at the same elevation
of the primary inlet pipe.
Both the NCVs and ASVs are balltype check valves. During normal operation, coolant
pressure forces the ball to the top of the shaft, blocks the top aperture of the valves and
therefore valves are closed. However, when primary flow rate decreases to certain level,
14
the ball falls down since under such a circumstance coolant pressure is not enough to
sustain the ball. As a result, valves are open. These configurations make natural
circulation possible.
As shown in Figure 15, the hot coolant leaving the core rises within the core tank, mixes
with cold coolant in the outlet plenum, reverses, flows through the NCVs and/or ANVs,
and finally flows back through the core region completing the natural circulation loop.
15
Figure 11 Cutaway schematic of the MITR [5]
16
C13 Cc4 C1 Control blade
 /absorber (6)
C1
C12 B8 B39
B1 C2
 fixed absorber In
Fuel radial arm (3)
Element
Core
Core tank structure
Figure 12 M1TR Core map showing fuel element position designations and major
core structures [5]
17
Fuel meat
CL
(d)
CL
Figure 13 Schematic of flow channel configuration of MITR (only 3 fuel plates and 1
supporting plate are shown in the schematic) [6]
18
Figure 14 Forced convection flow circulation path during normal operation [7]
19
Figure 15 Natural convection flow circulation path during LOF [7]
20
1.4 The Proposed Low Enrichment Uranium Fuel Design of the MITR
The high density LEU fuel that MITR currently plans to adopt is the monolithic
uranium and molybdenum (UMo) fuel. Currently, the development of UMo alloy
monolithic fuel is ongoing at Idaho National Laboratory [3, 8]. The mechanical,
physical, and microstructural properties in terms of both integrated and separate
effects of such a fuel were briefly discussed in a study by Burke et al [9].
A study at MIT [10] has demonstrated that LEU conversion is feasible using this alloy
monolithic fuel, which has a uranium density 15.5 g/cm 3 with 10 wt% Mo. The LEU
fuel element designed for the MITR conversion contains 18 fuel plates so that the heat
transfer area is larger than the current HEU design. This LEU configuration was
suggested by Ko [7], that the core tank pressure loading of LEU core should be
limited to be equal or less than that of the current pressure loading of the HEU core.
Tables 12, 13, and 14 [10, 11] compares the properties, as well as the fuel plate
dimensions of HEU and LEU. The thickness of the fuel meat, cladding and coolant
channel are reduced in the new design.
The MITR has two different flow channel configurations, full flow channel and side
flow channel, as illustrated in Figure 13. This is the unique design of the MITR that
a narrow space, which is roughly 50 % narrower than regular coolant channels, is in
between the end of fuel plates and their adjacent elements. These narrower flow
channels are called side channels.
The MITR will employ a transitional core conversion strategy, which means replacing
a few depleted HEU elements with fresh LEU elements for each cycle instead of
replacing all HEU elements at once. A preliminary analysis [12] concluded that in the
mixed core configurations higher Onset of Nucleate Boiling margin is expected in
LEU than that of HEU, therefore allowing the implementation of a transitional core
conversion.
21
Table 12 HEU and LEU fuel plate and fullchannel (interior channel) dimensions [10]
22
Table 13 HEU and LEU Fuel plate and sidechannel (outside channel) dimensions
[10]
Table 14 Composition and thermophysical properties of HEU and LEU fuel [11]
HEU LEU
Compound UAlx U10Mo
Fuel compound density 3.4 17.0
3
(g/cm )
Uranium density (g/cm3) 4.6 15.5
Thermal conductivity 0.42 0.17
(W/cm K)
Heat capacity (J/mol  K) 0.75 0.143
Melting temperature (*C) 1400 1135
23
1.5 Thesis Objectives
24
References
[11] Fact Sheet from National Nuclear Security Administration, "GTRI: Reducing
Nuclear Threats", Feb, 2011.
http://www.nnsa.energy.gov/mediaroom/factsheets/reducingthreats
[15] LinWen Hu, Gordon Kohse, "MITR User's Guide" Rev. 1 June 2008
[16] Sung Joong Kim, Yuchih Ko, Linwen Hu, "Loss of Flow Analysis of the
MIT Research Reactor HEULEU Transitional Cores Using RELAP53D",
proceedings of ICAPP '10, San Diego, CA, USA, June 1317,2010 Paper
10224
25
[110] S. J. Kim, "Memorandum: MITR ThermalHydraulic Parameters,"
MITNRL, June 2011
26
Chapter 2 Engineering Hot Channel Factors (EHCFs)
2.1 Introduction
27
Failure Limit
Margin for Correlation anc
Monitoring Uncertainties
Overpower Factor
Maximum Peak Steady State Condition
(i.e., at hot spot with engineering
uncertainties)
Engineering
Uncertainties
Nominal Peak Steady State Condition
(i.e., at hot spot)
Applicable Axial and
Local Flux Factor (for
LWR)
AxialAverage inRadial Peak Pin
Applicable Radial
Flux Factor
28
2.2 Historical Review on Engineering Uncertainty Treatment
In the very beginning, uncertainties involved in thermal design were addressed either
by directly using their values or using dimensionless factors. Systematic treatments
have been developed and employed in power reactors and research reactors to manage
engineering uncertainties. These methodologies apply different strategy to combine
uncertainties so that they can be categorized as direct deterministic method, semi
statistical method, and fully statistical methods.
The direct deterministic method is usually applied in the preliminary stage of core
design, directly taking all the parameters at their worst value assuming their
occurrence at the same time and same location, which is highly conservative.
Another strategy that has been employed in the core design of PWRs is treating
uncertainties using hot spot factors. All the engineering uncertainties are expressed as
dimensionless hot spot subfactors. Every subfactor is carefully defined and
represents different kinds of engineering uncertainties. Subfactors are then combined
either statistically since they are assumed to be independent, or multiplicatively,
therefore resulting in conservative estimates. The statistical combination of these
factors is defined as statistical method while the multiplication of these subfactors is
defined as deterministic combination.
For current core designs of PWRs, the fully statistical methods are widely employed
for its potential in uncertainties reduction. As summarized in Yang and Oka's study
[4], the methods can be further divided into two categories: the methods applying the
Root Sum Square (RSS) technique [5, 6, 7] and the methods applying Monte Carlo
technique [4, 8, 9]. Han [10] summarizes a general statistical formula used to
combine the uncertainties that the sensitivity of each parameter to departure from
nucleate boiling ratio (DNBR) is incorporated to estimate the overall uncertainties of
a PWR core.
29
2.3 Introduction to Engineering Hot Channel Factors (EHCFs)
2.3.1 EHCFs
Hot channel factors are dimensionless factors used to address the extent to which
actual reactor performance may depart from its nominal performance, owing to the
cumulative effect of variations of all primary design variables from their nominal
values. Hot channel factors are composed of contributions due to nuclear and
engineering considerations, which are assumed to be separative. Nuclear hot channel
factors, or known as power peaking factors express the peak to average ratio of the
nuclear power distributions radially and axially in the core, which are due to the
variation in neutron flux; engineering hot channel factors (EHCFs), which are
evaluated at constant neutron flux, express the uncertainties in local enthalpy rise,
heat flux and heat transfer coefficient due to the fabrication tolerance and flow
maldistributions [11].
Generally, EHCFs may arise from manufacturing dimensional tolerances on the fuel
elements or coolant channels, or dimensional changes of the fuel elements after
irradiation, or from deviations from an ideal flow pattern in the reactor core and
plenum chambers [3]. EHCFs may be categorized into three parts corresponding to
the change in parameters the uncertainties make contribution to: the heat flux in
reactor core, the film temperature rise in reactor core channel, and the temperature
rise or enthalpy rise in the channel [3, 12]. These three parts are illustrated as follows:
1. Uncertainties that influence the heat flux: Heat flux hot factor, FQ, is defined as
the ratio of the highest heat flux which could possibly occur anywhere in the
reactor core to the average heat flux.
Where the subscript hc refers to hot channel and nc refers to nominal channel.
2. Uncertainties in film temperature rise: film temperature rise hot factor, FAT, is
defined as the ratio of the maximum film temperature rise, which could possibly
occur anywhere in the reactor core channel to the average film temperature rise.
30
Where A T, is the increase in surface temperature (i.e. fuel cladding temperature)
3. Uncertainties in the temperature rise or enthalpy change in the channel: Coolant
temperature rise or enthalpy change in the channel hot factor, FH, is defined as the
ratio of the maximum coolant temperature rise which could possibly occur in any
fuel assembly of the reactor core to the average temperature rise.
Where A Tb is the increase in bulk temperature rise of between inlet and outlet.
or statistically
Where fQi are the subfactors involved in the deviation of heat flux from its nominal
value.
The EHCFs used for the MITR use the latter formula, which is discussed in section
2.4. The selection of subfactors and EHCFs for the thermalhydraulic analysis of the
hottest channel can have a significant impact on reactor safety margins. For instance,
the uncertainty in the heat transfer coefficient is a major contributor to the reduction
in thermalhydraulic safety margins, as indicated by Woodruff in 1997[12].
31
2.3.2 Common subfactors involved
Some common subfactors used in uncertainty analyses are summarized below, which
are taken from LeTourneau's study [2], which explicitly defined uncertainties
involved in reactor design. The first six factors contribute to unequal flow
distribution, and the last three are responsible for the changes in heat flux distribution.
In some analyses, although the complete independency of each subfactor is not
strictly verified, it is convenient and therefore somewhat conservative to consider
them as independent [3].
1. Even though the core geometry is ideal, the particular geometry of the reactor inlet
plenum and the entrances region to the individual coolant channels may give rise
to an unequal flow distribution. This factor is defined as plenum factor.
2. Deviations from the nominal design dimensions of the individual coolant channels
will cause unequal flow distribution among the channels. This is known as
channel tolerance factor.
5. Generally, for a compact coolant channel, such as a round tube, the coolant
enthalpy is taken as the mixingcup average over the cross section. Perfect fluid
mixing in the channel is thereby implied. However, for a noncompact channel,
like a wide, narrow, rectangular tube, it is convenient to assume perfect mixing
across the narrow direction and no mixing across the wide direction. In such a
channel, one may account for mixing in the wide direction by introducing a hot
channel factor less than unity called mixing factor. A similar interchannel mixing
factor may be defined where coolant channels are interconnected, as is the case
when rods or spheres are used as fuel elements.
6. If local or bulk boiling occurs in some channels due to the spatial variation of heat
32
generation in the reactor, the increased pressure drop per unit flow rate caused by
boiling will affect the flow distribution among the channels. The hot channel
factor is defined as the boiling factor.
7. If the volume of fuel material generating heat removed by a unit of surface area is
not uniform in the reactor core, the heat flux distribution will be affected. This
phenomenon may occur due to manufacturing tolerances applicable to the portion
of the fuel elements containing fissionable material. The hot channel factor is
called fuel element tolerance factor.
8. Similarly, if the number of fissionable atoms per unit volume of fuel material is
not constant in the reactor core, the heat flux distribution will be affected. The
disparity may arise from metallurgical tolerances on fuel material composition and
enrichment. The hot channel factor is named fuel density factor.
9. If the fuel material is separated from the coolant by cladding material, variations
in the thickness of this cladding around the perimeter of the heat transfer surface
of the fuel element due to manufacturing tolerances will cause variations in heat
flux distribution at the heat transfer surface compared to the symmetrical case.
The hot channel factor is defined as eccentricity factor.
The mixing factor may be determined wholly by experiment using dye or other tracer
techniques, or analytically using experimentally determined mixing coefficients. To
finish the calculation, one must know the flow distribution, the channel dimensions
and the coolant velocity. There's seems to be no general expression applicable to any
reactor.
The fuel element tolerance factor, the eccentricity factor, and the fuel density factor,
may be determined by application of the steadystate heat conduction equation to a
33
fuel element of nominal dimensions, and to a fuel element of worst allowable
dimensions which will result in a maximum heat flux, in a region of the same neutron
flux.
34
2.4 EHCFs Used in MITRII
The subfactors and statistically combined EHCFs used in the SAR [13] are
summarized in Table 21. Table 22 summarizes the definition for these subfactors
[2, 3, 13].
where n is the number of standard deviation that is incorporated into the subfactor,
o is standard deviation and p is the nominal value of parameters. However, how
these subfactors were obtained was not clearly documented in MITRII SAR [13], or
in the previous version dating back to 1970 [14]. There is no sufficient information
indicating the value of n in Eq. 26.
For example, the uncertainty for heat transfer coefficient is 1.20, according to table 2
1. If three standard deviations are assumed to be incorporated in this value, the
uncertainty of the heat transfer coefficient distribution is one third of 20%, which is
6.7%. Subfactors involved in subsequent calculation, as shown in Chapter 4, are
treated in this manner.
35
where fi represents subfactors and F is EHCF referring to FH, FQ and FAT in Table 21.
As can be seen in Table 21, subfactors were categorized into three parts contributing
to the uncertainties in enthalpy rise, film temperature rise, or heat flux respectively.
36
Table 21 Subfactors and EHCFs used in MITRII SAR [13]
Enthalpy Rise
Reactor power measurement 1.050
Power density measurement/calculation 1.100
Plenum chamber flow 1.080
Flow measurement 1.050
Fuel density tolerances 1.026
Flow channel tolerances 1.089
Eccentricity 1.001
FH Statistical 1.173
Film Temperature Rise
Reactor power measurement 1.050
Power density measurement/calculation 1.100
Plenum chamber flow 1.060
Flow measurement 1.040
Fuel density tolerances 1.050
Flow channel tolerances 1.124
Eccentricity 1.003
Heat transfer coefficient 1.200
FAT. Statistical 1.275
Heat Flux
Reactor power measurement 1.050
Power density measurement/calculation 1.100
Fuel density tolerances 1.050
Eccentricity 1.003
F0 , Statistical 1.123
37
Table 22 Description for subfactors used in MITRII SAR [2, 3,13]
Subfactor Definition
Even though the core geometry is ideal, the
Plenum factor particular geometry of the reactor inlet plenum
(Plenum chamber flow) and the entrances region to the individual
coolant channels may give rise to an unequal
flow distribution.
If the number of fissionable atoms per unit
volume of fuel material is not constant in the
Fuel density factor reactor core, the heat flux distribution will be
(Fuel density tolerances) affected. The disparity may arise from
metallurgical tolerances on fuel material
composition and enrichment.
Channel tolerance factor Deviations from the nominal design dimensions
(Flow channel tolerances) of the individual coolant channels will cause
unequal flow distribution among the channels.
If the fuel material is separated from the coolant
by cladding material, variations in the thickness
Eccentricity factor of this cladding around the perimeter of the heat
(Eccentricity) transfer surface of the fuel element due to
manufacturing tolerances will cause variations
in heat flux distribution at the heat transfer
surface compared to the symmetrical case.
38
References
39
[210] K. I. Han, "Technical Review on Statistical Thermal Design of PWR Core",
Journal of the Korean Nuclear Society, Vol. 16, No.1, March, 1984.
[213] MIT Nuclear Reactor Laboratory, "Safety Analysis Report for the MIT
Research Reactor," MITNRL 1102, August, (2011)
[214] Safety Analysis Report for the MIT Research Reactor (MITRII),
MITNE 115, October 1970.
40
Chapter 3 Limiting Safety System Settings
Similar to the description above, LSSS is defined in NRC Glossary [2] as " settings
for automatic protective devices related to those variables having significantsafety
functions. Where a limiting safety system setting is specifiedfor a variable on which
a safety limit has been placed, the setting will ensure that automaticprotective action
will correct the abnormalsituationbefore a safety limit is exceeded."
Hence, LSSS are limits established to guarantee that there is sufficient margin
between the normal operating conditions and the safety limits. The prevention of
nucleate boiling (ONB) within the coolant channels is chosen to derive LSSS in
thermalhydraulic analysis [3]. For narrow rectangular coolant channels, such as
those in the M1TR, Sudo et al. [4] suggested to apply the BerglesRohsenow
correlation [5] to predict the occurrence of the onset of nucleate boiling (ONB). This
suggestion was based on comparisons of several existing correlations with
experimental data.
The LSSS specifically for the MITR in forced convection mode [6] is set for:
a) The maximum reactor power,
b) The maximum steadystate average core outlet temperature,
c) The minimum primary flow rate, and
d) The minimum coolant level in the core tank.
If operating conditions are in the region below the LSSS curve shown in Figure 31
[3], it is guaranteed that boiling will not occur anywhere in the core under all credible
conditions and that means the safety limits will not be exceeded. The safety limits are
established to ensure the integrity of the fuel clad, to prevent fission product release.
41
For the MITR, both CHF and OFI are calculated and the one that would occur first is
adopted as the safety limits for conservatism. The analysis for the safety limit is
included in Chapter 6. The ONB limit is calculated based on the fuel clad
temperature. The analytical expression was presented in the MITR SAR report [3].
Both the bestestimate value for ONB and the LSSS calculated using engineering hot
channel factors (EHCFs) are presented in this section. The bestestimate ONB is the
LSSS obtained at the most limiting coolant channel (or fuel plate stripe for the hot
stripe approach described in section 3.7) that the parameters are taken at their nominal
values, e.g. without taking into account engineering uncertainties. In contrast, LSSS
calculated using EHCFs takes the uncertainties into consideration via the usage of
EHCFs.
There are numerous system parameters and notations involved in LSSS calculation.
Their definitions are summarized in Table 31.
42
Table 31 Parameters used in LSSS calculation
Symbols Definitions
Ff Core coolant flow factor, the ratio of the primary coolant flow that
actually cools the core region to the total flow, which is 0.921 [5]
dr Flow disparity factor of M1TR, the ratio of (minimum
flow/average flow) for coolant channels within a fuel assembly,
which is 0.93 [6]
WP Total primary coolant flow rate [3], which is taken as 1800 gpm
for LSSS calculation.
N, The number of coolant channels in the core region, which is 432
for the proposed LEU fuel design [3].
Fr Radial power peaking factor, which is assumed as 1.76 for LEU
core [3]
Fs Lateral power peaking factor such that FrF, is assumed to be 2.12
[7] for the hot stripe approach.
Fcore Core power deposition factor, the fraction of the fission power
deposited in the core region (fuel &coolant) of the core tank,
which is 0.965 [3]
Ffuel Fuel power deposition factor, the fraction of the core power
deposited in
the fuel elements, which is 0.94 [3]
rh Hot channel mass flow rate
P Reactor operating power
2
AH Heat transfer area per channel, which is calculated as 0.12357 m
PH Heat transfer perimeter per coolant channel, which is calculated as
0.2115 m
Tin Coolant temperature at channel inlet
Tout Coolant temperature at channel outlet
h Heat transfer coefficient
p System pressure
z Elevation in coolant channel
O(z) Normalized axial heat flux distribution factor
q"(z) Local heat flux
FAT EHCF featuring film temperature rise, which is 1.275 [3]
FH EHCF featuring enthalpy rise, which is 1.173 [3]
43
9
6 
Fcore, 2.0
Fgdf, =0.8
1.173
FAT = 1.275
4   
50 55 60 65 70 75 80
Figure 31 MITR HEU LSSS for forcedflow operation (twoloop) [3]
44
3.2 Derivation of LSSS
The basic idea of LSSS calculation is that the maximum cladding temperature is no
greater than the temperature that induces ONB, as expressed in Eq. 31.
For narrow rectangular channels, like channels in MITRII, Sudo et al. [4] suggested
to apply the BerglesRohsenow correlation [5] to predict the occurrence of the onset
of nucleate boiling (ONB).
The BerglesRohsenow correlation, which relates the wall temperature with applied
heat flux when ONB occurs, is used for the derivation of LSSS for MITR, as shown in
Eq. 32. The original form of correlation can be rewritten if the thermal properties are
specified, as shown in Eq. 33. The latter was used in the previous version of MITR
SAR [3] in which pressure was estimated at 1.3 bar corresponding to saturation
temperature 1070 C. This pressure corresponds to a coolant height of 10 feet (3 meters)
above the top of the fuel plates or 4" below overflow.
q 9O.0234
1156) 2.16
45
3.2.2 Cladding Temperature
The cladding temperature in Eq. 31 can be obtained using energy balance. The
subscript "hc" for heat flux refers to hot channel heat flux. The description for every
term used in the following equations is summarized in Table 31. As written in Eq. 3
4, the bulk temperature of coolant at certain axial level can be expressed in terms of
inlet coolant temperature and heat flux applied to the channel. The difference in bulk
coolant temperature and cladding temperature can be expanded to include heat
transfer resistance, as shown in Eq. 35.
T..,,(z)=7z+ .F
mhCpf 0 (Eq. 34)
Next, the channel inlet temperature in Eq. 35 can be expressed as a function of outlet
temperature and reactor operating power, as can be seen in Eq. 36. Substituting all
these terms into Eq. 33, the expression for LSSS is obtained as shown in Eq. 37.
T= PF,,,

WP f (Eq. 36)
where
P
q" (z)=  F ,CFOeF,O(z)
NcAH
As explained earlier, LSSS is used to set a limit so that ONB does not occur
everywhere in core region. To derive such a limit (upper limit for cladding
temperature), simply equalizing cladding temperature with temperature that induces
46
ONB, as shown in Eq. 38.
P F,,. F F qhcz)
Pe.
 j Pq",e(z)dz+F
WPc, MC, h
In Eq. 310, the objective is to derive maximum allowable outlet temperature Tour
when operating power P is specified, or to obtain maximum allowable operating
power P when outlet temperature Tot is fixed. In either case, the LSSS curve for
MITR LEU can be obtained similar to that of HEU as shown in Figure 31 [3].
47
3.2.3 General Form of LSSS Equation
There are many parameters in Eq. 310, which can be categorized into three groups:
system parameters, local properties and engineering hot channel factors. These
parameters can be combined to simplify the LSSS equation. As a result, Eq. 310 can
be reduced to,
These system parameters, local properties and EHCFs used in LSSS calculation are
described in detail in the following sections. Eq. 311 is derived in this study and is
applicable for most of the platetypefuel research reactors since ONB is the common
concern for these research reactors of narrow coolant channels and high power density.
Note that the coefficients C 1C 5 in Eq. 311 are different when applying different
methodologies to calculate LSSS. For the bestestimate LSSS in the most limiting
channel, every parameter is taken at their nominal values and EHCFs, the factors
characterizing accumulative uncertainties, are set as unity. For the LEU LSSS
calculated using EHCFs, all parameters are taken at their nominal values, but EHCFs
are set as what they were in HEU analysis [3] to account for the parametric
uncertainties. For the uncertainty propagation methodology, some key input
parameters are set as normal distributions reflecting for the parametric uncertainties
and EHCFs are set as unity since the parametric uncertainties are reflected using
parametric normal distributions in this methodology.
48
3.3 Parameters Used in LSSS Calculation
The system parameters, local properties, and EHCFs used in LSSS calculation are
summarized in this section.
The system parameters used in LSSS calculation can be found in Table 32. The LEU
fuel element consists of 18 fuel plates so that the heat transfer area is larger than the
original HEU design (15 plates). This LEU configuration was suggested by Ko et al.
[8] for reason that the core tank pressure loading of LEU core should be limited to be
equal or less than that of the current pressure loading of the HEU core. Total number
of flow channels is 432, which is calculated assuming 18 flow channels per element
and 24 fuel elements in core. Note that the number of channels 432 is obtained
assuming that two halfchannels form a full channel. This assumption simplifies the
analyses. Derivation of thermal hydraulic limits for halfchannels is not within the
scope of this study.
Since the neutronic analyses for the proposed LEU core were ongoing, core power
deposition factor Fcor, fuel power deposition factor Fjei, and radial power peaking
factor Fr are adopted from the previous version of MITRII SAR[3], which are 0.965,
0.94 and 1.76 respectively. These values were used in the LEU calculations at this
stage. The values will be updated in the near future to reflect the LEU core more
precisely [9].
The primary flow rate (kg/sec) is calculated using 1800 gallon per minute, which is
the current LSSS flow rate, and assuming average coolant temperature is 55 'C in
LEU core. The heat capacity, Cp, was taken at the aforementioned temperature
assuming a pressure of 1.3 bar. This pressure was also used in the uniform heat flux
assumption in [3]. Heat capacity at constant pressure Cp is not sensitive to the change
in pressure. Therefore, the assumption 1.3 bar made in the calculation has negligible
effect for the result.
According to the initial startup testing of the MITRII [6], about 92.1% of primary
flow enters core region of the MITRIL. The flow distribution in the reactor core was
also measured during the MITRII's initial startup testing. The minimum flow
through a fuel element is 93% of the average core flow rate [6]. The flow distribution
within a fuel element has also been measured experimentally using a dummy element
49
as 92.9% [6]. To calculate the worst case for a flow channel receiving minimum flow,
intuitively all of the three flow disparity factors should be included such that the
minimum flow for a channel is (92.1%x93%x92.9%) of the primary flow. However,
when calculating the mass flow rate used in Eq. 310, the factor 92.9% actually is
removed since the EHCFs are included in Eq. 310. The exclusion of the factor
92.9% is because the inclusion of EHCFs implies that the discrepancies among the
channels, such as flow rate, heat flux and so on, are taken into account in the
calculations. To avoid doublecounting mass flow rate discrepancy among the
channels, the factor 92.9% is therefore removed.
One thing has to be clarified here is that the hot channel does not necessarily receive
minimum channel flow. Instead, it is fairly possible that the channel receive
minimum flow does not possess peaking power. Therefore, the inclusion of flow
disparity for hot channel mass flow rate calculation is a conservative assumption.
Other geometry terms such as flow area and heat transfer area in Table 32 are derived
from the fuel geometry of the MITR. These geometry terms are discussed in detail in
section 3.4.
50
Table 32 Parameters used for analytical LSSS calculation
51
3.3.2 EHCFs
As can been seen in Eq. 310, two EHCFs are included in LSSS calculation: enthalpy
rise factor FH and film temperature rise factor FAT. According to the previous version
of the SAR [3], these two factors were the accumulative result of uncertainties
involved respectively in enthalpy rise and film temperature rise. The subfactors
involved in enthalpy rise and film temperature rise are summarized in Table 21 [3].
In this study, the values of EHCFs are directly adopted from [3] and used in one of the
methodologies to calculate LSSS.
52
3.3.3 Local Properties: Coolant pressure
The fueled region of a MITR channel is divided axially into ten equal distance nodes
in several thermalhydraulic computer models used for steadystate and transient
simulations, as can be seen in Ko's study [8]. Local properties in this study refer to
the fluid temperature, pressure, HTC and axial power distribution factor respectively
for each of these axial nodes. In the following sections, how these properties were
estimated is explained in detail.
The static pressure corresponding to 10 feet of coolant above the top of the fuel is 1.3
bar (which has a saturation temperature 107*C). Therefore, for the bottom node of the
fuel, the pressure is calculated to be the sum of 1.3bar plus the equivalent liquid
pressure of 10 ft water plus the equivalent liquid pressure of 0.5842 meter (fuel height)
water. The pressure of the top node of the fuel is the pressure of the bottom node
minus the pressure loss, where the pressure loss is the sum of frictional, gravitational
and nearly negligible acceleration pressure drop across the core.
The major contribution of the pressure drop is frictional pressure drop, which can be
expressed by Eq. 312 [10],
2
APfricti = f  )  )pV (Eq. 312)
where
Note that the friction factor f is taken from Wong's thesis [11], a friction factor
correlation developed for MIT finned rectangular channel.
The correlations used to calculate gravitational and acceleration pressure drop are
respectively shown in Eq. 314 and 315 [10].
53
where G is mass flux and p is fluid density.
The water properties used for pressure loss calculations are based on the
temperature/pressure assumption described in section 3.3.1. The summary for
pressure loss calculation and pressure at each node, and their corresponding saturation
pressure are summarized in Table 33 and Table 34. The total pressure loss is
calculated as 35,459 Pa, which is mostly contributed by the frictional pressure drop.
These pressure drops are calculated based on the estimated mass flow rate of the hot
channel, 0.2219 kg/sec, as shown previously in Table 32. In pressure drop
calculation, the thermal properties are assumed as constant for simplicity, and are
taken at temperature being 55*C and pressure at 1.3 bar.
54
Table 33 Summary for the MITR pressure loss calculation
Table 34 Pressure calculated for each node of the fueled region
2 1.61 113.5
3 1.57 112.8
4 1.53 112.0
5 1.49 111.2
6 1.45 110.4
7 1.41 109.6
8 1.37 108.8
9 1.33 107.9
10 (Top) 1.30 107.1
55
3.3.4 Local Properties: Axial Power Distribution and Coolant Temperature
The preliminary power distribution analyses conducted by ANL indicated that the
most limiting fuel element amongst the core is reference core 189 at the end of cycle
(denoted as 189EOC as follows) [7]. For conservatism, the power profile of this fuel
element is used to calculate LSSS. This LEU fuel axial power is bottompeaked, as
illustrated in Figure 32. The axial power peaking of 189EOC, the ratio of maximum
toaverage axial nodal power is 1.27.
Coolant temperature increases as coolant flowing through a heated channel, and the
increase in coolant temperature within certain flowing distance is related to local heat
flux. Since the axial power profile is bottompeaked, the increase in coolant
temperature is expected to be larger in the lower portion of the fueled region. This
trend can be observed in Table 36.
In this section, the values of heat capacity at constant pressure and mass flow rate are
assumed constant throughout the calculation. The coolant temperature for each node
is calculated assuming a reactor power of 8.4 MW, which is the presumed LSSS
power for MITR LEU core, and average core outlet coolant temperature of 60'C,
which is one of the specified conditions for LSSS. The inlet coolant temperature can
be obtained using these two assumptions via Eq. 36. The inlet temperature is
calculated to be 43C. Once the inlet coolant temperature is determined, the coolant
temperature at each node for the hot stripe can be obtained using the axial power
profile of 189EOC, as summarized in Table 35. The sensitivity study of coolant
temperature on LSSS calculation can be found in Chapter 5.
The LSSS for the MITR HEU configuration is 7.4 MW as documented in the SAR [3],
which is 20% larger than 6 MW, the steadystate power of the HEU configuration.
Similarly, the LSSS for the MITR LEU is expected to be 8.4 MW, which is 20%
larger than the target licensing LEU power 7 MW.
56
10
5
0
0
3
2
1
0.02 0.06 0.1 0.14
Table 35 Axial power distribution for the MITR LEU fuel (189EOC) [7]
1 0.117
2 0.115
3 0.123
4 0.127
5 0.127
6 0.123
7 0.109
8 0.084
9 0.04
10 0.02
57
Table 36 Hot channel coolant temperature calculated for core189
Node Coolant
Number Temperature(*C)
1 46.60
2 50.47
3 54.60
4 58.85
5 63.12
6 67.26
7 70.92
8 73.73
9 75.27
10 76.13
58
3.4 Heat Transfer Coefficient
Heat transfer coefficient plays an important role in LSSS calculation. This can be
seen later in Chapter 5 (Sensitivity Study). The purpose of this section is to obtain the
bestestimate value for heat transfer coefficients (HTC) at each node in the fueled
region of the MITR.
As can been seen in the existing studies [8, 11, 12], there are two ways to compute
HTCs for the MITR. One is the conventional DittusBoelter correlation [13] in
conjunction with the enlarged heat transfer area; the other is Carnavos correlation [14],
an empirical correlation for finned channel. The enhancement factor for heat transfer
area used in the former approach was calculated as 1.9 by S. Parra [9] and this value
is used for MITR's heat transfer calculation.
The fuel plate geometry of MITR and the notations for each parameter are firstly
introduced for the better understanding of the subsequent calculation of HTCs. Figure
13 shows part of the side view of a fuel assembly of the MITR noting that the
proposed design for LEU has 18 plates per assembly [8]. Table 37 summarizes some
derived geometry parameters for the MITR and these parameters will be used in
subsequent HTC computation. In this table, the term "nominal" refers to a situation
that geometry parameters are calculated as if the fins were not present, comparing to
the term "actual" calculating geometry parameters taking into consideration the
presence of fins. The coolant properties used for the HTC calculation are summarized
in Tables 34 and 36.
59
Table 37 Derived Geometry Parameters for MITR (LEU)
60
3.4.1 Carnavos Correlation and Geometry Analysis for MITR
=h.Dha Nu
Where Nu and the predicted value of is within 10% error.
k Pr*.
Nu, Re and Pr are Nusselt, Reynolds and Prandtl Number respectively. The subscript
"a" used in Eq. 316 represents that actual parameters used to compute for the
hydraulic diameter. The other terms in Carnavos correlation and their counterparts in
the MITR are given in Table 39.
*These Reynolds numbers were calculated based on hot channel flow with primary flow 1800 gpm.
The Reynolds number for average channel flow with primary flow 2000 gpm is between 7,700 and
14,000.
61
Table 38 The LEU geometry parameters in Carnavos correlation and their
counterpart in MITR
62
3.4.2 HTC Computed from DB Correlation and Carnavos Correlation
The DittusBoelter correlation (DB correlation) mentioned in this study refers to the
wellknown correlation of Eq. 317. It is interesting to know Wintertwon [16]
indicated that this widelyused equation was not originally proposed by Dittus F.W.
and Boelter L.M.K.[13], but was actually introduced by McAdams [17]. However, to
avoid confusion, the name DittusBoelter correlation is still kept in this study.
hD
Nu = = 0.023. Re*8 Pr*4 (Eq. 317)
k
Table 310 summarizes three approaches used to calculate HTC. Approach A and
approach B are based on DB correlation while approach C is based on Carnavos
correlation. In approach A, the actual wetted parameter and flow area were applied
and therefore the area enhancement factor is 1.0.
In contrast, the hydraulic diameter in approach B was calculated as if fins were not
present, and therefore the area enhancement factor is 1.9. There are no geometry
correction factors for DB Correlation, so geometry correction factors are set as 1.0 in
these two approaches. Approach C calculates HTC using Carnavos correlation, which
is developed for finned channel and therefore the area enhancement factor is 1.0.
In the previous sections, the temperature and pressure for each node has been
estimated. Therefore, HTC for each node can be calculated using approach A, B and
C respectively and are summarized in Table 310. The HTCs calculated by approach
C are the lowest among these three approaches. For conservatism, HTCs computed
from Carnavos correlation were taken for LSSS calculation.
63
Table 39 HTC calculated using DB Correlation and Carnavos Correlation
64
3.5 Best Estimate ONB
In section 3.2, a general form of LSSS equation was derived. Moreover, in sections
3.3, 3.4 and 3.5, all of the parameters required to obtain LSSS power, including
system parameters, EHCFs and local properties have been introduced. Combining of
all of these parameters, the coefficients in Eq. 311 can be computed but note that
these coefficients would vary if different methodologies were used, as explained in
section 3.2.
This section summarizes the bestestimate value for LSSS, which does not take into
account parametric uncertainties. That is, the two EHCFs in Eq. 311, FH and FAT,
were set as unity because EHCFs themselves reflect the accumulative result of
parametric uncertainties. The input parameters for LSSS calculation are set as a
single value and nominal values are used. The assumption made to compute the
LSSS for the hot channel is that LEU radial peaking factor is 1.76 [3].
Substitute the local properties of node #7 (as indicated in Table 36, pressure is 1.4 bar
and saturation temperature is 109.6*C) into Eq. 311, it becomes
466
T., = T,, +2.261 PO. 3.93. P (Eq. 318)
1 F reFFf.O(z)
C, =0.556[ Nc  AH0.463P0234
1 156
= 2.261
1082p .
ZP H F e F,.Ffel(z))dz
C4 = N, A" . = 3.36
mcp
1FcorFrFffieio(Z)
cr
AAH= h=
C,NeC5 .
2.632
h
Above LSSS calculations are for node #7, other nodes can be computed in a similar
manner. Node #7 is taken as an example above because this node has the most
65
limiting ONB margin, which can be seen later.
As can be expected in Eq. 318, when power increases, the negative contribution
brought by the last term on the RHS (3.93P) is larger than the positive contribution
brought by the second term (2.26 1PA6). As a result, when power increases,
allowable outlet temperature decreases. This trend agrees with the intuition that when
operating power increases, the maximum allowable outlet temperature should
decrease to prevent the occurrence of nucleate boiling.
66
16
14
12 node 1
 node 2
10 10. node 3
node 4
8
0  node 5
6 node6
node 7
  node 8
 node 9
2
 node 10
0
50 60 70 80 90 100
Tout (*C)
67
3.6 LSSS Calculated Using EHCFs
F.,f FydO(z)
C =0.556. [ N,  AH 1 156 10.463p0.0 = 2.261
1082p .
0 2 34
C2 = 0.463 p0 . = 0.466
F
C3 core = 2.062
WP CP
F  N A coreFFfuei(z))dz
C4  Nc  A= 3.941
th c,
1
F
Fcore rFfueO(z)
C5 = FAT' c H  3.356
h
The trend of Eq. 319 is same as Eq. 318. When power increases, the negative
contribution brought by the last term on the RHS (5.23P) is larger than the positive
contribution brought by the second term (2.261Po.466). As a result, when power
increases, allowable outlet temperature decreases.
Similar to the bestestimate ONB, the result of node #7 at outlet temperature 60'C was
taken as the reference value (10.48 MW) for this methodology because this node has
the most limiting LSSS power, as demonstrated in Figure 34.
68
16
14
12
node 1
 node 2
10
A node 3
8 node 4
  node 5
4  node 8
  node 9
2   node 10
0
50 55 60 65 70 75 so
Tout (*C)
Figure 34 LSSS power computed using EHCFs for corel89 on each node
69
3.7 LSSS Calculated Using Hot Stripe Technique
Given the same conditions, using hot stripe factor leads to more limiting results
comparing to using radial peaking factor because the radial power peaking factor is
replaced with the hot stripe factor, and the latter is roughly 20% larger. Divide one
fuel plate into four strips, adopt the one has the highest flux as shown in Figure 35,
and perform LSSS calculation. Similar hot stripe approach has been used in the
thermal hydraulic analyses of Missouri Research Reactor [19].
Fe
PF. + Pjjq'h(z)dz + FT he
W,c, rhc, h
<[ q] hc 1(Z)0.463
56
+sat =0 (Eq. 37)
1.8 1082p1
Where
P
q"c(z)= FfiFc,,corF,0(z)
h Nc AH
70
Rewrite the heat flux term in Eq. 37 describing the most limiting hot stripe such that,
P
qhs(z) = F FF, (z)
NAH
Where FrFs is 2.12 and subscript hs refers to hot stripe, so that Eq. 37 becomes,
PF'"
re (zhdz+
 + Zd
WPc, mc ,q. h
Table 32 summaries the system parameters used in hot stripe LSSS calculation. In
hot stripe approach, it is assumed that there is no lateral mixing and the flow is the
same as that in hot channel approach. Table 310 summarizes the thermalhydraulic
conditions used in hot stripe LSSS calculation for each node. The coolant
temperature for each node is calculated assuming the reactor power is 8.4 MW, which
is the presumed LSSS power for MITR LEU core, and average core outlet coolant
temperature is 600C, which is one of the specified conditions for LSSS. The inlet
coolant temperature can be obtained using these two assumptions via energy
conservation. Once the inlet coolant temperature is determined, the coolant
temperature at each node for the hot stripe can be obtained using the axial power
profile of 189EOC, as summarized in Table 310. Note that the coolant temperature
at outlet is higher than the case based on power radial peaking factor, as shown in
Table 37. The temperature is higher is because the heat flux of the most limiting
channel calculated based on hot stripe consideration (2.12) is of greater value than
using radial peaking power factor (1.76).
71
Table 310 T/H conditions used in hot stripe LSSS calculation for each node
Normalized Coolant
Node # Power factor Pressure (bar) Temperature
(189EOC) ("C)
0
0c.
0
0 2 4 6 8 10 12 14 16 18 20 22
Distance from Bottom of Fuel (inches)
Figure 35 Heat fluxes of strips on MIT LEU core 189EOC [6]
72
3.7.1 Best estimate ONB using hot stripe technique
The bestestimate ONB using hot stripe technique is obtained by setting EHCFs in Eq.
320 to be unities, since the parameters are taken at their nominal values. The most
limiting ONB occurs at node #7, which is 11.1 MW with outlet temperature being
60*C, as shown in Figure 36.
73
16
14
7 4 . node 1
12 node 2
yAn 3
10 ~node
8  Mnode 4
o  * node 5
6  node 6
4  node 7
node 8
2
node 9
0
50 55 60 65 70 75 80  node 10
Tout (*C)
74
3.7.2 Hot stripe LSSS using EHCFs
As demonstrated in Figure 37, the most limiting node for 189EOC is node #7. The
analytical LSSS for node #7 using hot stripe approach is calculated as 8.36 MW.
75
16
14
12
node 1
 node 2
10
A node 3
node 4
84i
node 7
4 + node 8
node 9
2  node 10
0
50 55 60 65 70 75 80
Tout (*C)
76
3.8 Summary
It has been shown that node #7 predicts the most limiting LSSS power and therefore
for conservatism, the LSSS power of this node is adopted for the subsequent analyses.
The LSSS powers illustrated in Figures 38 310 are all LSSS powers of node #7
obtained based on different approaches. The analytical approach takes into account
the parametric uncertainties via the usage of EHCFs while the bestestimate ONB
does not take into the parametric uncertainties but provides the bestestimate ONB
results. Given the same conditions, the LSSS power calculated using hot stripe
technique predicts much lower LSSS power than the conventional radial power
peaking factor since the heat flux of the most limiting channel calculated based on hot
stripe consideration (2.12) is of greater value than using radial peaking power factor
(1.76).
Figure 38 compares the bestestimate ONB based on radial peaking factor and hot
stripe factor. As expected, the one based on hot stripe factor, 11.1MW, is more
limiting than that based on radial peaking factor, 14MW. The similar trend can be
also observed in LSSSs calculated using EHCFs that the one based on hot stripe factor,
8.36MW, is more limiting than that based on radial peaking factor, 10.48MW, as
shown in Figure 39. Table 312 summarizes these LSSSs calculated using different
approaches. For conservatism, the LSSSs based on hot stripe factor are adopted as the
final results, as shown in Figure 310. These analytically obtained LSSSs for node #7
are going to be used for the subsequent analyses in the following chapters.
77
Table 311 Summary for LSSS powers calculated using different approaches
16
14
12
 10
4BE (radial peaking)
0
50 55 60 65 70 75 80
Tout (*C)
Figure 38 Bestestimate ONB comparison using radial peaking factor and hot stripe
factor
78
16
14
12
10
I
8 +Using EHCFs (radial peaking)
a
0~ 6  Using EHCFs (hot stripe)"
U)
U) 4
U)
I
2
0
50 55 60 65 70 75 80
Tout (*C)
Figure 39 LSSS calculated using EHCFs based on radial peaking factor and hot
stripe factor
16
14
12
10
I
8 0 BE (hot stripe)
0
6 . Using EHCFs (hot stripe)
U)
U) 4
U)
I
2
0
50 55 60 65 70 75 80
Tout (*C)
Figure 310 Hot stripe LSSS calculated using EHCFs and best estimate approach
79
References
[33] MIT Nuclear Reactor Laboratory, "Safety Analysis Report for the MIT
Research Reactor," MITNRL1102, August, (2011)
[37] E.H. Wilson, N.E. Horelik, F.E. Dunn, T.H. Newton, Jr., Linwen Hu, and J.G.
Stevens, "Power Distributions in Fresh and Depleted LEU and HEU Cores of
the MITR Reactor," ANLRERTRTM1203, Argonne National Laboratory
(2012).
[39] S. Parra, The Physics and Engineering Upgrade of the Massachusetts Institute
of Technology Research Reactor, Ph.D. Thesis, MIT Nuclear Engineering
Department, 1993.
80
[311] Susanna Wong, LinWen Hu, Mujid Kazimi, "New Friction Factor
Correlation For The MIT Reactor Fuel Elements", Reduced Enrichment Test
and Research Reactors (RERTR) Conference Beijing, China , November 15,
2009
[315] Sung Joong Kim, Yuchih Ko, Linwen Hu, "Loss of Flow Analysis of the
MIT Research Reactor HEULEU Transitional Cores Using RELAP53D",
proceedings of ICAPP '10, San Diego, CA, USA, June 1317,2010 Paper
10224
[316] R.H.S. Winterton, "Where did the DittusBoelter Equation Come From?" Int.
J. Heat Mass Transfer. Vol.41 Nos 45, pp 809810, 1998.
[318] X5 Monte Carlo Team, "MCNP  A General Monte Carlo NParticle
Transport Code, Version 5, Volume I: Overview and Theory"LAUR031987
[319] E.E, Feldman, "Implementation of the Flow Instability Model for the
University of Missouri Reactor (MURR) That is Based on the Bernath
Critical Heat Flux Correlation" ANL/RERTR/TM1 128,July 2011
81
Chapter 4 LSSS Calculation Using Uncertainty Propagation
Technique
As discussed in section 3.2.3, the coefficients in Eq. 311 would vary if different
methodologies were used. In this chapter, the methodology uncertainty propagation
technique is presented to obtain LSSS.
4.1 Introduction
The input parametric distributions used in this methodology were assumed as normal
distributions at this stage due to the insufficient, or the incomplete understanding of
underlying physical mechanism of experimental data. However, more realistic
parametric distributions, i.e. distributions characterizing of skewness and/or kurtosis,
can be used as input distributions in the future if we have prior knowledge on them.
Some key input parameters in Eq. 39, such as heat transfer coefficient h, primary
coolant flow rate w,, hot channel mass flow rate rh and reactor power P are set as
normal distributions and denoted as <h>, <Wp>, <rh >, and <P> as follows. Other
input parameters, which are of insignificant importance on LSSS calculation, are
treated as constant values. These parameters are treated in the same manner as
analytical approach and therefore their notations remain unchanged. As a result, Eq.
39 becomes,
82
<P>F F z <q"z
'*,+
Tu, <W >c, <m>c
<q",(z)>dz+Fsr
<
, <h>
[ [<  h ( 10.463P00234 0
>
(Eq. 41)
Tat 1.8 1082p 1 56
Where < q"h,(z)>= e,,,F,,,FFO(z) and the subscript hs means hot stripe.
Nc AH
Note that EHCFs FH and Fr in Eq. 41 were set to unity since the parametric
uncertainties were taken into account by setting key input parameters as normal
distributions. Consequently, Eq. 41 becomes Eq. 42,
T]] 1 .. (z)
1082p'.
56 463,1.15=0 (Eq. 42)
"' 1.8
Rewrite Eq. 42 by expanding hot channel heat flux in terms of power,
rz <P> <P>
______ j ( NH A Fco,,Fu,,FFO(z))dz F ,,F ,,FFO(z)
< P > F,,,, o Nc AH NC A core f
TW
" <W > c, < rh > cH <h>
<>FF FF
1 . A
NI core fuel , r (z)
 1.8 1082p1.156 ].463,0*4=0 (Eq. 43)
Solving Eq. 43 for LSSS power given input parametric distributions and other single
valued input as discussed in sections 3.2.33.5, LSSS power is obtained in a form of
distribution, representing the accumulative result of input parametric uncertainties.
83
4.2 Monte Carlo Simulation
Monte Carlo simulation is typically used when the model is complex, nonlinear, or
involves several parameters of uncertainties. It is essentially a sampling method with
inputs randomly generated from the probability distributions, which are in good
agreement with the actual data or best represent the current state of knowledge to
simulate the process of sampling from an actual population. The goal of a Monte
Carlo simulation is to simulate realistic situations to make predictions based upon the
given confidence intervals. A simulation can typically involve over 10,000
evaluations of the model so that sufficient data can be gathered. Some variance
reduction techniques, including modifying probability distributions to favor events of
greater interests and splitting/rouletting to change the number of particles in certain
regions, might be applied to approach smaller variances of the prediction results
depending on the situations [4,5]. In this study, none of any variance reduction
techniques were used at this stage.
Monte Carlo simulation was performed on Oracle Crystal Ball to generate and
combine normal distributions for several key input parameters. These normal
distributions are randomly sampled with specified mean values and standard
deviations. The mean values of these input parametric distributions are the nominal
values used in the analytical approaches in Chapter 3. The standard deviations of
these input parametric distributions are either obtained based on the SAR [1] or by
uncertainty propagation based upon their interrelationship with known parametric
distributions, such as the fabrication tolerance distribution of water gap distance.
The quality of random number generator is associated with the credibility of Monte
Carlo simulations. As pointed out in Oracle Crystal Ball User's guide [2], "For no
starting seed value, Crystal Ball takes the value of the number of milliseconds elapsed
since Windows started." That is equivalently to say, if no starting seed value is
specified in the Sampling dialog box, the cycle of random numbers may be found
84
repeated after several billion trials. The iteration formula used in the random number
generator is:
85
4.2.2 Model validation on Crystal Ball
Some simple tests were conduct to assess the reliability of this software. In these
simple tests, some models with exact known analytical solutions were chosen to
validate the results computed by Oracle Crystal Ball.
For example, if two normal distributions are multiplied, each of them with mean 1.00
and the standard deviations of 0.10 and 0.05 respectively, the predicted mean and
standard deviation based on analytical solution are respectively 1.00 and 0.01118.
Using a sample size of 10,000 in Monte Carlo Simulation on Oracle Crystal Ball, the
simulated results are in good agreement with the analytical solution: the averaged
mean of 5 runs is 0.999161 and the averaged standard deviation is 0.112184. The
error predicting mean value and standard deviation in this case is respectively 0.08%
and 0.34%.
Next, using the same analytical model but sample size was increased to 100,000. The
error of the predicted mean value and standard deviation are reduced to 0.02% and
0.07% respectively, which are fairly close to the exact analytical solution. However,
the reduction in variation is obtained at the expense of longer simulation time, which
is proportional to 1/sqrt(N), where N is sample size.
Another more complex model with known analytical solution was tested using sample
size 100,000. The model and the test results are described in Table 41. The error of
predicted mean value and standard deviation for this model is 0.02% and 0.33%,
which is in good agreement with the exact analytical solution. Therefore, for all
analyses presented in this chapter a sampling size of 100,000 is chosen.
86
Table 41 The input description for the model used to validate Oracle Crystal Ball
Standard
Input Parametric Distribution Mean Deviation
A 1.0 0.1
B 1.0 0.2
C 1.0 0.03
D 1.0 0.04
E 1.0 0.05
Table 42 Comparison between analytical solution and results using Crystal Ball
87
4.3 Uncertainty of input parameters for the MITR
As explained earlier in section 4.1, four input parameters including primary coolant
flow rates, heat transfer coefficient (HTC), hot channel mass flow rate (HCMFR) and
power, were set as normal distributions to obtain LSSS power. To generate normal
distributions, mean values and standard deviations were specified on Crystal Ball first,
and then Monte Carlo simulation was performed. How these mean values and
standard deviations were determined is discussed in this section.
The mean value of primary coolant flow rate, 111.38 kg/sec, was converted from 1800
gallon per minute under the assumption that temperature is 55*C, the average coolant
temperature in the core. The primary flow rate 1800 gpm was documented in MITR
II SAR as the LSSS. As a result, 111.38 kg/sec is specified as the mean value for the
primary coolant flow rate distribution.
As for the specification for the distribution's standard deviation, the uncertainty for
flow measurement is 1.05 for enthalpy rise and 1.04 for film temperature rise, as
shown in the EHCFs table in MITRII SAR. These values can be also found in Table
3.3. For conservatism, 1.05 is adopted in the analysis. However, how the subfactors
were obtained in Table 33 was not clearly documented in MITRII SAR. In a study
covering the statistical thermal design procedure of super critical LWR, Yang et al [6]
incorporated three standard deviations (n=3 in Eq. 21 that is recalled in this section)
to obtain the relevant engineering subfactors. Therefore, in this study, it is assumed
that the subfactors in SAR were derived in the same manner and that three standard
deviations were incorporated.
where n is the number of standard deviation that is incorporated into the subfactor,
a is standard deviation and p is the nominal value of parameters.
88
Figure 41 shows the primary coolant flow rate distribution as one of the input
parametric distributions for LSSS calculation. According to the famous 689599.7
rule, about 99.7% of the value falls within three standard deviations of the mean for a
normal distribution. That is, there is 99.7% probability that the primary coolant flow
rate used in LSSS calculation are within the range of 106.33 and 117.55 kg/sec,
reflecting minor flow fluctuation as well as flow measurement uncertainty in real
situation.
89
100.000 Tdils NosneW D*iinuion 99.999 Olsplayed
3,200
_otforC miee Use
~~~2,800
~~
~ 2,400
8,000
Figure 41 Primary coolant flow rate distribution as an input for LSSS calculation
90
4.3.2 Heat Transfer Coefficient
The mean values for the HTC distributions are calculated using Carnavos correlation
as illustrated in section 3.4. The uncertainty for heat transfer coefficient estimation is
1.20, as documented in the EHCFs table in the SAR. However, again, how this sub
factor was obtained is unclear. Therefore, the subfactors in MITRII SAR were
assumed to be three standard deviation values (3a) in this study. That is, the
uncertainty (1Y) of the heat transfer coefficient distribution is one third of 20%, which
is 6.7%.
Figure 42 shows the mean value and standard deviation of the HTC estimated for
node #7 for 189EOC core. Note that, the mean value of HTC for each node is
different due to the different local properties, but the standard deviations for them are
all assumed to be 6.7%.
91
100.00m Tdals Norma Disttion 99,994 Dispied
HTC (W/m2K)
  ~  ~ 3,300
Notfor Comme i/ Use 3,000
2,700
2,400
2100
1,800 .
SdDev 1 3 900o
  0
9,000.00 10,000.00 11,000.00 12,000.00 13,000.00 14,000.00 15,00.00
92
4.3.3 Hot Channel Mass Flow Rate
The mean value of hot channel mass flow rate distribution is 0.2219 kg/sec, which is
obtained from primary coolant flow rate divided by the number of channels, along
with taking into account the core coolant flow factor and flow disparity factor, as
explained earlier in Chapter 3.
The uncertainty for hot channel flow rate is associated to the variations in water gap
distance because of fabrication tolerance. Due to multichannel core design with
constant pressure drop in core region, when water gap distance is smaller than design
specification, less mass flow pass through the channel; whereas water gap distance is
larger than design specification, more mass flow pass through the channel.
The fabrication data for the LEU core cannot be determined at this stage. Therefore,
it is assumed that the fabrication distribution of LEU water gap is the same as that of
HEU's.
The data for the water gap distance of HEU was collected and analyzed in this study.
The data of channel scanning results of HEU was taken from element #243 to element
#364 (fabricated from 1994 to 2008), fabricated by Babcock & Wilcox [7]. Water gap
tiptotip distance was measured at 14 axial levels for each fuel element that total
1708 data entry were analyzed to obtain the fabrication distribution of water gap, as
depicted in Figure 4.3. The abscissa of Figure 4.3 is the deviation from the nominal
distance of water gap distance. The historic standard deviation (3a) for HEU water
gap distance was found as 5.4 mils, according to HEU water gap data analysis in
Figure 4.3. As explained earlier that the fabrication of LEU water gap distance is
assumed to be the same as that of HEU. Consequently, the standard deviation for
LEU water gap distance is 1.8 mil (lo).
The distribution in Figure 4.3, as reexamined by the Oracle Crystal Ball using bestfit
technique, is a normal distribution. Since water gap distance distribution is a normal
distribution, the maximum/minimum possible value of water gap distance is close to
the nominal value of LEU fuel (72 mils) plus/ minus three standard deviations (5.4
mils), which is 77.4 and 66.6 mils respectively.
Given the asfabricated distribution of water gap distance, the distribution of HCMFR
can be obtained by assuming that frictional pressure drop, a major contributor to total
pressure drop, is the same for all flow channels. That is,
93
(AP)NORMN = (AP)OFFNOR (Eq.45)
2
p L PV
p
D, 2 (Eq.46)
and the frictional factor developed empirically for the MITR is [9]
f, = 0.316.(e)0.2s1
(Eq.47)
Next, combining Eq. 45  47, a conclusion can be made that the pressure drop
where A represents the crosssectional flow area of a channel. Therefore, Eq. 45 can
be expressed as
02
(i
75
 A'" De1.25 L
0 2
P1)NOMVAL _ l.75 1'.7s. es L*

P')OFFNRMAL (Eq.48)
For simplicity, an assumption is made that the pressure drop is independent of Tvg ;
that is, pressure drop are assumed independent of the property terms listed in Eq. 48.
In addition, flow area A and hydraulic diameter De are proportional to water gap
94
Since the distribution for water gap distance is known (99.7% of water gap distance
value falls within the range 72 + 5.4 mils, as discussed earlier), the distribution of
HCMFR can be obtained using Eq. 410.
95
DatAnalysis: DatSeries
004
63
64
003 6
36
32
i
006 .
24
201
Is
12
3
Figure 43 istorical offnormal water gap distance value for HEU collecting from
1994 to 2008
HCMFR (kg/sec)
2,700
2,400
5 21800
1.500
2900
 0
0.1900 0.2000 02100 .0.2200 0.2300 0.2400 0.26W0 0.2600
Mean 0.2219 Std. Dev. 0.0095
96
4.3.4 Power
As shown in Eq. 311 and Figure 311, LSSS is specified such that average outlet
temperature can be obtained if reactor power is given, or alternatively, a reactor power
(LSSS power) can be obtained if average outlet temperature is given. If the former
approach is used, reactor power is one of the inputs and therefore a power distribution
representative of power measurement uncertainty should be used as one of the input
parametric distributions. If the latter approach is used, reactor power is first solved
from the LSSS equation (Eq. 43, but <P> = P), and then the overall uncertainty of
power is obtained by propagating the power measurement uncertainty with the
uncertainties obtained in the previous step. The latter approach is used in this study
since the goal is to calculate the maximum allowable reactor given average outlet
temperature is 60*C.
The standard deviation for power is determined as 3.73%, which is the statistical
combination result (Eq. 22) of the uncertainties of reactor power measurement and
power density measurement/calculation listed in the EHCFs table (Table 3.3). Again,
the uncertainties of reactor power measurement and power density
measurement/calculation listed in the EHCFs table are assumed referring to three
standard deviations. Therefore, the one standard deviation for reactor power
measurement and power density measurement/calculation is respectively 1.66% and
3.33%. Combine these two uncertainties using root mean square (Eq. 22), the one
standard deviation for power is then obtained as 3.73%.
97
4.4 Results
In Chapter 3, it has been demonstrated that the analytical LSSS calculated based on
hot stripe factor (2.12) is more limiting than that based on radial power peaking factor
(1.76), and for conservatism the former is adopted. Therefore, the methodology
proposed in this Chapter is also focus on calculating LSSS based on hot stripe factor.
Since some input parameters used in LSSS calculation are set as distributions, LSSS
power obtained based on these parameters is also in a form of distribution. For
conservatism, LSSS power at (mean  3a) value is taken as the reference value for
this methodology. Figure 45 shows the (mean  3o) value of LSSS power for each
node. Like in Chapter 3 node #7 predicts the most limiting LSSS power in this
technique, as depicted in Figure 45.
When outlet temperature is specified as 60*C, the LSSS power at (mean  3a) value
for node #7 is 9.1 MW, as depicted in Figure 45. This LSSS power is taken as the
reference value for the uncertainty propagation technique.
The data fitting performed on Crystal Ball indicates that the LSSS power distribution
obtained from Eq. 43 is also a normal distribution. Since this LSSS power is a
normal distribution, there is about 99.85% possibility that the actual LSSS power
limitation is higher than the (mean  3a) LSSS power limitation 9.1MW. That is to
say, the probability of ONB occurrence is roughly 0.15% when operating power is
9.1MW.
98
16
14 *
+ node 1
U node 2
10 A node 3
x node 4
8
node 5
6 0 node 6
4
node 7
 node 8
2 +node9
 node 10
0
50 55 60 65 70 75 80
Tout(*C)
Figure 45 LSSS power of each node using uncertainty propagation technique
99
4.5 Summary
100
Table 43 Input parametric distributions used in uncertainty propagation
methodology
Normally Uncertainty
distributed Mean value Specification Source
parameters
1.00+ See ECHF table. Sub
Primary 111.938 kg/sec (converted 0.0167(1a) factors documented in
coolant flow from 1800 gpm assuming T Mean+3 or MITR SAR is assumed
rate <Wp> =55 *C) =05 that 3 sigma was
incorporated
0.2219 kg/sec, calculated This distribution was
from 1.00+ derived from water gap
MFR in hot [(111.938/432)*0.921*0.93], 0.0426(10) distance distribution
where core coolant flow where mean value is
channel <m> factor =0.921, flow disparity Mea+28 72.0 mil and (mean +
factor = 0.93, and number of 3y) value are 66.6 and
channels =432 77.4 mils
the calculated heat transfer 1.00+ See ECHFs table. Sub
Heat transfer coefficient ranges from 0.067(0) factors documented in
coefficient <h> 10094 to 12914 based on Mean+3 a MITR SAR is assumed
local properties for each 1.20 that 3 sigma was
node. ~1'0incorporated
The one standard
deviation for reactor
power measurement and
power density
Mean value of LSSS power 1.00 0.0373 measurement/calculation
LSSS Power is computed that outlet (lo) is respectively 1.66%
coolant temperature and Mean+3 . and 3.33%. Combine
other parameters are Mean+3 these two uncertainties
specified. using root mean square
(Eq. 22), the one
standard deviation for
power is then obtained
J _as 3.73%.
101
16
14
~12
10  Using EHCFs
8
+BestEstimate
6
4_UUncertainty
Propagation
2
0
50 60 70 80
Tout(*C)
102
Table 44 Summary for LSSS power obtained using different methodology
LSSS Power
Methodology Description (When Tout is
60"C)
EHCFs were set as 1.0. The input
parameters were set at their
nominal values and as single values
Bestestimate to obtain LSSS power in an 11.1 MW
analytical manner. This dataset
shows the results of node #7, the
most limiting among the nodes of
189EOC.
EHCFs were set as 1.0. Some key
input parameters were set as normal
distributions to obtain a LSSS
power distribution. This dataset
Uncertainty shows the LSSS power at (mean 9.1 MW
propagation 3*S.T.D.) value that are calculated
by Crystal Ball based on node #7,
the most limiting among the nodes
of 189EOC.
EHCFs were set as what they were
documented in the SAR. The input
parameters were set at their
nominal values and as single values
EHCFs to obtain LSSS power in an 8.3 MW
analytical manner. This dataset
shows the results of node #7, the
most limiting among the nodes of
189EOC.
103
References
[41] MIT Nuclear Reactor Laboratory, "Safety Analysis Report for the MIT
Research Reactor," MITNRL 1102, August, (2011)
[42] EPM Information Development Team, "Oracle Crystal Ball User's Guide
Fusion Edition" Release 11.1.1.3.00, 1988
[43] N. Metropolis, , "The beginning of the Monte Carlo Method", Los Alamos
Science (1987 Special Issue): ppl25130, 1987
[49] S. Wong, L. W. Hu, M. Kazimi, "New Friction Factor Correlation For The
MIT Reactor Fuel Elements", Reduced Enrichment Test and Research
Reactors (RERTR) Conference Beijing, China, November 15, (2009)
104
Chapter 5 Sensitivity Study of LSSS
As explained earlier in section 3.3, flow disparity factor is defined as the ratio of
(minimum flow/average flow) for the coolant channels within a fuel element, which
was taken as 0.864 in the previous version of MITRII SAR[1]. This factor is updated
and used in LSSS calculations based on discussions with experts from MITNRL and
ANL.
As documented in the SAR citing experimental results from the startup test [2], 0.864
is the multiplicative result of the other two factors, 0.93 and 0.929. The first factor
0.93 represents the minimum flow through a fuel element is 93% of the average core
flow rate. The second factor 0.929 represents the ratio of the minimum channel flow
rate to the average channel flow rate within a fuel element is 0.929. As discussed
earlier, it is suggested to take the factor 0.929 out from the flow disparity factor to
avoid doublecount. Therefore, flow disparity factor was changed from 0.864 to 0.93,
and the latter was used in this study for LSSS calculation.
Table 51 summarizes the calculated LSSS before and after the updating of flow
disparity factor. When the outlet temperature is fixed at 60'C, the update in flow
disparity factor results in approximately 0.4 MW increase in LSSS power. These
results are reasonable. As flow disparity factor becomes larger, it means flow
distribution moves towards closer to uniform distribution, and therefore there is more
coolant for the minimal flow channel. This fact surely gives more room to the upper
bond of safety limits LSSS power, as can be seen in Table 51.
105
Table 51 Changes in LSSS due to the change in flow disparity factor
106
5.2 Coolant Density Estimation
One of the LSSS criterions is that primary flow is at least 1,800 gallons per minute
(gpm) when the primary coolant pumps on both loops are active. As shown earlier in
Chapter 3, primary flow rate (kg/sec) and hot channel mass flow rate (kg/sec) are
involved in LSSS calculation. Coolant density is required when converting gallon per
minute into kilogram per second.
Table 52 shows how LSSS power would change with respect to the change in coolant
density at channel inlet when outlet temperature is fixed at 60*C, which is also one of
the LSSS criterions. The results show that the change in LSSS power due to the
change in coolant density is of insignificant importance. This effect is negligible
when coolant average temperature ranges from 40*C to 60*C with changes in LSSS is
within 0.03 MW.
107
Table 52 The resulting change in LSSS power when outlet temperature is fixed
at 60cC
108
5.3 Heat Transfer Coefficient (HTC)
Heat transfer coefficient plays an important role in LSSS calculation for its impact on
cladding temperature. Carnavos correlation was used in this study to estimate HTC
due to its conservatism as explained earlier in section 3.4. The HTC calculated for
node #7 at core 189EOC is 12436 W/m 2 K.
Table 53 shows how the LSSS power of node #7 at core 189EOC would change with
respect to the change in HTC when outlet temperature is fixed at 60*C. Figure 51
depicts the sensitivity of LSSS power on HTC. As expected, higher HTC means
better heat transfer, and therefore LSSS power, which plays similar role as safety limit
does, is expected to be higher. This trend can be observed in Table 53 and Figure 5
1 that overestimation of HTC results in larger LSSS power. The results show that
LSSS power is sensitive to HTC so that the error of LSSS power is roughly within
10 %, or equivalently 0.9MW, given acceptable estimation error in HTC, say
within + 16%.
109
Table 53 The change in LSSS power with respect to the change in HTC when outlet
temperature is fixed at 60'C
HTC Change in HTC Analytical LSSS Power Resulting Change in LSSS Power
2
[W/(m K)] 1%] [MW] [%]
5436 56.3% 4.58 45.2%
6436 48.2% 5.24 37.4%
7436 40.2% 5.85 30.1%
8436 32.2% 6.42 23.2%
9436 24.1% 6.96 16.9%
10436 16.1% 7.46 10.9%
11436 8.0% 7.93 5.3%
12436 0.0% 8.36 0.0%
13436 8.0% 8.78 5.0%
14436 16.1% 9.17 9.6%
15436 24.1% 9.54 14.1%
16436 32.2% 9.89 18.2%
17436 40.2% 10.23 22.2%
18436 48.2% 10.54 26.0%
19436 56.3% 10.84 29.5%
60%
40%
0 20%
20%
0
1U
40Y%
60%
Error in HTC
Figure 51 Sensitivity of LSSS power on HTC
110
5.3.2 Effect of Variation in Viscosity for Heat Transfer Correlation Calculation
* Entrance length for turbulent flow is z/D = 2540. The diameter of the flow channels in MITR is
2e3 m, therefore the entrance length is 0.05m0.08m, which is about onetenth of fuel rod length.
111
Table 54 Summary for correlations typically used to compute HTC in research
reactors
112
Since some HTC correlations take properties from wall temperatures, or from the
average of bulk and wall temperatures, iterations are required to compute HTCs based
on wall temperatures using the equation below,
Where p is viscosity and T is coolant temperature. The average mean absolute error
of Eq. 52 is within 1% in the range shown in Figure 52 comparing to the viscosities
computed using the International Association for the Properties of Water and Steam
(IAPWS) 1995 formulation [7], which has been widely used for general and scientific
purposes. Table 55 summarizes the HTCs and cladding temperatures at node #7 of
189EOC core using Eq. 51 in conjunction with different HTC correlations.
113
0.0008
0.0007
0.0006
M 0.0005
0.0004
 Equation 52
0.0003 ''LAPWS Formulation 1995
0.0002
0.0001
0
0 20 40 60 80 100 120
Figure 52 Comparison between IAPWS 1995 Formulation and simplified viscosity
formula
Table 55 HTCs and outlet clad temperatures computed using different correlations
114
As can be observed in Table 55, after taking into account the effect of viscosity
variation in the radial direction, the HTC computed from modified DB correlation is
6% larger than the original DB correlation. This 6% increase in HTC, based on
Figure 51, could result in roughly 5% in LSSS power, which is of minor importance.
As demonstrated, Carnavos correlation predicts the most limiting HTC and cladding
temperature, and therefore is adopted in LSSS analyses. However, Carnavos
correlation does not have a counterpart incorporating the varying viscosity in radial
direction as the modified DB correlation does. Consequently, how the varying
viscosity in radial direction could affect the HTC computed based on Carnavos
correlation is unknown at this stage, and it is expected that the ongoing experiment
conducted at the MITNRL can bridge this gap afterwards.
115
5.4 Local Fluid Temperature
In previous analyses we realized that LSSS is sensitive to the change in HTC. HTC
was calculated using Carnavos correlation in this study. Among parameters in
Carnavos correlation, viscosity is the one that is strongly dependent on local fluid
temperature. The purpose of this section is to investigate how the measurement error
of fluid temperature could affect the change in HTC and LSSS power.
As can be seen in Table 56, measurement error in local temperature has insignificant
impact on HTC, resulting in roughly the same magnitude of change in LSSS power.
As can be seen, 5 *C measurement error in local fluid temperature results in +2.5%
change in HTC, resulting in about +2% change in LSSS power. Consequently, it is
concluded that the change in LSSS power is within 0.3 MW if fluid temperature
measurement error is within 5*C.
116
Table 56 The change in LSSS power with respect to the change in local temperature
(viscosity) when outlet temperature is fixed at 60*C
Change in
Local Temperature HTC Change in HTC LSSS Power LSSS
[*C] 2
[W/(m K)J [%] [MW] Power
[%]
65 11409 5.7% 7.91 3.7%
66 11475 5.1% 7.94 3.4%
67 11540 4.6% 7.97 3.0%
68 11604 4.0% 8.00 2.7%
69 11668 3.5% 8.03 2.3%
70 11731 3.0% 8.06 2.0%
71 11794 2.5% 8.09 1.6%
72 11856 2.0% 8.12 1.3%
73 11918 1.5% 8.14 1.0%
74 11979 0.9% 8.17 0.6%
75 12040 0.4% 8.20 0.3%
76 12100 0.0% 8.22 0.0%
77 12160 0.5% 8.25 0.3%
78 12219 0.9% 8.27 0.6%
79 12277 1.4% 8.30 0.9%
80 12336 1.9% 8.32 1.2%
81 12393 2.4% 8.35 1.6%
82 12451 2.9% 8.37 1.9%
83 12507 3.3% 8.40 2.1%
84 12564 3.8% 8.42 2.4%
85 12620 4.3% 8.45 2.7%
117
5.5 Summary
The sensitivity analyses in this chapter give out a general idea how LSSS power could
change due to the uncertainties in parameters/properties, or due to the update of flow
disparity factor. Conclusions for the sensitivity analyses are summarized below.
(1) The update of flow disparity factor results in roughly 0.4 MW increase in
LSSS power. This updated disparity factor was used in this study for
analytical, bestestimate, and uncertainty propagation LSSS power
calculations.
(2) The estimation uncertainty in coolant density at channel inlet has been
demonstrated to be negligible for LSSS power calculation.
(3) HTC is important for LSSS calculation. LSSS power is sensitive to HTC so
that the error of LSSS power is roughly within 10 %, or equivalently
+0.9MW, given acceptable estimation error in HTC, say within + 16%.
118
References
[51] MIT Nuclear Reactor Laboratory, "Safety Analysis Report for the MIT
Research Reactor," MITNRL 1102, August, (2011)
[55] S. Kakag, R.K. Shah and W. Aung, Handbook of SinglePhase Heat Transfer,
A WileyInterscience Publications (1987)
[57] W. Wagnera. and A. PruBb., "The IAPWS Formulation 1995 for the
Thermodynamic Properties of Ordinary Water Substance for General and
Scientific Use", J. Phys. Chem. Ref. Data, Vol. 31, No. 2, 2002
119
Chapter 6 Safety Limit Calculation
6.1 Introduction
Safety limits are established to secure the integrity of the fuel cladding, that is to say,
to avoid fuel overheating. Critical heat flux (CHF) is normally used as the criterion
for fuel overheating. However, considering the MITR's flow channel is a
multichannel design, flow instabilities could possibly occur prior to reaching CHF
limitations. When onset of flow instability (OFI) occurs, it would have the effect of
reduced flow rate due to flow instability, hence lowering the CHF in the hot channel.
In the safety limit calculations, both CHF and OF are calculated and the lower one is
adopted as the safety limit for conservatism.
Since coolant in the MITR flows through parallel flow channels, Ledinegg instability
[1], or flow excursion instability could be a particular concern in such a parallel flow
path design, especially for research reactors having narrow flow channels. Figure 61
shows a demand curve [2, 3] describing the total pressure loss in a heated channel
versus the mass flow rate. As depicted in Figure 61, before entering the minimum
point B, which is defined as OFI (the onset of flow instability), the pressure drop
through the channel decreases as mass flow rate decreases. Beyond this point, as
mass flow rate decreases, pressure loss continually increases until fluid becomes
single vapor phase.
Operating between point B and point C in Figure 61 is considered unstable. The
trend in the region AB is reversed at point B because plenty of vapor generation is
there mixing with the liquid core flow. At this point, a small negative perturbation in
mass flow rate at point B could result in the shifting to B' ,which leads to CHF
because the increase in pressure drop in single channel results in flow diversion to
other channels, or even to bypass flow path. As a result, substantial loss of channel
cooling could lead to burnout. The bubble blockage in flow channels is a dominant
cause for premature CHF in low pressure system, such as lowpressure research
reactors due to flow instability.
CHF is essentially a boiling process transiting from nucleate boiling to film boiling.
Typically this transition is accompanied by a large increase in temperature on the
surface due to the sudden decrease in heat transfer performance. This localized
overheating could cause failure of equipment, and therefore it is important to
120
investigate when such a mechanism occurs, if there is any indication prior to its
occurrence, and if possible, any approaches to take to improve heat transfer during
such an undesirable transition.
Due to the extremely complex nature of twophase flow with heat transfer, in spite
there have been numerous experiments conduct and theoretical models developed, the
mechanism of CHF is still not fully understood. For nuclear industry, design of
reactors therefore requires sufficient margin with regard to CHF to minimize the
possibility of cladding failure due to overheating.
121
constant q'(z) liquid
Constant h y,
gas
A
I9
IL
a L q'(z)
TOE
L / B
D
two phase single phase
O'region
01
liquid
0
122
6.2 Onset of Flow Instability
Under some circumstances, for safety concerns a limit is set for the maximum
possible channel power due to a phenomenon called excursiveflow instability or
Ledinegg instability if coolant is likely undergoing a transition from single phase to
twophase conditions when flowing through heated channels.
6.2.1 Introduction
Research reactor is different from power reactor given that research reactor is
primarily designed to generate neutrons for research purposes, instead of achieving
high power conversion efficiency. For this reason, a research reactor usually has
higher power density to attain relatively high neutron flux densities. High power
density is achieved by the compact core structure design with flow channels of small
hydraulic diameters.
For such a narrow channel design, the onset of significant void (OSV), the onset of
flow instability (OFI) and other relevant phenomena could occur. As can be seen in
Figure 62 [3], before entering to region I, subcooling prevails in that the flow is
mainly liquid singlephase flow. If heat is continuously applied, the heating surface
temperature increases, eventually exceeding the saturation temperature by a certain
extent, the incipient boiling then occurs, which is also known as ONB, referring to the
incipience of bubble nucleation on the heating surface. At this stage, bubbles do not
detach from the heating surface but grow and collapse before their contact with the
surrounding subcooled liquid. That is, bubbles are restricted to the immediate vicinity
of the heating surface (wall voidage).
In region II, bubbles begin to detach from the heating surface, which is defined as the
onset of significant void (OSV). At this stage, bubbles grow, detach from the heating
surface, and coexist with core flow, but are condensed when encountering the colder
zone in subcooled liquid core flow. The local increase in void content is important to
nuclear reactors because it influences reactivity via the change in neutron moderation
effectiveness, therefore changing the dynamic behavior of reactors.
The point ONB, OSV in Figure 62 can be referred in the pressure dropmass flux
curve in Figure 63. OFI refers to the minimum point of the curve in Figure 63 [4].
As explained earlier, operating beyond the point OFI is considered unstable because
the trend beyond this point is reversed due to mixing of vapor generation with the
123
liquid core flow, therefore resulting in flow's diversion to other channels leading to
burnout.
Al
FEGION I REGION 11
a
OWALL DETACHED
VIOIO$4E VODAGE
IIT THERMODYNAMIC
I ff.~ EOUIUSRNJUM
VOID PROFILE
ACTUAL
I g PROFILE
Figure 62 Void fraction variation along a uniformly heated channel [3]
Superbeated
singlephase Wall Subcooled
vapor Flow Bulk boiling voidage singlephase liquid flow
AP
G
Figure 63 Schematic diagram of the demand curve for a heated channel with constant
heating rate [4]
124
6.2.2 Calculation of OFI for the MITR (Analytical Approach)
The point OFI was determined using a steadystate energy conservation equation,
which was proposed by J.E. Kowlaski [5],
FI= (Eq.61)
R c,(T.,  TO
where
rhOFI is the channel mass flow rate when OFT occurs (kg/sec)
Q is the channel power (W)
R is the channel outlet subcooling ratio, (T., Tm)I(Tw,  T)
The channel outlet subcooling ratio, R, can be determined from one of the following
three equations [5, 6],
L
R= for 30 < L<300 [6], (Eq. 62)
1+25 D, D,
Lh
R = 0.21 In D 0.258, for 70 < 1L< 250 [7], (Eq. 63)
D,) D,6
125
Therefore, we should apply Eq. 62 to calculate R. The parameters used in Eq. 61 to
calculate OFI is summarized in Table 61.
Substitute the parameters of Table 61 into Eq. 61 to calculate how much channel
power/reactor operating power it takes to induce OFI. The conversion from operating
power to channel power is also summarized in Table 61. The result shows that 12.46
MW is predicted to induce OFI.
126
Table 61 Parameters used to calculate OFI
Saturation temperature
Saturation temperature, Tat 1070C corresponds to pressure
is 1.3 bar
0
Channel inlet temperature, Ti, 42 C Assumption
Hot channel mass flow rate rh 0.2219 kg/sec See section 3.3
power P (MW) to channel power =( 432 )0.965 0.94 .2.12 System parameters
=432
Q (W)Q(W) =(4.451e3).P
127
6.2.3 Calculation of OFT for the MITR (Uncertainty Propagation Technique)
The results obtained in the previous section were obtained in a manner that every
parameter used is at their nominal value, without taking into account the parametric
uncertainty. In this section, the uncertainty of water gap distance, which is due to
fabrication tolerance, is incorporated to calculate OFI power.
As documented, the historical fabrication tolerance for HEU water gap distance is 5.4
mil (3a value). The fabrication tolerance of LEU is assumed to be the same as that of
HEU. Since water gap distance is a distribution, the parameters associated with water
gap distance, such as channel subcooling ratio R and channel mass flow rate, are also
distributions based upon the relation between water gap distance.
The equivalent diameter of the coolant channel De is a function of water gap distance.
Therefore, the channel outlet subcooling ratio R also changes with water gap distance.
In addition, as demonstrated earlier in Eq. 410, channel mass flow rate is associated
with water gap distance so that water gap of different dimension results in different
mass flow rate. How these parameters correlated with water gap distance and its
relevant geometry dimension are summarized in Table 62.
Lh
The uncertainty for reactor power is also taken into account such that 3.73% is used in
the OFT analyses, which is similar to the LSSS analyses.
According to the Monte Carlo simulation on Crystal Ball, the reactor power at (mean
3a) value inducing OFI is 10.41 MW. Recall that the most limiting LSSS power
using uncertainty propagation is 9.1 MW, as concluded in Chapter 4. These imply
that the margin between OFT and operating power is sufficient given the fact that the
proposed relicensing power of the MITR is 7 MW.
128
Table 62 Parameters set in a form of distribution for OFI/CHF calculation
129
6.3 Critical Heat flux
6.3.1 Introduction
Sudo et al. [8] proposed a CHF correlation scheme for research reactors using flat
platetype fuel, as shown in Figure 64. For the MITR, forced convection is the heat
transfer mechanism during normal operation that coolant flows upward through the
vertical rectangular channels.
The CHF correlation used for forced upflow in MITR is a modified version proposed
by Sudo taking into account the effect of channel outlet subcooling [8], that was
applied in high mass flux region as depicted in Figure 64,
. ,g.61 5000 .
qCHF =0.005 G .(1+ .ATsuo) (Eq. 66)
G*
G
G Agp (P,  Pg)
C
ATs*u,o = T,,),
a(T,,
hfg
F 10.5
= ( 0j (characteristic length)
g ( p,  p, )
g is the acceleration due to gravity,
pf is fluid density,
130
p8 is gas density,
Regarding CHF for natural convection, Sudo suggested the minimum CHF for upflow
forced convection that corresponds to a condition that the flow become stagnant, the
low mass flux region in Figure 64, is used as the CHF for natural convection. The
flow is under countercurrent flow limitation (CCFL) while the water moving
downward coexists with the steam/bubbles moving upward. CHF is closely related to
CCFL, and the correlation used for such a situation, which also applies to natural
convection, is
.A
qCHF=0.7  vWIA (Eq.67)
Am [+(, /Pf )o.2s
131
Low mans ffua Medium mass flux IHigh mass flux
ase of
A'TUBO
qICIIF
()
GG )
Figure 64 CHF correlation scheme proposed for research reactors using platetype
fuel (adopted from [8])
132
6.3.2 Root Mean Square Error of Sudo's CHF Correlation
The modified CHF correlation (Eq. 66) is a bestfit result from experimental data.
Figure 65 [8] compares CHF's prediction using Eq. 66 and the experiment results.
Both upflow and downflow experiment results are shown in this figure while the
coolant in the core of the MITR flows upwards. As can be seen in Figure 65, Sudo's
CHF prediction allows the root mean square error (RMSR) of 33 percent to the lower
limits of the experimental data.
Indeed, some experimental data are 33% outside the upper limits of Sudo's CHF
prediction, as shown in Figure 65, but this fact is of insignificant importance to the
safety limit calculation because safety analyses focus on the lower limit of CHF
prediction, instead of the upper limit. Therefore, 33% RMSR is adopted as the basis
for CHF prediction uncertainty for subsequent analyses, as shown in section 6.3.4.
133
100,
100 o UPFLOW .
*DOWNFLOW o
6
+33% 33%
000
E
10 o 
LL0 10
10 0
q* (
Figure 65 Comparison between Sudo CHF prediction and experiment result[8]
134
6.3.3 Calculation of CHF for the MITR (Analytical Approach)
Eq. 66 is used to estimate the forced convection CHF for the MITR. Table 63
summarizes the parameters used in CHF calculation for the MITR. The CHF on the
hottest spot of the MITR is calculated in this section. The hottest spot is node #5 of
189EOC core, which is close to the midplane of the heated region, as can be seen in
Figure 3.2. Therefore, the properties used for CHF calculation is based on the
estimated local condition of this node. The estimated pressure for this node is 1.49
bar and saturation temperature is 111. PC.
The calculated CHF for node#5 is 3.22 x10 6 W/m 2 , which is equivalent to 70 MW
reactor power with a hot stripe factor of 2.12. However, as shown earlier in Sudo's
study [8] that 33% reduction should be made to reflect the uncertainty in Sudo's CHF
correlation. As a result, a conservative estimate of forced convection CHF using
analytical approach is taken as 2.16 x106 W/m.
In contrast, the CHF calculated for the natural convection mode using Eq. 67 is
2.307 x 10 4 W/m 2, which corresponds to a reactor power of 504 kW with a hot stripe
factor of 2.12. However, upon taking into account the uncertainties associated with
reactor power measurement (5%) and power density analysis (10%) as documented in
the EHCFs table, the reactor power corresponding to a dryout condition becomes 448
kW. A reactor power of 400 kW is conservatively adopted as the safety limit as
proposed in this study.
135
Table 63 Parameters used in CHF calculation for the MITR
136
6.3.4 Calculation of CHF for the MITR (Uncertainty Propagation Technique)
The results obtained in the previous section were obtained in a manner that every
parameter used is at their nominal value, without taking into account the parametric
uncertainty. In this section, the uncertainty of water gap distance, which is due to
fabrication tolerance, along with the correlation error stated in Sudo's study [8], are
incorporated to calculate CHF power.
and q*HF in Eq. 68 are in brackets < >, representing that they are in a form of
distribution as input to calculate CHF. How these parameters relate to water gap
fabrication tolerance can be found in Table 62.
whr=<T Cf (T<O,>
P FO,,,, F , F,.Fs
in N .c,<ih >
<G> <cm> 1
.
2gp,(p,p,) <A> Vgp,(pfpg)
As documented, the historical fabrication tolerance for HEU water gap distance is 5.4
mil (3cy value). The fabrication tolerance of LEU is assumed to be the same as that of
HEU. Since water gap distance is a distribution, the parameters which are directly
linked to water gap distance, such as hot channel mass flow rate and mass flux are
also distributions. Since mass flux changes with water gap distance, coolant outlet
temperature also changes. The latter can be verified by energy balance. How these
parameters correlate with the change in water gap distance is summarized in Table 64.
Note that HCMFR is used as the mass flow rate for hot stripe analyses in this study.
137
limit of Sudo's correlation, assuming 33% corresponds to 3a such that 11% error
corresponds to la. This uncertainty was specified as STD when generating the
normal distribution representing CHF distribution.
The CHF calculated at (mean3a) value is 2.14 x 106 W/m 2, which is equivalent to
47 MW reactor power with a hot stripe factor of 2.12.
138
Table 64 Parameters that were set in a form of distribution for CHF calculation
4.3.3.
A= channel width x ( water gap + 2 x fin A=5.8623E02 x ( 6 +2 x 2.5400E
Actual flow area, A height)  number of fins per channel x 04)  220 x 6.4516E08
single fin area
Hot channel mass G =(HCMFR)/(actual flow area) Combination result of the above
flux, G two
Outlet temperature, (Channel power/(HCMFR* Cp)) +
To'tChannel outlet temperature Ti., where HCMFR is correlated to
""_ _ _S as shown above
Since all data falls within 33% of
the lower limit of Sudo's
correlation, assuming 33%
CHF, q* Critical heat flux predicted using Sudo's corresponds to 3a so that 11%
correlation error corresponds to la is used as
the STD for the normal
distribution representing the error
of CHF
139
6.4 Comparison between OFI and CHF
As summarized in Table 65 that the heat flux inducing OFI is about 12 times smaller
than that of CHF. Note that CHF is a localized phenomenon while OFI is a universal
effect within a flow channel. That is, the CHF calculated in Table 65 is specifically
for the hottest point of 189EOC and the heat flux inducing OH is calculated for the
whole channel. To take into account this difference, the axial peaking factor of
189EOC, 1.26, should be incorporated into the analysis. As a result, after taking into
account the axial power peaking factor, the heat flux inducing OH is roughly 9.5
times smaller than CHF. Which means, for the transients characterized by ascending
heat flux due to loss of cooling or unexpected power excursion, OH occurs prior to
CHF. Therefore, OH is adopted as the safety limit for the proposed LEU design in
this study for conservatism.
140
Table 65 Comparison between OH and CHF
141
6.5 Summary
Both CHF and OFI are analyzed in this study to decide which one should be adopted
as the safety limit, which is set to avoid the overheating of fuel cladding, for the
proposed LEU design. These two criterions are calculated respectively using
analytical approach and uncertainty propagation methodology. The CHF and OFI at
(mean3a) value using uncertainty propagation methodology are adopted for
conservatism.
The CHF calculated at (mean3a) value is 2.14 x 106 W/m 2, which is equivalent to
47 MW reactor power with a hot stripe factor of 2.12 while the OFT power at (mean
3a) value is 10.4 MW. That is, OFT is more limiting than CHF and therefore is
adopted as the safety limit. The margin between the LSSS power using uncertainty
propagation 9.1MW and the calculated OFT power 10.4 MW is 1.3 MW, which is
sufficiently enough from the perspective of safety.
142
References
[61] M. Ledinegg, "Instability of flow during natural and forced circulation." Die
Wirme (translation in USAECtr1861) 614: 891898, 1938.
[65] J.E. Kowlaski, et al., "Onset of Nucleate Boiling and Significant Void On
Finned Surfaces", ASME, FED 99:405411, 1990.
[66] R. H. Whittle and R. Forgan, "A Correlation for the Minima in the Pressure
Drop Versus Flow Rate Curves for SubCooled Water Flowing in Narrow
Heated Channel," Nuclear Engineering and Design, Vol. 6, 1967.
[67] T. Dougherty, et. al., Boiling Channel Flow Instability, ASMEJSME Thermal
Engineering Proceedings, Vol. 2, ASME, 1991.
[68] Y Sudo and M. Kaminaga, "A New CHF Correlation Scheme Proposed for
Vertical Rectangular Channels Heated From Both Sides in Nuclear Research
Reactors", Journal of Heat Transfer, Vol.115 pp. 426434, May 1993
143
Chapter 7 Natural Convection LSSS Calculation
7.1 Introduction
The MITR is designed to be passively safe such that natural circulation and anti
siphon valves (NCVs and ASVs) come into play whenever forced convection, the
main heat removal mechanism during normal operation, is not sufficient to remove
heat from the core region during transients. The antisiphon valves make natural
circulation possible. Figure 14 [1] illustrates the flow path for natural circulation
comparing the flow path for forced convection during normal operation, as depicted
in Figure 15 [1]. Four NCVs were located at the bottom of the core tank while two
ASVs were installed inside the core tank at the same elevation of the primary inlet
pipe.
Both the NCVs and ASVs are balltype check valves. During normal operation,
coolant pressure forces the ball to the top of the shaft, blocks the top aperture of the
valves and therefore valves are closed. However, when primary flow rate decreases to
certain level, the ball falls down since under such a condition coolant pressure is not
enough to sustain the weight of the ball. As a result, valves are open enabling natural
circulation.
As depicted in Figure 14 [1], the hot coolant leaving the core rises within the core
tank, mixes with cold coolant in the outlet plenum, reverses, flows through the NCVs
and/or ANVs, and finally flows back through the core region completing the natural
circulation loop.
144
7.2 Introduction to RELAP5/Mod3.3
This code was developed for the U.S. Nuclear Regulatory Commission (NRC) for use
in rulemaking, licensing audit calculations, evaluation of operator guidelines, and as a
basis for a nuclear plant analyzer. In cooperation with several countries and domestic
organizations that were members of the International Code Assessment and
Applications Program (ICAP) and its successor organization, Code Applications and
Maintenance Program (CAMP), the NRC developed RELAP5/MOD3, a code version
suitable for the analysis of all transients and postulated accidents in LWR systems,
including both large and smallbreak lossofcoolant accidents (LOCAs) as well as the
full range of operational transients.
The principal new feature of the RELAP5 series was the use of a twofluid, non
homogeneous, nonequilibrium, hydrodynamic model for transient simulation of the
twophase system behavior.
Note that the coolant channel of MITR is finned. However, RELAP5/MOD 3.3 does
not have heat transfer package specifically for finned hydraulic geometry. The
alternative approach taken in this study is, treating finned channel as smooth channel
and adjust some of the geometry parameters accordingly (material thermal properties
were also changed accordingly).
145
7.3 RELAP5 Input Deck for the MITR
The RELAP5 steadystate input deck for the MITR was prepared by S. J. Kim, a
senior researcher at MITNRL [3]. Several minor edits were made in the input deck in
this study to calculate LSSS during natural convection.
The flow channels of the proposed LEU core are simulated using pipe components.
These pipes are axially divided into ten nodes. Node 1 is the node at the bottom of
the channels and node 10 is the node at the top end. The flow channel receiving the
highest power in the LEU core is defined as the hot channel in RELAP5 input
hydrodynamic model while other flow channels are defined as average channels. Hot
channel model is constructed in accordance with the actual dimension of a single flow
channel in MITR while other flow channels were lumped into a large one simulated as
average channel, as depicted in Figure 71 [3].
Heat structure was attached to the hydrodynamic model to simulate the power
generated in fueled region. The power distribution within the models, both in axial
direction and radial direction, were set in accordance with the power distribution in
MITR. The power generated within the hot channel model is featured by hot stripe
peaking factor 2.12 and both the axial power distribution of hot channel and average
model is set using 189EOC core.
Four natural convection valves and two antisiphon valves were modeled for natural
convective heat removal, which is the heat transfer mechanism during LOF. In the
input deck, these valves were respectively lumped into one for simplicity. This
procedure might not be able to truly reflect the local flow conditions, especially in the
vicinity of the valves, but from the perspective of bulk flow, the result is agreeable.
Under normal conditions, the MITR operates with a primary coolant flow rate of
146
2,000 gpm. If primary coolant flow rate drops below 1900 gpm, a scram signal is
automatically actuated. Followed by the initiation of low flow scram signal, the shim
blade magnets are deenergized and then all six shim blades drop at the core periphery.
Two events might result in low primary coolant flow rate: loss of offsite electric
power and pump coast down accident due to pump hardware failure. The later might
be caused by malfunctions of motors or failures of pump power supply. In both case,
the MITR loses primary flow via pump coast down. The pump coast down curve
used in RELAP5/mod3.3 is,
where Q is gallons per minute and t is time in seconds. This new curve fit was
obtained when the heat exchanger outage was performed [8].
To simulate the scenarios of LOF, relevant trips settings and tabulated data in input
deck were constructed. The simulated LOF scenarios begin with reactor scram signal
actuation at 0.0 seconds. In contrast to step reactivity insertion applied in MULCH,
which was mentioned in Ko's study [1], ramp reactivity insertion was applied in
RELAP5/MOD3.3 simulation that 7.5 beta of reactivity was introduced between 1.3
and 2.3 seconds after reactor scram. Shim blade insertion is assumed to have one
second delay to reflect the signal transmission delay in real situation.
The assumption that the shim blade insertion takes 1.3 seconds is based on MITR
shim bank integral curve [9]. The heat removal mechanism during LOF transient is
natural circulation. During which natural circulation valves and antisiphon valves
are open since the pressure applied on the balls blocking on the aperture of valves
decreases during LOF. In RELAP5/MOD3.3 simulation, these valves were manually
set to open on 4.4 seconds after the scram signal.
147
Hot leg
105 20113
Mixing area 1
106 107
204
108
Flow shroud
02208
210
3011 401 501
.21L ~~~
FFuel110 bottom
148
7.4 Natural Convection LSSS Calculation
Two cases are analyzed in this study using RELAP5/mod3.3: both the ASVs and
NCVs are open and only NCVs are open. The first case describing the lowpower
operation situation without forced primary flow while the second case may occur if
the coolant level drops below the ASV (about 6.4 feet above top of the core).
The calculations are performed assuming that the reactor is at 1 MW before a loss of
primary flow (LOF) occurring at t=300 second and the data required for LSSS
calculation is retrieved at where reactor power is at 100 kW, as depicted in Figure 73.
Coolant inlet temperature is assumed as 60 *C, because the resistance temperature
detector (RTD) located at the outlet pipe level cannot measure the instantaneous core
outlet temperature due to slow coolant mixing in the upper core tank region during
natural convection.
The criterion for the LSSS during natural convection is the avoidance of ONB, which
is the same as that in forced convection. Recall that the margin to ONB ATONB can be
calculated using,
Where
0.0234
( hs ) 2.16
7ONB 1082p +
1.8
ZH q h(Z
Tcla dT Tin + ,(Hqhs)dz+FA hs
mc h
The difference between calculating LSSS in natural convection and forced convection
is that the HTCs, pressures, and HCMFR are directly retrieved from RELAP5 instead
of using hand calculation. This is because (1) this study mainly focuses on the forced
convection LSSS, which also receives more attention in license application, and (2)
for simplicity.
The margins to ONB for both cases are summarized in Table 71 and Table 72. To
make sure the HTCs computed by RELAP5 is based on laminar flow condition, which
is expected to occur in natural convection, Reynolds number is also retrieved from the
149
data. As can be seen in Tables 71 and 72, the Reynolds number for each node is
well below 2,200, indicating that viscous effect are dominant and laminar flow
prevails.
The margins to ONB are adequate during natural convection since the minimum
margin to ONB is about 36.4*C for both cases, as demonstrated in Tables 71 and 72.
150
1.20E+06
1.OOE+06
8.00E+05
0 6.OOE+05
0
U
M 4.OOE+05
cc
2.OOE+05
O.OOE+00
0 200 400 600 am 1000 1200 1400 1600
Time (s)
Figure 72 Decay power changes with time (initial power is set as 1MW)
Table 71 Cladding temperature and temperature inducing ONB at each node (both
the NCVs and ASVs are open)
node 1 node 2 node 3 node 4 node 5 node 6 node 7 node 8 node 9 node 10
Reynolds 451 455 458 461 464 467 468 469 468 467
Number
Hot stripe HTC 1698 1704 1710 1716 1720 1725 1728 1729 1728 1727
(W/m 2 K)
Pressure at hot 1.26 1.26 1.25 1.24 1.24 1.23 1.23 1.22 1.22 1.21
stripe (bar)
Heat flux at hot 4053 3883 4080 4149 4120 3961 3444 2561 1260 632
stripe (W/m2)
Cladding
temperature at 64.09 64.96 66.15 67.26 68.30 69.20 69.70 69.71 69.08 68.78
hot stripe (*C)
Temperature that 10724 107.09 107.00 106.86 106.72 106.57 106.40 106.15 105.80 105.51
induces ONB ("C) 107.0 10 _____ 0
___0 00
Margin to ONB 43.15 42.13 40.85 39.60 38.42 37.37 36.70 36.44 36.72 36.73
("C) I I I II
151
Table 72 Cladding temperature and temperature inducing ONB at each node (Only
NCVs are open)
node 1 node 2 node 3 node 4 node 5 node 6 node 7 node 8 node 9 node 10
Reynolds 482 486 489 493 496 498 500 501 500 499
Number
Hot npe HTC 1725 1732 1738 1743 1748 1752 1756 1757 1756 1754
Pressure at hot 1.26 1.26 1.25 1.24 1.24 1.23 1.23 1.22 1.22 1.21
stripe (bar)
Heat flux at hot 4128 3951 4151 4221 4190 4029 3502 2604 1280 642
stripe (W/m 2)
Cladding
temperature at 64.10 64.96 66.15 67.26 68.29 69.19 69.68 69.69 69.06 68.76
hot stripe (*C)
Temperature that 107.27 107.09 107.00 106.87 106.73 106.58 106.40 106.15 105.80 105.51
induces ONB (*C)
Margin to 43.17 42.14 40.86 39.61 38.44 37.39 36.72 36.46 36.75 36.76
ONB(0 C) _ _ _ _ _ I_
________
152
7.5 Summary
153
Table 73 Calculated Coolant Temperature Rise and Film Temperature Rise for
Natural Convection Operation
Tin (*C) 60 60
154
References
[75] W. H. Rettig et al. RELAP3  "A Computer Program for Reactor Blowdown
Analysis", IN1445. Idaho National Engineering Laboratory. February 1971
[76] K. V. Moore and W. H. Rettig. RELAP4  "A Computer Program for Transient
ThermalHydraulic Analysis. ANCR1127", Idaho National Engineering
Laboratory. March 1975.
[79] MITR Staff, "Safety Analysis Report for the MIT Research Reactor
(MITRIII)", Chapter 4, MIT Nuclear Reactor Laboratory, July 1999.
[710] Research reactor core conversion guidebook, Vol. 4: Fuels (Appendices IK),
IAEATECDOC643, IAEA.
155
Chapter 8 Conclusions and Recommendations
8.1 Conclusions
In this study, several project accomplishments have been achieved: (1) Propose a
generalized equation that can be used to calculate LSSS for platefueltype research
reactors, (2) Calculate the analytical forced convection LSSS, OFT and CHF for the
MITR, which are respectively 8.3, 12.4 and 70 MW, (3) Quantify coolant channel
fabrication uncertainty, which is the dominant sources of engineering uncertainties,
and based upon which derive associated parametric uncertainties, (4) Develop and use
uncertainty propagation technique for the MITR calculating forced convection LSSS,
OFI and CHF, which are respectively obtained as 9.1, 10.4 and 47 MW using Oracle
Crystal Ball, (5) Calculate the LSSS for the proposed MITRLEU core for natural
convection modes using RELAP5/mod3 for two cases: both NCVs and ASVs are
open and only NCVs are open, and the minimum margin to ONB is about 36.4'C for
these two cases.
In the uncertainty propagation methodology LSSS was taken at (mean3o) value. The
calculation shows that the probability of ONB occurrence is roughly 0.15% when
operating power is 9.1MW. This result not only permits 0.8 MW additional margins
comparing to analytical approach, but also indicates the low probability of ONB
occurrence at this power level. Besides, there is adequate margin between 9.1 MW
and the targeting license power of the MITR 7 MW, making possible the proposed
power uprate from 6 MW to 7 MW.
156
8.2 Recommendations for This Study
(1) Flow and heat transfer behavior at the channel edge would be a concern:
The MITR has narrow rectangular coolant channels and therefore heat transfer
rate at the corner of coolant channel may be lower, comparing to the center region
due to reduced turbulent convection. The flow and heat transfer behavior near the
edge of the channels could be investigated using CFD software. This effect could
have significant influence if power peaking, which requires further neutronic
calculation, is not negligible near the end of fuel meat.
157
(5) Analyses based on halfchannels:
The MITR has a unique design such that there are fullchannels and halfchannels
in the core. Note that the total number of flow channels 432 used in this study is
based on the assumption that two half channels form a full channel, which
simplifies the analyses. This study only focuses on the fullchannels of the MITR
assuming that the most limiting result throughout the core is on one of the full
channels. For licensing, it is suggested to conduct similar analyses on half
channels for completeness.
158
Appendix A.
RELAP5 Input File for Natural Circulation LSSS of MITR (Steady State)
* time step
*
201 300. 1.09 .005 3 100 1000 500 * SJK 5/12/2010) time step control 3,
minimum time step=0.005 sec
*
301 count 0
302 cputime 0
303 dt 0
304 dtcrnt 0
*
401 time 1 ge null 1 10000. 1* SJK 071409 no ASV trip within specified
operation duration
402 time 1 ge null 1 10000. 1 * SJK 071409 no NCV trip within specified
operation duration
403 time 1 ge null 1 10000. 1 * SJK 071409 no pump trip within specified
operation duration
*
* hydrodynamic components
*
159
1000201 0. 1.02+5 328.0 * initial p, T, by MULCH S.S. compinent #1 YCK
032207
*
160
1080101 .130.76 .0988 0.90. .76.00001 .387 11000 *YCK 022807
1080200 103 1.05+5 333.15 * initial p, T
*
161
2020000 ASV valve
2020101 105010002 203010001 .007674 6.90 7.90 100 1.0 1.0 1.0 * 2 valves
2020201 1 0. 0. 0. * initial flow rate
2020300 trpvlv * trip valve
2020301 401 * trip 401
*
162
*
163
2110000 rgn4toi sngljun * region 4 to inlet plenum
2110101 210100002 110010001 .029 2.05 2.05 100 * YCK 022807
2110201 1 112.5 0. 0. * initial flow rate
*
164
13021601 0,0,0,0,4.922345,10 * right boundary condition, insulated * SJK
072709 LEU 24 elements
13021701 1000 9.931029e02 0. 0. 1 * axial source distribution, KYC 2011127:
431 avg. in 24 LEU elements
13021702 1000 1.043853e01 0. 0. 2 *SJK 062609
13021703 1000 1.119480e01 0. 0. 3 *SJK 062609
13021704 1000 1.216004e01 0. 0. 4 *SJK 062609
13021705 1000 1.193117e01 0. 0. 5 *SJK 062609
13021706 1000 1.134406e01 0. 0. 6 *SJK 062609
13021707 1000 1.022956e01 0. 0. 7 *SJK 062609
13021708 1000 9.075249e02 0. 0. 8 *SJK 062609
13021709 1000 7.035308e02 0. 0. 9 *SJK 062609
13021710 1000 6.169577e02 0. 0. 10 *SJK 062609
130218000
13021801 .0021253,10.,10.,0.,0.,0.,0.,1.0,10 * additional left boundary * SJK
5/12/2011
130219000
13021901 .0021253,10.,10.,0.,0.,0.,0.,1.0,10 * additional right boundary * SJK
5/12/2011
*
165
4030201 1 .259 0. 0.* initial flow rate * YCK 032407
*
166
5020001 10 * number of nodes
5020101 4.19343,10 * area * SJK 5/12/2011
5020301 .06478,1 .05683,9 .06478,10 * node lengthss
5020601 90.,10 * vertical angles
5020801 .00001,.0018820,10 * roughness, Dw *SJK 5/12/2011
5021001 11000,10 * volume control flags
5021101 1020,9 * junction control flags
5021201 103,1.08+5,320.4,0.,0.,0.,10 * initial pressure, temperature
5021300 1 * use mass flows below
5021301 8.91,0.,0.,9 * initial junction flow rates * YCK 032407
*
* tables
*
167
20100352 1.292e6 *
373.15
20100353 1.331e6 *
473.15
20100354 1.387e6 *
573.15
20100355 1.443e6 *
673.15
20100356 1.507e6 *
773.15
20100357 1.542e6 *
873.15
20100358 973.15 1.586e6 *
*
* point kinetics
*
168
Appendix B.
RELAP5 Input File for Natural Circulation LSSS of MITR (Restart file)
* time step
*
201 375. 1.09 .000125 23 200 1000 500 * SJK 5/12/2011 time step control 23, max
time step=0.000125 sec
202 1500. 1.09.005 23 20 1000 500 * SJK 5/12/2011 time step control 23, max time
step=0.005 sec
*
301 count 0
302 cputime 0
303 dt 0
304 dtcrnt 0
*
403 time 1 ge null 1 0.0 1 * pump trip at restarting, i.e., @t= 0.0 and subsequent
coastdown *SJK 070309
401 time 1 ge timeof 403 4.4 1 * trip ASV at t = 4.4 and latch *SJK 070309
402 time 1 ge timeof 403 4.4 1 * trip NCV at t = 4.4 and latch *SJK 070309
611 403 and 403 n 1. * SJK @Rxtrip*
*
169
*
* tables
*
170
20200100 react 611 * General table 1, scram reactivity
20200101 0. 0. * t, reactivity ($)
20200102 1.3 0. * SJK 071509
20200103 2.3 7.5 * SJK 071509
20200104 3.3 10.0 * SJK 071509
20200105 10000. 10.0 * SJK 071509
*
* point kinetics
*
171