Titanium-Water Heat Pipe Radiators for Space Fission Power System Thermal Management

Kuan-Lin Lee, Calin Tarau*, William G. Anderson and Derek Beard

Advanced Cooling Technologies, Inc., Lancaster, PA 17601, United State


For future space transportation and planetary exploration mission power applications, NASA Glenn Research Center (GRC) is currently developing a small-scale nuclear fission system (i.e. Kilopower system), which has an operable range of 1 to 10 kWe and a design life of 8 to 15 years. The thermal management system of Kilopower system involves two types of heat pipes: high temperature alkali metal heat pipes that are used to transport thermal energy from the nuclear reactor to the Stirling convertors hot end and titanium water heat pipes that are used to remove the waste heat from the convertors cold end and transport it to the radiators for ultimate rejection. This paper presents the development of the titanium water heat pipes, which are featured with a special wick structure design: it has bi-porous screened evaporator and screen-groove hybrid wick in the adiabatic and condenser sections. This will allow the heat pipe to (1) operate in space with zero gravity (2) operate on planetary surface with gravity-aided orientation (3) be tested on ground with slight adverse gravity orientation and (4) to startup smoothly after being frozen. Under a research and development program, several freeze/thaw tolerant heat pipes were designed, fabricated and experimentally validated. Later, various heat pipe radiators were developed and tested in a thermal vacuum chamber (TVC). Test results successfully demonstrated that the titanium heat pipes with radiator attached are able to transfer the required power at the working temperature of 400K under space-like testing conditions with a thermal resistance of 0.019°C/W while the total heat pipe radiator weight is less than 0.73kg.


Nuclear fission power thermal management; Titanium-water heat pipe; Hybrid wick; Freeze-thaw tolerance


A      :  liquid flow area

Dh     :  hydraulic diameter

fRe  :  Poiseuille number for a trapezoidal groove

fv      :  vapor friction factor in a circular pipe

hfg     :  latent heat of vaporization

P      :  pressure

Q      :  heat input

r       :  meniscus radius

Seff    :  shear area between liquid and vapor flow

V      :  speed of vapor flow

z       :  axial direction

θ       :  inclination angle (rad)

ρ       :  density

μ       :  viscosity (Pa-s)


For future space transportation and surface power applications, NASA Glenn Research Center (GRC) in partnership with Los Alamos Laboratory (LANL) is leading the efforts to develop a small-scale nuclear fission power system (i.e. Kilopower System). This system is designed to provide 1 to 10 kW of electricity through Stirling conversion (Gibson, et al., 2017). The Kilopower nuclear ground testing completed on March 21, 2018 successfully demonstrated the Kilopower reactor operational modes (startup, steady-state and transient in space simulated environment etc.) (Gibson, et al., 2018). The thermal management architecture of the Kilopower system is illustrated in Fig. 1. The thermal energy of the nuclear reactor core is transferred to the Stirling convertor’s hot end through a series of high temperature sodium heat pipes (Beard, et al., 2017). After the energy conversion, the waste heat from the convertor’s cold end is transferred to the radiator panels through multiple titanium heat pipes and ultimately rejected into the space environment (Hay & Anderson, 2015)  (Beard, et al., 2017).

To reject 1 kW of waste heat, each heat pipe needs to carry at least 125 Watts of waste heat during nominal operation and 250 Watts in marginal conditions. The required operating temperature of the Stirling convertor cold end is 400K (~125°C). In this working temperature range, water is the ideal working fluid and titanium is the lightest envelope. The titanium water heat pipes for Kilopower cold end cooling are required to survive and operate under the following four situations:

  • Space operation (with zero-gravity).
  • Ground testing condition with a slight adverse gravity elevation (i.e. the evaporator is slightly higher than the condenser) to simulate microgravity operation.
  • Planetary surface operation with gravity-aided orientation, which is the simplest scenario.
  • Start-up after being exposed to the launch condition. In such scenarios, heat pipes are orientated with the evaporator above the condenser while ambient temperature can be potentially lower than the freezing point of water.

To meet the operation situations (1)–(3), axial grooves seem to be the best choice of wick structure due to its long distance liquid transport capability. Nevertheless, using axial grooves as the wick structure along cannot satisfy the startup condition after launch (condition (4)). When the Kilopower system is prepared for launch, the system will be oriented such that the evaporator of the heat pipe will be above the condenser. The heat pipe will de-prime and all of the working fluid (i.e. water) will be collected at the bottom of the heat pipe. During freezing conditions, liquid water accumulated at the bottom of the pipe will freeze and expand, damaging the pipe wall. Also, in order to ensure that there is sufficient liquid in the evaporator at start-up is necessary. The screen-grooved hybrid wick structure that was incorporated within the titanium water heat pipes is the solution for both freezing and start-up issues described above. The design of heat pipe with hybrid wick structure has been demonstrated by ACT  (Ababneh, et al., 2016).

Fig. 1 Kilopower system conceptual design and the thermal management system

Funded by a Small Business Innovation Research (SBIR) award, Advanced Cooling Technologies, Inc (ACT) developed the titanium water heat pipes attached with radiator panel for Kilopower waste heat rejection. These titanium heat pipes have bi-porous screen wick structure in the evaporator and have axial groove structure in the adiabatic section and condenser. This paper presents the development of titanium water heat pipes with radiator for Kilopower system waste heat rejection, including the hardware design, prototype development, deliverable heat pipes assembly and thermal performance experimental validation in a relevant environment.

Heat pipe development

Bi-porous screen evaporator  

To accommodate different interfacing surfaces with the Stirling convertor, two evaporator configurations were designed. The direct interface evaporator is shown in Fig. 2(a), which directly clamps to the cylindrical portion of the Stirling convertor. The waste heat generated from the Stirling convertor transfers into the C-shape evaporator through its curved inner annular surface. The second evaporator (Fig. 2(b)) interfaces with the Stirling convertor via a Cold Side Adapter Flange (CSAF). In this interfacing mode, the waste heat from the Stirling convertor conducts through the CSAF and enters the evaporator from its flat bottom surface. Special wick structure within the evaporator is shown in the cross section views of the evaporator (Fig. 2(c) and Fig. 2(d)). Fig. 2(c) is the cross section viewing from the top and Fig. 2(d) is the cross section of the A-A’ plane. As shown, two types of titanium screen with different pore sizes (i.e. bi-porous screen) were strategically inserted:

  • Fine screen meshes attached adjacent to the heating surfaces (i.e. inner annular and flat bottom) provide sufficient capillary action to pull the liquid working fluid from the grooved section into the heating surfaces.
  • Layers of coarse screen mesh occupying the remaining interior volume of the evaporator provide storage capability to hold the entire liquid inventory of the grooved section during the launch condition. ACT named it the “accumulator”.

During the launch, the liquid inventory that normally is contained in the grooves will be stored in the accumulator to prevent working fluid accumulation at the bottom of the pipe and damaging of the pipe because of the frozen liquid. As the A-A’ cross section (Fig. 2(d) shows, a 0.1” clearances (~0.1”) between the wall and the coarse meshes within the evaporator were left to accommodate the volume expansion of working fluid while freezing. During start-up, the liquid originally held within the accumulator will be capillary driven by the primary screen, vaporized, condensed in the condenser then re-distributed into the grooves. The primary screen (i.e. fine screen) will be always saturated with the working fluid. Liquid flow paths and the pressure drop across the accumulator was minimized by extending the grooved section and directly attaching it to the primary (fine) screen, which can be seen from the evaporator section view (Fig. 2(c)).

Fig. 2. Titanium water heat pipe evaporator design (a) evaporator directly interfaced with the cylindrical portion of Stirling convert (b) evaporator interfaced with Stirling convertor via CSAF (c) bi-porous screen distribution and working fluid transport direction inside the C-shape evaporator (d) cross section of A-A’ plane

 Grooved adiabatic and condenser sections

The remaining portion of the titanium heat pipe (i.e. the adiabatic section and the condenser) has an axial groove structure for liquid return during space operation. The geometry of grooves was designed based on a one-dimensional mathematical model developed by Do (Do, et al., 2008). In this model, the momentum equations of liquid phase and vapor phase (eqn. 1 and 2) were first solved with a given groove geometry and heat input. The variation of meniscus radius along grooves can then be determined by solving the Young-Laplace equation (eqn. 3). The relationship between the meniscus radius and the hydraulic parameters (e.g. diameter, liquid flow area and shear surface) within the grooves were determined by the groove geometry. The iteration process was performed by the MathCAD to determine the maximum power capability.

An extensive trade study was performed to identify the optimum groove geometry that has the highest heat transport capability with lower structural mass and thermal resistance. Specifications of the finalized groove profile are summarized in Table 1.  The groove profile manufactured by electrical discharge machining (EDM) is shown in Fig. 3. For performance validation, ACT fabricated a 120 cm long groove heat pipe and tested its heat transport capability in various orientations and working temperatures. The test results match the model prediction reasonably well (Beard, et al., 2017).

Table 1. Specifications of the grooved section

Pipe OD 1.59 cm
Adiabatic length 15.24 cm
Condenser length 100.0 m
Groove number 48
Groove depth 0.1 cm
Groove shape Trapezoidal
Min. wall thickness 0.05 cm

Heat Pipe Fabrication

The grooved titanium envelopes were manufactured by electrical discharge machining (EDM) in multiple sections (30.5 cm long). Then, one of the sections was bent at ~82° to avoid interference with the Stirling convertors. To avoid the damage of grooves and thin pipe wall during bending, the pipe was first filled with water-soluble particles (e.g. salts) and then bent to the designated curve. After bending, the rest of the sections were joined together through electron beam (EB) welding. The full-length (~1 m) grooved pipe was then coupled with the bi-porous screened evaporator through the screen-groove hybrid joint. End caps with pinch tube were then welded to the end of condenser. The heat pipe was then charged with DI water, and a small amount of argon as non-condensable gas (NCG) to facilitate the heat pipe startup process. Fig. 3 shows the titanium heat pipe before working fluid charging.

Fig. 3 Titanium water heat pipe with bi-porous screen evaporator (CSAF interface) and hybrid wick for the rest of the pipe

Heat pipe testing (without radiator)

Test setup

The experimental setup to test the thermal performance of the prototypes is shown in Fig. 4(a). Heat was applied to the evaporator by mock heaters, which simulate the waste heat from a Stirling convertor. Cooling was provided by several chiller blocks with liquid nitrogen (LN). Chiller blocks were mounted on the top of the condenser section of the heat pipe. Temperature distribution along the heat pipe was measured by 10 T-type thermocouple probes attached to the surface of CSAF evaporator and along the adiabatic and condenser sections.  To estimate heat pipe performance in microgravity, the heat pipe was orientated in such a way that the level of evaporator was slightly (< 0.25 cm) above the condenser. Heat pipe working temperature was maintained through active cooling. The system was thermally insulated to minimize heat leaks. As mentioned, NCG (argon) was charged to facilitate heat pipe startup.

Fig. 4 Experimental apparatus for titanium water heat pipe thermal performance testing (a) test system (b) Stirling convertor mock heaters

Heat transfer performance

Test results of the prototype heat pipe at the working temperature of 100°C is shown in Fig. 5.  To shorten the duration of testing, the pipe was started by orienting it in gravity-aided inclination and applying an initial power of 500W. Once the pipe started operating (t = 1800 second), it was placed in the slightly adverse gravity orientation (~ 0.25 cm) and the power was turned down to 125W. Heat pipe reached a steady-state at 3500 second. A uniform temperature distribution can be seen from the instantaneous temperature profile (Fig. 6). The temperature difference between evaporator and condenser was 2.8°C. Subsequently at t =6200 second, power was turned up to 200W and then to 250W. The heat pipe reached another steady state at t=9000 second. The corresponding temperature profile is shown in Fig. 7. At Q=250W, temperature difference increased to 4.9°C.

Fig. 5 Temperature profile variation with time (working temperature of 100°C, evaporator was elevated 0.25 cm higher than the condenser).

Fig. 6. Instantaneous temperature distribution along the heat pipe at t= 5500 sec (Q=125W)

Fig. 7. Instantaneous temperature distributions along the heat pipe at t= 9100 sec (Q=250 W)

Fig. 8 below shows the thermal resistance of the prototype heat pipe at working temperature of 125°C with different heat inputs (50W – 450W) and in various orientations (horizonal, 0.25cm adverse gravity, 0.51cm adverse gravity and gravity-aided). As the figure indicates, the heat pipe prototype is capable of carrying 450W of heat at 0.25 cm adverse gravity elevation. The average thermal resistance of the prototype heat pipe shown in the plot is 0.01 °C/W.

Fig. 8. Thermal resistance dependence on different heat inputs (working temperature at 125°C)

Freeze/thaw Startup performance

To test whether the heat pipe can survive and startup smoothly after being exposed to launch conditions, the heat pipe orientated in a slight against gravity orientation was first operating steadily at 125°C with 250W heat input (state 1 in Fig. 9). Freezing process begins at 1000 seconds, which is done by turning-off the heater and spreading LN to the evaporator and the mock heater. The condenser is cooled down by running LN to through the chiller blocks. The freezing process stops at 1800 seconds, when the highest temperature of evaporator (indicated in Fig. 9) drops below -50°C. To startup the pipe, heat was re-applied to the system incrementally from 125W to 250W. It can be observed from Fig. 9 that the heat pipe successfully survived and smoothly recovered back to the original state (state 1). As the working temperature reached 125°C at t=10500 seconds, the LN active cooling was ON again to maintain the working temperature. The heat pipe operated at the same steady state parameters as in the beginning of the test (state 1). Temperature distribution along the pipe at both the frozen state (t=1900 seconds) and the normal operation (t=11000 seconds) are shown in Fig. 10. This test demonstrates that the bi-porous screen evaporator design and a small amount of NCG presence enable the heat pipe to both survive and revolver from a deeply frozen condition.

Fig. 9. Freeze/thaw startup performance of the titanium water heat pipe

Fig. 10 Corresponding instantaneous temperature distributions along the heat pipe at frozen state (t= 1900 seconds) and normal operation (t=11000 seconds)

Deliverable heat pipe (with radiator) development

List of deliverables

After validating the heat transfer capability and freeze-thaw tolerance of the titanium water heat pipe prototypes, seven titanium heat pipes were further fabricated as the deliverables to NASA for further testing. Six of them have aluminum radiator panels attached. The specifications of the deliverable heat pipe radiators are listed in Table 2.  It can be seen from the table that 6 heat pipes with radiator attached are full-length (~1 m long) and 1 heat pipe without radiator attached is half-length (~60 cm long), which is used for shock and vibration testing only. As Fig. 11 depicts, the length of the heat pipe groove section depends on the interface locations with the Kilopower system:

  • HP-A and HP-B are designed to interface with a top Stirling convertor via CSAF. As 11 shows, the location of top Stirling convertor will be closer to the radiator side, so the pipe length is relatively shorter.
  • HP-C and HP-D are designed to directly interface to the cylindrical surface of the top Stirling convertor
  • HP-E and HP-F with a longer grooved pipe length will interface with the bottom Stirling convertor via CSAF.
  • HP-G is the half-length heat pipe without radiator attached, only for shock and vibration testing.

Table 2. List of deliverable heat pipes

HP No. Evaporator Type Grooved Pipe Length (cm) Radiator Attached
A CSAF Interface 113.03 Y
B CSAF Interface 113.03 Y
C Direct Interface 113.03 Y
D Direct Interface 113.03 Y
E CSAF Interface 130.18 Y
F CSAF Interface 130.18 Y
G Direct Interface 60.96 N

Fig. 11. Locations of titanium water heat pipe evaporator (blue) and the Stirling convertors of Kilopower system

Heat pipe/radiator Assembly

Thin aluminum panels (0.03 cm thick) were integrated with the condenser section of the titanium heat pipe as radiator panels. The total area of the radiator panel was designed to be 15.24 cm by 91.44 cm. To reduce the thermal expansion mismatch impact from bonding two dissimilar materials (Ti and Al), a novel active solder called S-bond was employed (S-bond Technologies, n.d.). The bonding integrity and the interface (S-bond based) had been validated by ACT in previous research (Beard, et al., 2017). After S-bonding, the surfaces of the radiators were painted with Aeroglaze A276 to increase surface radiation emissivity. The deliverable heat pipes with integrated aluminum radiator panels are shown in Fig. 12 below. After charging with water and argon, the full-length heat pipes were ready for testing in a thermal vacuum chamber.

Fig. 12 Photographs of deliverable heat pipes with radiator (left) and half-length heat pipe for shock and vibration testing (right)

Thermal vacuum chamber (TVC) test

To simulate space operation, the deliverable heat pipes with radiator were tested in a thermal vacuum chamber in a slight against gravity orientation (~0.25 cm elevation). The test setup is shown in Fig. 13. Heat was provided by a mock heater and the radiation cooling was provided by two aluminum plates with LN pipelines embedded (cold walls). The heat pipe radiator was sandwiched between two cold walls without physical contacts. In this experiment, the cold wall temperature was maintained at 300K. Locations of thermocouples (TCs) are shown in Fig. 14. Four TCs were attached to the evaporator, two TCs were attached to the adiabatic section, which is the curved portion of the pipe. TCs attached to the condenser section (# 7 to #17) were directly welded to the radiator surface instead of the heat pipe surface. The last section of the condenser, which is 3.81 cm away from the fill tube was considered to be the NCG reservoir, being thermally monitored by TCs #18, #20 and #21. An additional temperature probe (TC #19) was attached to the radiator fin tip. The test procedure is as follows:

  • Place heat radiator on the test stand with a slight adverse gravity orientation.
  • Close chamber door and evacuate the chamber until the internal pressure is lower than 1 torr.
  • Apply heat to the evaporator incrementally from 125W to 250W.
  • Maintain the average cold wall temperature at 300K through LN active cooling.

Fig. 13 Heat pipe radiator thermal performance test setup

Fig. 14. TC map of heat pipe radiator

TVC test results

Fig. 15 shows the thermal performance of HP-F, which is considered to be the most representative design because it has the longest grooved section and the most complex evaporator geometry (CSAF type), making the liquid return more challenging than others. The heat pipe radiator started up and reached the steady-state immediately as the heat was applied at t=700 seconds. Heat input was then gradually increased to 250W with the sink temperature maintained at 300K. No dry-out were observed, proving that the heat pipe radiator is able to reject 250W of heat in a space-like environment.  The temperature profiles that correspond to two steady-states are shown in Fig. 16. When Q=125W, vapor temperature (i.e. adiabatic section) is 120°C, the base temperature of the radiator (condenser) is 119.8°C and the fin tip temperature (purple line) is 95°C. The fin efficiency and the overall thermal resistance are 0.583 and 0.019 °C/W. When Q= 250W, vapor temperature increases to 180°C, the radiator base temperature is 179.4°C and the fin tip temperature is 135.4 °C. The corresponding fin efficiency and the overall thermal resistance are 0.54 and 0.023 °C/W. Note that the overall thermal resistance mentioned here includes the resistance of heat pipe, the contact resistance of S-bond joint and the resistance across the thin aluminum panel.

Fig. 15 Thermal performance test results of heat pipe radiator “F”

Fig. 16 Instantaneous temperature distributions along the heat pipe (a) heat input of 125W (t=3740 seconds) (b) heat input of 250W (t=7676 seconds)

Other heat pipe radiators (from HP-A to HP-E) were tested through the identical procedure. All heat pipes successfully rejected the nominal power in a space-like condition. The corresponding thermal resistance and the total weight of each heat pipe radiator are summarized in Table 3. It can be seen that the heaviest deliverable weights 0.73 kg. The half-length heat pipe was qualified through IR imaging as Fig. 17 shows. Uniform temperature distribution can be observed.

Table 3. Summary of HP radiator performance

HP radiator Thermal Resistance (°C/W) Total weight


A 0.075 0.695
B 0.036 0.685
C 0.057 0.665
D 0.048 0.655
E 0.019 0.730
F 0.036 0.715


Fig. 17 IR image of HP-G in a horizontal orientation (heat is applied through a heat gun from the left)


ACT successfully developed a series of titanium water heat pipes with aluminum radiator attached that are capable of cooling the space nuclear fission power system (i.e. Kilopower). These heat pipes are featured with unique bi-porous screen structure in the evaporator and axial groove structure in the adiabatic and condenser sections to accommodate various operating conditions (microgravity operation, planetary operation and launch). Through experimental testing, the heat transfer capability and freeze/thaw recovery performance of heat pipes in a slight adverse gravity orientation were demonstrated. As a final task of the development, ACT assembled 7 titanium water heat pipes, which had various evaporator designs and grooved pipe length. Six of the heat pipes were integrated with aluminum panel through S-bond technology as the radiator. Thermal performance of titanium water heat pipe radiators was tested in a thermal vacuum chamber with radiation cooling only. All six full-length heat pipe radiators successfully transported the nominal power at the working temperature of 400K in the space-simulated orientation. The thermal resistance of the heat pipe radiator showed 0.019 °C/W and the heaviest heat pipe radiator is less than 0.73 kg. The deliverables (7 titanium water heat pipes) have been sent to NASA GRC for further performance validation.


This research was sponsored by NASA Glenn Research Center under Contract NNX15CC06C. ACT would like to thank Maxwell Briggs, Marc Gibson, Jim Sanzi, and Lee Mason for their support and helpful discussions during the program. In addition, special thanks are addressed to Philip Texter, Larry Waltman and Tim Wagner who were the technicians on the program at ACT. Thanks are also due to Derek Beard and Rebecca Hay, former R&D engineers at ACT who contributed to the development of Ti-water heat pipes prototypes under this program.


Ababneh, M. T., Tarau, C. & Anderson, W. G., 2016. Hybrid Variable and Constant Conductance Heat Pipes for Lunar and Martian Environments and High Heat Flux Space Applications. Jeju, s.n.

Beard, D. et al., 2017. High Temperature Water Heat Pipes for Kilopower System. Atlanta, GA, s.n.

Beard, D., Tarau, C. & Anderson, W. G., 2017. Sodium Heat Pipes for Space and Surface Fission Power Generation. Altanta, GA, s.n.

Do, K. H., Kim, S. J. & Garimella, S. V., 2008. A mathematical model for analyzing the thermal characteristics of a flat micro heat pipe with a groove wick. International Journal of Heat and Mas Transfer , Volume 51, pp. 4637-4650.

Gibson, M. A., Oleson, S. R., Poston, D. I. & McClure, P., 2017. NASA’s Kilopower Reactor Development and the Path to Higher Power Missions. Big Sky, MT, s.n.

Gibson, M. et al., 2018. The Kilopower Reactor Using Stirling TechnologY (KRUSTY) Nuclear Ground Test Results and Lessons Learned. Cincinnati, OH, s.n.

Hay, R. & Anderson, W. G., 2015. Water-Titanium Heat Pipes for Spacecraft Fission Power. Orlando, FL, s.n.

S-bond Technologies, n.d. S-bond Technolgies. [Online]
Available at: http://www.s-bond.com/

Have a Question or Project to Discuss?