вход по аккаунту



код для вставкиСкачать
An Integrated Analysis of a NERVA Based
Nuclear Thermal Propulsion System
Cite as: AIP Conference Proceedings 813, 870 (2006);
Published Online: 01 February 2006
Hans Ludewig, Lap-Yan Cheng, Lynne Ecker, and Michael Todosow
Review of Helium and Xenon Pure Component and Mixture Transport Properties and
Recommendation of Estimating Approach for Project Prometheus (Viscosity and Thermal
AIP Conference Proceedings 880, 559 (2007);
3D Reacting Flow Analysis of LANTR Nozzles
AIP Conference Proceedings 813, 858 (2006);
Lunar Fission Surface Power System Design and Implementation Concept
AIP Conference Proceedings 813, 942 (2006);
AIP Conference Proceedings 813, 870 (2006);
© 2006 American Institute of Physics.
813, 870
An Integrated Analysis of a NERVA Based
Nuclear Thermal Propulsion System
Hans Ludewig, Lap-Yan Cheng, Lynne Ecker, and Michael Todosow
Energy Sciences and Technology Department, Brookhaven National Laboratory, Upton, NY 11973
631-344-2624, [email protected]
Abstract. This paper presents results and conclusions derived from an integrated analysis of a NERVA based
Nuclear Thermal Propulsion (NTP) system. The NTP system is sized to generate a thrust of 70,000 N (15,000 lbf),
and have a specific impulse (Isp) of 860 s. This implies a reactor that operates at 350 MWth and has a mixed mean
propellant outlet temperature of 2760 K. The integrated analysis will require that self-consistent neutronic/thermalhydraulic/stress analyses be carried out. The major code packages used in this analysis are MCNP, RELAP, and
ANSYS. Results from this analysis indicate that nuclear data will have to be re-generated to cover the wide
temperature range, zone loading will be necessary to avoid entering the liquidus region for the fuel, and the
effectiveness of the ZrC insulator will have implications for bi-modal applications. These results suggest a path
forward in the development of a viable NTP system based on a NERVA reactor should initially concentrate on fuel
and structural materials and associated coating development. A series of safety related criticality determinations
were carried out addressing water immersion following a launch incident
Keywords: Nuclear Thermal Propulsion, NERVA.
PACS: 28.50.Ky.
A design study of Nuclear Thermal Propulsion (NTP) based rocket must start with the definition of mission
requirements. These requirements connect the engine design to the remainder of the space vehicle performance
parameters. In the initial study to be outlined below, thrust only NTP systems will be considered. The assumed
mission requirements for the initial study are shown in Table 1.
TABLE 1. Mission Requirements.
Specific Impulse
Total Full Power Thrust Time
Desired Value
71600 N
860 s
2 hrs.
Not Specified
The above requirements imply the overall reactor design goals shown in Table 2.
TABLE 2. Reactor Design Goals.
Thermal Power
Mixed Mean Outlet Temperature
Propellant Flow Rate
Engine Cycle
Expansion Ratio
Goal Value
350 MW
2750 K
8.13 kg/s
Full Flow
Two different reactor designs are considered. The first design to be considered is based on the Small Nuclear
Reactor (SNR) NERVA design. In this case the fuel elements will consist of a carbon based composite consisting of
CP813, Space Technology and Applications International Forum—STAIF 2006, edited by M. S. El-Genk
© 2006 American Institute of Physics 0-7354-0305-8/06/$23.00
(U-Zr-C)-C, appropriately coated by a metal carbide to protect the surfaces from the hot hydrogen propellant. In
addition, this design has tie-tubes that tie the outlet grid to the inlet grid support structure, and add moderation in the
form of a metal hydride. The second design is based on the GE-710 configuration. In this case the fuel elements
consist of a W-UO2 cermet, appropriately coated to retain fission products and protect the surface from attack by the
hot hydrogen propellant. Besides the different fuel element designs these two concepts differ in a fundamental
manner. The first design has an epi-thermal neutron spectrum, while the second design has a fast neutron spectrum.
These two differences have an impact on the fissile requirements (uranium-235 loading), safety issues, and engine
Details of the reactors are given in Table 3 below.
TABLE 3. Details of Reactors.
Core Diameter (cm)
Reflector Thickness (cm)
Core Height (cm)
Number of Fuel Elements
Fuel Element Pitch (cm)
Coolant Passage Diameter (cm)
Number of Coolant Passages per Element
Number of Tie-Tubes
Fuel Form
Uranium Density (gm/cc)
Moderator Form
Number of Control Drums
Control material
The objective of this work is to carry out an analysis of a NTP reactor system that is internally self consistent. In
order to satisfy this condition power distributions and multiplication factors must be determined from a neutronic
analysis, coolant temperature, pressure and heat transfer coefficients must be determined from a thermal-hydraulic
analysis, and solid component temperatures, stresses, and deformations must be determined from a heat conduction
and stress analysis. This inter-connected analysis method will be discussed in more detail below.
The calculations to be carried out for the proposed reactor designs need to recognize that all reactors must be
critical, the heat generated in the core must be removed, and the component stresses need to be below acceptable
limits. A scheme linking the above three calculational disciplines is shown in Fig. 1. It is seen that once a critical
configuration has been established, it is necessary to estimate the heat generated in the various core and reflector
volumes. Following this step, the thermal and flow conditions are determined corresponding to the heat deposition
distribution. In addition, the heat transfer coefficients at the solid volume/coolant interface surfaces are determined.
These values are used as boundary conditions for detailed thermal calculations of the solid components, and
subsequently the associated thermal stresses can be determined. The linking among the various calculations is
iterative, and it is expected that several iterations will be necessary to arrive at a converged overall solution for the
critical configuration, heat deposition distribution, temperature distribution, and implied thermal stresses.
In the calculations outlined below the following codes are used:
Neutronics calculations will be carried out using the Monte Carlo code MCNP (Los Alamos National Laboratory,
1999) and the associated nuclear data library (ENDF/B-IV). Inputs to this calculation are core composition,
geometry, component temperatures, and coolant temperatures and pressures. Output will be multiplication factors
and power distribution.
Fluid dynamics and heat transfer calculations will be carried out using the system code ATHENA (Idaho National
Laboratory, 2003). Input to this code will be power distribution, and system configuration. Output will consist of
coolant temperature and pressure, and heat transfer coefficients.
Solid component temperature, stress, and distortions will be determined using the ANSYS code (Swanson Analysis
Systems, 2004). Input to this calculation will be power distribution, heat transfer coefficients, and coolant
temperature. Output will be solid component temperature distribution, thermal stresses, and solid component
FIGURE 1. Linkage of Analysis Methods.
The reactor design descriptions used as the base for these analyses were published reports prepared by Los Alamos
National Laboratory (Durham, 1972) in the case of the NERVA concept, and General Electric (1973) in the case of
the GE-710 concept.
The MCNP model for the NERVA small engine reactor is shown in the cross section through the core of the reactor
in Fig.2. The partial elements surrounding the core consist of graphite reflector. The fuel element tie-tube structure
is shown in Figure 3. A MCNP model of equal detail was also created for the GE-710 (General Electric, 1973)
reactor. Details of the reactor configuration is shown in Table 3.
A detailed volume and junction model of the engine system shown in Fig. 4 was prepared for the ATHENA code.
Both a hot channel and an average channel are included in the core region. Heat structures are included for the
reflector, core, tie-tube, and nozzle volumes. Heat deposition is included for all of the volumes except the nozzle
volume. Detail temperature distributions, resulting thermal stresses, and including stresses due to mechanical loads
can be carried out using the ANSYS code. In the current analysis only the temperature distribution will be estimated.
The configuration to be analyzed includes half of a fuel element, a twelfth of a tie-tube at one end and a sixth of a
tie-tube at the other end (see Figure 3). The geometric dimensions are consistent with those used in MCNP, and the
thermal properties are consistent with those used in ATHENA. Finally heat transfer coefficients and coolant
temperatures are determined by ATHENA, and power distributions are obtained from MCNP.
The most important neutronic related results are a determination of the critical configuration and heat deposition in
the fuel and structural components. The MCNP models as described above were used to determine the multiplication
factor for two control drum positions, heat deposition in selected volumes, and safety related criticality
determinations. In the case of the NERVA based reactor, criticality estimates assumed uranium loading of 600
mg/cc, enriched to 93 %. This uranium density implies a fissile loading of 64 kg of uranium-235. In the case of the
GE-710 based reactor a 50/50 mixture of tungsten to uranium oxide was assumed. This implied a fissile loading of
252.3 kg of uranium-235 in this case.
FIGURE 2. Cross Section Through NERVA Reactor.
The multiplication factors for the NERVA core described above are given in Table 4.
TABLE 4. Multiplication Factors for NERVA Reactor.
Control Drum Position
No control drums present
Control Drums Out
Control Drums In
Multiplication factor
1.06287 (±0.00062)
1.04133 (±0.00062)
0.92111 (±0.00060)
It is seen that with the control drums out, poison in the furthest position from the core, there is sufficient margin to
allow for any reduction in multiplication factor introduced by a potential “real” design exercise. Furthermore, the
swing in multiplication factor between drums out and drums in is 0.12022, which is ~ $ 17.0. The neutron lifetime
for this core is approximately 21 µs.
Heat deposition estimates in the fuel and tie-tubes were made to be consistent with the input to the ATHENA code.
These estimates thus recognize an average channel and a hot channel. The total heat deposited in the core is given
below in Table 5 as a function of axial slice.
TABLE 5. Heat Deposition in Core (MW).
Axial Interval (m)
Fuel region
0.0 – 0.18
0.18 – 0.36
0.36 – 0.54
0.54 – 0.72
0.72 – 0.9
The summary of all the heat deposited is given in Table 6.
Tie-Tube (down)
Tie-Tube (up)
A comparison between the initial temperature assumptions and those determined after the first iteration are shown in
Table 7 (all temperatures in K) for both selected solid components and coolant in the core. Deviations in these
temperatures comparisons indicate the desirability of carrying out the
series of calculations for at least one more iteration.
In the five core volumes the coolant and solid temperatures were
initially assumed to be the same. Results from the ATHENA
calculation indicates that the coolant temperature starts at a lower
value, but then increases to higher values at the exhaust. The solid
(fuel) temperature is seen to be several hundred degrees higher than the
coolant, and significantly higher than the initially assumed values. It
should be noted that the ATHENA solid temperatures listed in Table 6
are volume averaged values. The multi-dimensional ANSYS
calculation results (using heating values from MCNP and heat transfer
coefficients from ATHENA) in a range of temperatures, bracketing the
ATHENA value. The maximum temperature is always higher than the
ATHENA value, and is presumably the maximum that the fuel would
have to tolerate. The difference between the minimum and maximum
at any location within the fuel is ~ 300 K, and the thermal stress is
directly proportional to the implied temperature gradient.
FIGURE 4. NERVA Engine System.
FIGURE 3. Fuel Element and Tie-Tube
The initial tie-tube (d) coolant temperatures are quite similar
(within ~ 10 K) to those determined by ATHENA. However
the initial zirconium hydride temperatures are quite different
from those determined by either ATHENA or ANSYS, with
the ANSYS temperature determinations being the most
credible estimates. The initial tie-tube (u) coolant temperatures
are higher than those calculated by ATHENA. The solid (ZrC)
temperatures as determined by ANSYS is significantly above
those which were assumed and those determined by ATHENA.
This difference is due to the multi-dimensional nature of the
ANSYS calculation, which recognizes the close proximity of
the fuel element, while the other estimates do not recognize
this thermal connection. Though thermal contact conductance
can be modeled by using the conduction enclosure model in
ATHENA it is not as accurate and convenient to implement as
using the features in ANSYS. It is clear that it is necessary to
re-determine the energy deposition using the neutronic codes
with the revised temperature distributions. These revised
values could then be used one more time to validate the
temperature distribution. The final result would be a reactor
analysis with consistent neutronic, thermal and stress analyses.
TABLE 6. Summary of Heat Deposition (MW).
Radial reflector
Axial grid/shield
Heat Deposition
Initial criticality estimates for the GE-710 cermet based reactor are given below in Table 8.
The results, shown in Table 8, indicate that the control drums are worth approximately $ 1.0, which is low.
Conceivably a slightly lower fissile loading and the use of enriched boron might increase the control drum worth.
TABLE 7. Comparison of Selected Temperatures Following Heat Transfer Calculations.
Core 1
Core 2
Core 3
Core 4
Core 5
Tie-tube (d) 1
Tie-tube (d) 2
Tie-tube (d) 3
Tie-tube (d) 4
Tie-tube (d) 5
Tie-tube (u) 1
Tie-tube (u) 2
Tie-tube (u) 3
Tie-tube (u) 4
Tie-tube (u) 5
Initial Coolant
Initial Solid*
ANSYS Solid*
890.0– 974.3
1709 – 1977
2326 – 2724
2635 – 3064
2593 – 2962
111.3– 151.5
133.0– 167.0
137.9– 161.5
122.3– 135.8
89.3– 93.7
273.6– 881.8
448 – 1651
611 – 2209
712 – 2481
685 – 2436
* Core 1 – 5 = Fuel , Tie-tube (d) 1 – 5 = ZrHx , Tie-tube (u) 1 – 5 = ZrC
TABLE 8. Multiplication Factor as a Function of Control Drum Position.
Control Drum Position
Control Drums Out
Control Drums In
Multiplication factor
1.00940 (±0.00057)
1.00271 (±0.00061)
In addition to the above neutronic and thermal-hydraulic calculations a series of safety related criticality
determinations were carried out to determine the variation of multiplication factor corresponding to the following
Reactor (core and reflector) surrounded by light water,
Reactor surrounded and flooded by light water,
Reactor surrounded by dry sand (GE-710 cermet reactor only), and
Core surrounded by dry sand (GE-710 cermet reactor only).
In the case of the base case NERVA reactor, corresponding to a configuration that is dry (no hydrogen) and at
ambient temperature the resulting multiplication factors are given below in table 9.
TABLE 9. Initial Safety Related Criticality Results for NERVA Based Reactor.
Configuration description
Core dry – drums in
Reactor surrounded by light water core dry – drums in
Reactor surrounded and flooded by light water – drums in
Multiplication factor
These results indicate that surrounding the core with light water increases the multiplication factor by approximately
0.033. However, flooding the core in addition to surrounding the core results in a large increase in criticality, and
would clearly lead to a serious situation. The slight increase due to surrounding the reactor with water is due to
reflection at the bottom of the core.
The results presented in Table 9 indicate that removal or displacement of the water in the core is a potentially good
way of reducing the magnitude of the multiplication factor. It was thus proposed to fill each coolant passage (10792
coolant passages) with a Teflon string, and if necessary the Teflon could be infiltrated with a neutron poison
(boron). Teflon is relatively soft and should easily slide in and out of the coolant passages. Furthermore, it consists
of carbon and fluorine, neither of which are as good as hydrogen as a neutron moderator. A series of calculations
was then carried out in which all the coolant passages contained a Teflon string with an effective diameter of 0.2 cm
(used for displacing water in the coolant passage). The remainder of the core would still be flooded, and the outside
of the reactor would still be surrounded by water. The results of the multiplication factor determinations are shown
in Table 10.
The results below indicate that only displacing the water in the fuel elements is not enough to reduce the criticality
below unity. A neutron poison needs to be added to the Teflon strings to ensure a sub-critical assembly will result
following surrounding the reactor with water and flooding of the core with water.
TABLE 10. Flooded Core With Teflon Strings Inserted in Coolant Passages for NERVA Based Reactor.
Boron content (atom percent)
Multiplication factor
A series of safety related criticality determinations were carried out for the GE-710 based reactor at this stage to
estimate the effects of immersion in sand with the core dry, and immersion in water with the core dry and flooded.
These estimates were all carried out on the reactor with the control drums in the “in” position. The two stage water
immersion calculations were carried out to estimate the variation in multiplication factor as the core floods. Finally,
a possible re-mediation method is also explored. The results are shown in Table 11.
TABLE 11. Safety Related Multiplication Factors for GE-710 Based Reactor.
State of Reactor
Core dry, drums in
Water surrounding reactor
Water surrounding and flooding reactor
Sand surrounding reactor
Sand surrounding core (No reflector)
Sand surrounding core (core loaded with polyethylene)
Multiplication factor
1.00271 (±0.00061)
1.01844 (±0.00062)
1.02285 (±0.00056)
1.03060 (±0.00060)
0.95308 (±0.00050)
These results indicate that the multiplication factor increases when the core is surrounded by water, but not flooded.
The increase is primarily due to increased reflection along the outlet plenum end of the core, which is generally
considered to be unreflected in the “operational” configuration. Once the core is flooded the multiplication factor
drops precipitously, due to the shift in the neutron spectrum to a lower energy, and the subsequent neutron poisoning
effect of the tungsten and rhenium in the core. However, it should be noted that the reactor might never get to the
flooded state, since the reactivity input of external reflection is ~ $ 2.0. If the reactor buries itself in the surrounding
sand (assumed to be silicon dioxide with a density of 1.5 gm/cc), either with or without its reflector, then it is seen
that the multiplication factor increases by approximately the same amount. The solution to this dilemma is to ensure
that the neutron spectrum is guaranteed to be at least epi-thermal, thus ensuring that the core will be sub-critical
regardless of the type and state of the reflector. This situation can be achieved by inserting polyethylene “rods” into
all the coolant passages. In this manner, in the event of an immersion in either water or sand, the core will be well
Results showed that both reactor concepts considered in the above study would be viable candidates to satisfy the
mission requirements stated in Table 1. They are both critical with reasonable fissile material loading, and the
average power density in the fuel regions is in an acceptable range implying reasonable temperature gradients and
thermal stresses (the latter statement was not verified by analysis). The highest fuel matrix temperature in both cases
is well in excess of the 2750 K required for the coolant. A more careful design should investigate the possibility of
zone loading the core. In the case of the composite (NERVA) core the cooler portions (close to the inlet plenum)
could have a higher fissile density (~ 900 mg/cc), and at the exhaust end (temperatures above ~ 2500 K) the loading
could be lower (~ 200 mg/cc). The cermet fuel temperature limits are not clear, experiments (General Electric, 1973)
have indicated that the highest temperature possible is close to the maximum coolant temperature. Finally, the loss
of both uranium and fission products will be controlled by migration through the fuel matrix and then by
evaporation and sublimation from the coolant passage surface. These phenomena will need careful study, since
temperature goals have increased over those values acceptable in the past (Lyon, 1973; Storms, 1992). The ZrC
insulator is effective at thermally separating the tie-tubes from the fuel elements, which has implications for bimodal operation. Finally, the above discussion and preliminary conclusions indicate the need for carrying out at
least one complete cycle around the linked methods described in Fig. 1. The result of these calculations will lead to a
more converged and consistent concept design.
In both cases studied above, in order to prevent a criticality incident upon immersion in either water or dry sand, it
will be necessary to launch the reactor with criticality control devices inserted in the core. This poison addition
displaces water (substituting a relatively poor moderator in place of the good moderator), and also increasing the
neutron absorption. In the case of the GE-710 reactor all the coolant channels must be filled with a good moderator,
preferably hydrogenous material (polyethylene). In this manner the neutron energy spectrum is guaranteed to shift to
lower energies. Under these conditions the high absorption cross sections of tungsten and rhenium will act as
poisons and ensure that the reactor is always sub-critical, regardless of the reflector conditions (water or dry sand
reflection on all sides leads to a reactivity increase). A viable mechanical design of the core insertion devise needs to
be conceptualized, and validated by carrying out a series of prototypic experiments. In addition, a safety analysis of
the operation of the mechanism will be needed.
The most important research and development activities that need to be addressed in the near term (~ 5 years) are:
Fuel composition and associated coating or cladding options should not be limited to those candidates considered in
the past. In addition, work carried out in the FSU (Anghaie and Knight, 2002) on (U-Zr-Nb-C), and work on the
infiltration of graphite by UC2, developed during the SNTP program (Ludewig, et al., 1996), should be considered.
Coating technology has advances since the earlier attempts at creating NTP fuel, and it has been found that multiple
coatings perform more satisfactorily (Barletta, et al., 1993; Adams, et al., 1993) at mitigating mass loss. Finally, it
will be necessary to develop a fuel performance code in order to better predict system operation. Attempts have been
made in the past (Storms, 1992) to create such a code. Structural materials and their coating and cladding,
Moderator/reflector selections (for those cases using moderators) have a significant effect on the fissile loading,
potentially engine mass (influencing thrust/weight), and on post accident criticality behavior. Radiation damage and
the ability of the moderator to survive the space environment are two important effects to be considered. Fuel
element thermal hydraulic and mechanical validation test need to be closely integrated in the reactor core design
effort. The primary task of the effort will be to validate the mechanical and chemical compatibility of the element
design under steady state and transient conditions. Development of suitable instrumentation to monitor engine
performance, particularly during start-up. Some of this instrumentation already exists to monitor the operation of
chemical rocket engines. However, there will be significant differences that will need research and development
efforts. Engine component tests will be initially followed by an integrated engine test, excluding the reactor. This
test will include the turbo-pump assembly, propellant management system, and thrust vector control system. The
reactor response will be represented by a simulator and by an appropriate hydrogen source to operate the TPA. A
full size prototypic ground test assembly can be conducted to carry out final validating tests of the fuel element
design, and the engine arrangement. In addition, mass loss at full operating power and duration can be validated
during this test.
= First core volume, consists of both a hot channel and an average channel
TT-1 (d) = First downward tie-tube volume, consists of both a hot and average channel
TT-1 (u) = First upward tie-tube volume, consists of both a hot and average channel
= Reflector volume
= Nozzle tube volume
= Nozzle exhaust chamber volume
This work was performed by the Brookhaven National Laboratory (BNL) for the U.S. Department of Energy (DOE)
in support of the National Aeronautics and Space Administration (NASA). BNL is managed for DOE by
Brookhaven Science Associates, LLC, under contract DE-AC02-98CH10886. Any opinions expressed in this paper
are those of the authors and do not necessarily reflect the views of DOE or NASA.
Adams, J.W., Barletta, R.E., Svandrlik, J. and Vanier, P.E. “Performance of CVR Coatings for PBR Fuels,” in Proceedings of the
Materials Research Society, Boston, MA, 1993.
Anghaie, S. and Knight, T. “Status of Carbide Fuels: Past, Present and Future,” in Symposium on Space Nuclear Power and
Propulsion, edited by M.S. El Genk, AIP Conference Proceedings, 0-7354-0052, 2002.
Barletta, R.E., Vanier, P.E., Dowell, M.B. and Lennartz. J.A. The Development of CVR Coatings for PBR Fuels, in Proceedings
of the Materials Research Society, Boston, MA, 1993, pp. 852-856.
Durham, F.P., Nuclear Engine Definition Study Preliminary Report Vols I,II, and III, LA-5044-MS, Los Alamos National
Laboratory, Los Alamos, NM, 1972.
Idaho National Laboratory ATHENA Code Manual Volume I: Code Structure, System Models and Solution Methods, INEELEXT-98-00834 Rev 2.2, Idaho Falls, ID, 2003.
Los Alamos National Laboratory MCNPX Users Manual-Version 2.1.5, TPO-E83-G-UG-X-00001, Los Alamos, NM 1999.
Ludewig, H., et al. “Design of Particle Bed Reactors for the Space Nuclear Propulsion Project,” Progress in Nuc. Eng. 1, 1-65,
Lyon, L.L. Performance of (U,Zr)C-Graphite Composite, and (U,Zr)C Carbide Fuel Elements in the Nuclear Furnace 1 Test
Reactor, LA-5398-MS, Los Alamos National Laboratory, Los Alamos, NM, 1973.
Nuclear Energy Division, General Electric 710 High Temperature Gas Reactor Program Summary Report, Vols I – VI, GEMP600, Cincinnati, OH, 1973.
Storms, E. The Behavior of ZrC1-x and UyZr1-yC1-x in Flowing Hydrogen at Very High Temperatures, LA-12043-MS, Los Alamos
National Laboratory, Los Alamos, NM, 1992.
Swanson Analysis Systems, Inc. ANSYS 8.1 User’s Manual Vols I – V, Houston, PA, 2004.
Без категории
Размер файла
827 Кб
Пожаловаться на содержимое документа