- Email: [email protected]
Hydrogen inactivation of liquid metal heat pipes
International Journal of Heat and Mass Transfer 92 (2016) 920–928
Contents lists available at ScienceDirect
International Journal of Heat and Mass Transfer journal homepage: www.elsevier.com/locate/ijhmt
Hydrogen inactivation of liquid metal heat pipes Jonas M. Leimert ⇑, Marius Dillig, Jürgen Karl Friedrich-Alexander-University of Erlangen-Nuremberg, Institute of Energy Process Engineering, Fürther Str. 244f, 90429 Nürnberg, Germany
a r t i c l e
i n f o
Article history: Received 16 February 2015 Received in revised form 15 September 2015 Accepted 18 September 2015
Keywords: Heat pipe Hydrogen Gasification Heatpipe Reformer Hydrogen permeation
a b s t r a c t When heat pipes are applied in atmospheres containing hydrogen, e.g. in presence of syngas, a small amount of hydrogen permeates through the wall of the containment into the heat pipe and accumulates as non-condensable at the end of the condenser. If the hydrogen is not removed during operation, this mechanism causes a rapid deactivation of the heat pipe starting from the condensation zone. This paper demonstrates the impacts of this deactivation on heat pipe operation and heat transfer rates and presents a formal description of the mechanism. Possible countermeasures to avoid this problem are presented and will be discussed, in particular the choice of safe operation conditions and the use of hydrogen windows based on metal membranes. The hydrogen deactivation was investigated both experimentally and theoretically. We will focus on the kinetics of the inactivation process as well as the equilibrium state when applying hydrogen windows for hydrogen removal. The results will be discussed in comparison with modelling approaches. The deactivation time from beginning of hydrogen sweep to complete inactivation of the heat pipe is in the range of 1–3 h. In the experiments a nickel hydrogen window was applied in a sodium heat pipe. With this measure the inactive length could be limited to 25–31% of the heat pipe length in hydrogen atmosphere. Ó 2015 Elsevier Ltd. All rights reserved.
1. Introduction Heat pipes are used for efficient and isothermal heating of processes in energy process engineering and in chemical industry applications requiring high heat fluxes. They consist of a metal housing containing a working fluid which can be evaporated in the so-called evaporation zone by applying heat. The vapour flows through the housing to the condensation zone where heat is removed causing the working fluid to condense. The liquid working fluid flows back to the evaporation zone driven by capillary or gravitational force. The system works isothermally as the heat transfer is only driven by evaporation. Ongoing research topics for the application of high temperature heat pipes at FAU-EVT are the carbonate looping process, biomass gasification using the Heatpipe Reformer and SOFC stacks with integrated heat pipes. For all of these applications sodium heat pipes with high-temperature steel containers are used. In processes involving gaseous hydrogen the heat pipes deactivate due to hydrogen permeation to the inner side as shown in Fig. 1. A small amount of the hydrogen from the process permeates through the container wall into the heat pipe, accumulates at the cold end of the heat pipe and the heat pipe deactivates starting from the condensation zone. Due to the constant working fluid ⇑ Corresponding author. http://dx.doi.org/10.1016/j.ijheatmasstransfer.2015.09.058 0017-9310/Ó 2015 Elsevier Ltd. All rights reserved.
circulation the hydrogen partial pressure in the active part of the heat pipe remains zero and equals the vapour pressure of the working fluid in the inactive part. When applying a cooling duty, the inactive part has a much lower temperature than the rest of the heat pipe as no working fluid can condense in this region. Heat transfer only happens through conduction and diffusion with thermal resistances several degrees of magnitude higher than in the operating heat pipe. This phenomenon is called a ‘‘cold finger” [18]. We already described the heat pipe inactivation phenomenon in previous papers on the Heatpipe Reformer in [13] and thermal management of SOFC stacks using heat pipes in [5]. During the EU-project ‘‘BioHPR” (Ref. ENK5-CT-2000-00311) the inactivation was investigated by Groll et al., who examined different coatings like ZrO2 ; Y2 O3 and oxide layers to reduce hydrogen permeability of the housing [10]. Richardson et al. proposed a methane steam reforming reactor using heat pipes for heat transfer for isothermal operation and fast load changes [24,23]. However, in bench-scale experiments he encountered a complete inactivation of the heat pipe after 19 h of operation. The re-activation of the heat pipe took 28 days due to low the low temperature in the hydrogen-filled cold finger [24]. North and Anderson discussed heat pipe inactivation as a problem for the operation of heat pipes in bimodal space nuclear systems and proposed passive systems to remove the hydrogen from the heat pipe during operation [18,19]: For the construction of a heat pipe which can work in hydrogen containing atmospheres
J.M. Leimert et al. / International Journal of Heat and Mass Transfer 92 (2016) 920–928
921
Nomenclature Latin letters A surface area, m2 c concentration, – D diffusivity, mol m1 s1 D diameter, m 1 EA activation energy, kJ mol H reduction factor, – j molar flux, mol m2 s1 l length, m n partial pressure exponent, – n_ molar flow, mol s1 P permeability, mol m1 s1 Pa0:5 P0 permeability constant, mol m1 s1 Pa0:5 p pressure, Pa vapour pressure, Pa pLV 1 R gas constant, kJ mol K1 S solubility, Pa0:5 t time, s T temperature, K x X-coordinate, m Dx wall thickness, m
Subscripts az active zone g gas H housing HP heat pipe H2 hydrogen iz inactive zone l liquid MeOx metal and oxide Me metal W hydrogen window Abbreviations HP heat pipe HPR Heatpipe Reformer MFC mass flow controller OLS ordinary least squares estimation SOFC solid oxide fuel cell SS stainless steel
Greek letters time coordinate, s
s
it is possible to either reduce the hydrogen flow into the heat pipe or enhance the flow leaving the heat pipe using a so-called ‘‘Hydrogen Window”. For the first approach the container should comprise of a material with a very low hydrogen permeability. Also, the operating temperature of the heat pipe plays an important role as the hydrogen permeation is strongly temperature dependent. Unfortunately, in most applications the operating temperature can not be chosen freely. It is also possible to use coatings which reduce the hydrogen solubility on the surface [8,18]. The only part where the partial pressure of the hydrogen inside the heat pipe is high enough for an efficient removal is the inactive part. The hydrogen flow out of the heat pipe is greatly diminished because of the lower temperature of the container due to the deactivation. The construction of the hydrogen window should therefore keep the temperature of the material as high as possible and use a material with a high hydrogen permeability, e.g. nickel alloys or palladium.
2. Materials and methods It is important to investigate the mechanisms of hydrogen transportation through dense materials as well as the working principles of heat pipes in order to design a hydrogen tolerant heat pipe. Heat pipes consist of the metal container for the heat transfer fluid, made of a steel pipe, a mesh structure on the inner side of the tube for the transport of the condensed fluid to the evaporation zone and the heat transfer fluid itself.
2.1. Fundamentals of heat pipes Heat pipes are passive heat exchangers which rely on a evaporation–condensation mechanism: When heat is supplied to the one side the working fluid evaporates and flows to the cold end where it condenses. The condensate is transferred back to the evaporation zone by gravitational or capillary force. The operating pressure of the heat pipe equals therefore the vapour pressure of the working fluid pLV at the respective operating temperature T HP . The heat transfer of a heat pipe can be limited by the properties of the working fluid, heat pipe geometry and capillary structure, these limitations are discussed thoroughly in [6]. For the following considerations it is assumed that the heat transfer is only limited by radiation and convection on the heat pipe outer surface and therefore the active length of the heat pipe.
2.2. Hydrogen permeation through metals
Fig. 1. Scheme of the inactivation of heat pipes due to hydrogen permeation.
The permeation of diatomic gases like hydrogen through a dense body is divided into several steps: The adsorption of the hydrogen at the surface, the dissociation into protons and dissolution into the metal body, which is sometimes referred to as solution of the hydrogen. Then the diffusion through the metal in form of protons, which is in most cases the rate-limiting step of
922
J.M. Leimert et al. / International Journal of Heat and Mass Transfer 92 (2016) 920–928
the whole process and finally the recombination and desorption from the solid [16,21,26]. The dissolution of a gas in a solid body can be described using Sievert’s law [21]:
ci ¼ S pni
ð1Þ
In this equation ci is the concentration of the dissolved component, S the solubility and pi the partial pressure of the component in the gas phase. For a diatomic gas like hydrogen the exponent is n ¼ 0:5 as the gas dissociates after the solution in the metal. The resulting molar flux jH2 can be calculated using Fick’s first law [21] assuming a linear concentration profile:
jH2 ¼ D
dc Dc ¼ D dx Dx
ð2Þ
Eq. (1) can be applied to this equation for the concentration of the dissolved component. The resulting equation only depends on the partial pressure of the hydrogen on both sides of the solid wall p1 resp. p2 , the wall thickness of the membrane Dx and the coefficients of solubility and diffusion S and D:
jH 2 ¼
DS ðp n pn2 Þ Dx 1
ð3Þ
The permeability P, which is the product of the coefficients of solubility and diffusion is used more often in literature. To calculate the molar flow of hydrogen n_ H2 the resulting equation has to be multiplied with the area of the solid A:
n_ H2 ¼
PA n ðp pn2 Þ Dx 1
ð4Þ
As mentioned above, the partial pressure exponent n is assumed to be 0:5. However, in some cases also values ranging from 0.5 to 1 were found experimentally due to limitations in the dissolution. This can be caused by surface contaminations or oxide layers. For the following calculations the exponent was assumed to be 0.5 as this value is commonly used in literature, which allows easy comparability of the results. The calculation of the hydrogen molar flow in the presence of dense oxide layers is discussed in [16]. The reduced permeability can be described by using the reduction factor H.
H¼
n_ Me _nMeOx
ð5Þ
Möllenhoff reports about oxide layers on austenitic steel 1.4876 which reduced the permeation of hydrogen by a factor of up to 520 [16,17]. For the operation of heat pipes in hydrogen atmosphere such a dense oxide layer would influence heat pipe behaviour significantly [15]. The permeability itself is a temperature activated property and can be calculated with the Arrhenius equation using the permeability constant P 0 , permeability activation energy EA and ideal gas constant R:
EA
P ¼ P0 e RT
ð6Þ
Table 1 shows a collection of permeability constants and activation energies for possible heat pipe housing and hydrogen window materials. 2.3. Material considerations The hydrogen tolerant heat pipe comprises of the housing, the working fluid and the hydrogen window. In the following section the main features of these components are discussed in regards of their suitability in a heat pipe working in hydrogen atmosphere. Additionally, a model was developed to describe the time dependence and equilibrium state of the inactivation procedure. 2.3.1. Working fluid As discussed above, for the temperature regime from 600–1000 °C the working fluids of choice are mostly liquid alkali metals like lithium, sodium or potassium. As the vapour pressure of these metals rises with their position in the periodic table, lithium is suitable for higher temperatures than sodium or potassium. For the removal of hydrogen from the heat pipe the internal pressure of the hydrogen buffer, which equals the vapour pressure of the working fluid, is of importance as it enhances the hydrogen permeation out of the heat pipe. 2.3.2. Housing The housing of the heat pipe is normally made of high temperature steel, sometimes also refractory metals like Tantalum, Titanium or Tungsten are discussed because of their higher strength which allows lower wall thicknesses [7,18]. Because of the higher price compared to steel housings these materials are not considered for an application in process engineering. A serious problem for steel housings in the desired temperature regime is sigma phase embrittlement. The buildup of the sigma phase leads to a decrease of the notched impact strength of the affected steels of up to 80% [22]. The formation of the sigma phase depends mainly on the composition of the steel regarding the three main components, which are chromium, iron and nickel. Nickel base alloys with nickel contents of more than 40% are not affected by the sigma phase, while nickel contents below 11% can also lead to sigma phase embrittlement due to nickel rich clusters in the steel [12,25]. The hydrogen permeability is of course also affected by the main components and especially by the nickel content as nickel has a comparatively high permeability for hydrogen: As a result, nickel base alloys show a higher permeability than steels with iron as main component [2]. Fig. 2 supports this finding as the permeability of most literature values are directly proportional to their nickel content. For the application in hydrogen rich atmosphere heat pipe housings should therefore not comprise steels with a high nickel content. This poses an optimisation problem for the
Table 1 Permeation rates of different materials for high temperature heat pipe or hydrogen applications. P 0 [mol=m s Pa0:5 ]
P at 800 °C [mol=m s Pa0:5 ]
Refs.
Material
Temperature range [K]
303 SS
773–1173
3:6010
7
67,000
1:9710
10
[9]
304 SS
823–1173
5:10107
71,000
1:781010
[9]
316 SS
500–930
7
2:3610
63,496
2:021010
[30]
Inconel
500–930
9:30107
69,000
4:071010
[9]
Palladium
370–900
8:11107
15,464
1:43107
[28]
Tantalum
253–944
5:80109
20,203
5:58108
[28]
Nickel
673–1123
4:65107
55,204
9:551010
[9]
Tungsten
1100–2400
7:60107
1,321,923
1:071013
[28]
EA [J=mol]
J.M. Leimert et al. / International Journal of Heat and Mass Transfer 92 (2016) 920–928
923
The easiest approach for the hydrogen window is to apply nickel tubes directly into the gas channel of the heat pipe at the cold end of the condenser as shown in Fig. 4. The hydrogen will be collected at this position of the heat pipe due to the vapour flow of the working fluid. The disadvantage of this concept is that the heat pipe and the hydrogen window will cool down partially resulting in a ‘‘cold finger” and lower heat transfer rates due to the formation of an inactive zone. This leads to a decrease in hydrogen permeability of the hydrogen window in the inactive zone as the permeation is a temperature activated property. The length of the inactive zone can be influenced by the geometry of the hydrogen window. 2.4. Mathematical model for hydrogen inactivation Fig. 2. Permeability of stainless and nickel base steels as a function of the nickel content (temperature: 850 °C; permeability measurements from [9,14,27,30]).
container material as steels with low nickel content tend to sigma phase embrittlement. For the experiments shown in this work the steel Sandvik 253 MA (material number: 1.4835) was used. It combines low formation of sigma phase as it also shows 1% sigma phase after 2000 working hours at 800°C with a low nickel content of only 11%. The exact composition of the steel is shown in Table 2. 2.3.3. The hydrogen window The hydrogen permeating into the heat pipe during operation has to be removed continuously. One possibility is the construction of an open system where the hydrogen can be sucked out of the heat pipe as proposed by North and Anderson in [19]. This solution ensures the operation of the whole heat pipe without the buildup of a cold finger. The disadvantage is the increased effort in control and safety engineering as the potentially hazardous working fluid must not be sucked out of the heat pipe. Additionally alkali metals tend to plug small tubes when cooled down. Furthermore, alkali metals form solid hydrides in the presence of gaseous hydrogen in the temperature regime of about 400 °C, which causes severe plugging problems. Alkali hydride formation in heat pipe operation is also discussed in the experimental section. For these reasons the focus in this paper lies on passive systems which remove the hydrogen by permeation. For an efficient removal of the hydrogen the hydrogen window must consist of a material with high hydrogen permeability. The materials of choice are palladium, tantalum or nickel because, as shown in Table 1, their permeability is at least one order of magnitude higher than that of steels which are used for the housing. There are also some restrictions regarding corrosion and material compatibility: Palladium is not compatible with alkali metals in this temperature regime [18], Tantalum can withstand the alkali metals but is attacked by traces of oxygen which leads to material failure [20]. In contrast to those materials, the attack of sodium on nickel is moderate in the range of 2,5–25 lm/a [4]. Because of its low price and high durability, nickel will be used in the following experiments. Table 2 Composition of the housing material 253MA (1.4835) according to inspection certificate. Element
Mass fraction [%]
Carbon Silicium Manganese Phosphorus Chromium Nickel Cerium
0.075 1.50 0.59 0.022 20.94 10.86 0.04
In the following part a model for the inactive length of a heat pipe exposed to hydrogen is developed. This can be used in further works as a layout tool for the dimensioning of a hydrogen window. Additionally a model of the inactive length of a heat pipe as a function of time will be shown. To describe the deactivation of the heat pipe, a few assumptions have to be made, which are also illustrated in Fig. 1: The hydrogen will only enter the heat pipe in the active zone as the temperature of the housing is significantly higher in this part of the heat pipe. Furthermore the partial pressure of the hydrogen is zero as the hydrogen will be transported to the inactive zone immediately after permeation. The heat pipe is isothermal with one discrete temperature each in the active and inactive zone. The hydrogen in the inactive zone has the same temperature as the heat pipe working fluid, i.e. the working temperature of the heat pipe. This condition holds for heat pipes with a small inactive zone. In a real system the hydrogen temperature will be in between the heat pipe operating temperature and the temperature of the housing in the inactive zone depending on the heat transfer conditions. The hydrogen leaves the heat pipe only via the hydrogen window as the permeability of the window is at least one order of magnitude higher than that of the steel housing. Hydrogen solubility in liquid sodium has been neglected. According to Wittingham [31] hydrogen solubility in sodium at approx. 0.56 bar, is only 110 ppm and therefore not considered in dynamic considerations. Formation of alkali hydrides is discussed in the corresponding section. Assuming molar fluxes of hydrogen entering through the steel housing and leaving through the hydrogen window, n_ H respectively n_ W and an accumulated amount of hydrogen inside the heat pipe nacc , the hydrogen balance of the heat pipe can be developed as follows:
n_ H n_ W ¼
dnacc dt
ð7Þ
The molar flow of hydrogen through the steel housing and the hydrogen window can be calculated using (4):
n_ H ¼ P H
AH 0:5 pH2 p0:5 H2 ;HP D xH
n_ W ¼ PW
AW 0:5 pH2 ;iz p0:5 H2 ;permeate DxW
ð8Þ ð9Þ
These equations can be simplified by assuming that the partial pressure of hydrogen in the active zone of the heat pipe pH2 ;HP is zero as the hydrogen will be transported to the inactive zone immediately after permeation. This simplification is also valid for
924
J.M. Leimert et al. / International Journal of Heat and Mass Transfer 92 (2016) 920–928
the pressure on the outer side of the hydrogen window pH2 ;permeate as it is flushed with inert gas to remove the permeated hydrogen. The hydrogen pressure in the inactive zone pH2 ;iz equals the vapour pressure of the working fluid pLV at the operating temperature of the heat pipe. The area of the housing responsible for the hydrogen permeation AH;az depends on the length of the inactive zone liz , the diameter of the housing DH and the overall length of the heat pipe lHP . It is assumed that hydrogen only permeates in the active zone of the heat pipe as the permeation is much lower in the colder inactive zone and the driving pressure is also lower due to the hydrogen on the inside.
AH;az ¼ pDH ðlHP liz Þ
ð10Þ
The area of the hydrogen window where hydrogen permeates out of the heat pipe AW;iz can be calculated accordingly. It is assumed that the hydrogen only permeates through the hydrogen window:
AW;iz ¼ pDW liz
ð11Þ
The accumulated hydrogen in the heat pipe corresponds to an inactive volume, which can be calculated using the ideal gas equation. The hydrogen temperature and pressure are assumed as the heat pipe working temperature and the working fluid vapour pressure, respectively:
nacc
VH p ¼ 2 LV RT HP
ð12Þ
Assuming this hydrogen volume is always located at the top end of the condenser where it forms the inactive zone, the inactive length of the heat pipe can be calculated using the cylindrical geometry:
liz ¼
V H2 p ðD 4
H
2xH Þ2
ð13Þ
2.4.1. Model for time-dependence of inactivation The inactivation of the heat pipe with respect to the time can be calculated by integrating (7) over time by numerical integration:
Z nacc ðtÞ ¼ 0
t
n_ H ðsÞ n_ W ðsÞdðsÞ
ð14Þ
2.4.2. Model for inactive length The equilibrium state at t ! 1 of the inactive length can also be estimated. In equilibrium state the change in accumulated hydrogen becomes zero, resulting in a simple equation:
n_ H n_ W ¼
dnacc ¼0 dt
ð15Þ
The inactive length of the heat pipe can then be expressed as:
DxW PH DH p0;5 liz H2 ¼ 0;5 lHP DxH PW DW p0;5 LV þ DxW P H DH pH2
ð16Þ
If no hydrogen window is used, the hydrogen window permeability may be set to zero or to the permeability of the steel housing if it is used as hydrogen window. The inactive length was calculated for two cases: A hydrogen window tube 3 0.3 mm using Nickel and degassing only over the heat pipe housing made of stainless steel SS304 (material number: 1.4301) with permeability values from Gorman and Nardella [9]. The length of the hydrogen window was 400 mm. Fig. 3 shows the modelling results with DT being the temperature difference between active and inactive zone. The inactive length is decreasing with the heat pipe operating temperature as the heat pipe inner
Fig. 3. Equilibrium calculation of the inactive length of sodium heat pipes with degassing via housing or with a nickel hydrogen window at 1 bar and 10 bar hydrogen pressure; DT: temperature difference between active and inactive zone; DH = 33.7 mm, xH = 3.2 mm, DW = 3 mm, xW = 0.3 mm, LW = 400 mm.
pressure is rising. The temperature of the cold end has a high impact on the inactive length which stresses the importance of a good isolation on the cold end. The Nickel hydrogen window decreases the inactive length by a factor of about two. However, with a low temperature drop in the inactive zone a high active length is also possible without hydrogen window. This could be achieved by insulation or supplementary heating of the inactive zone. 3. Experimental For the study of the deactivation phenomena a heat pipe test rig was erected at FAU-EVT. It can be used for heat pipes with an outer diameter of up to 50 mm and a length of up to 1500 mm. The heating is supplied by a Sandvik furnace with a heat duty of 5 kW. The heat pipe is fixed by four steel clamps. It is possible to change the operating angle of the heat pipe, e.g. for measurements of maximum heat transfer. The heat pipe is provided with an enclosure of 500 mm to allow different gas atmospheres for the testing procedure, see Fig. 4. The inactivation is measured by six thermocouples which detect the wall temperature of the heat pipe. The heat pipe housing is made of Sandvik 253 MA with three layers of steel mesh made of 1.4841. The mesh has a wire thickness of 0.160 mm and a wire space of 0.200 mm. The upper ending of the heat pipe has a compression fitting so that different materials for degassing can be tested. The heat pipe can be evacuated by a steel pipe leading to a ball valve and a vacuum pump. The heat pipe has an overall length of 1200 mm with an evaporation zone of 500 mm, an adiabatic zone of 200 mm and a condensation zone of 500 mm. At the beginning of the experiments the heat pipe was heated to the desired operation temperature. When stationary state was reached, the double wall of the heat pipe was flushed with hydrogen. The deactivation of the heat pipe was recorded using six thermocouples type K. After the whole condensation zone of the heat pipe was deactivated the reactivation procedure was started by flushing with nitrogen. 4. Results and discussion The results are divided into three parts: First the results from the deactivation measurements are shown with a comparison to
J.M. Leimert et al. / International Journal of Heat and Mass Transfer 92 (2016) 920–928
925
Fig. 6. Measured inactive length plotted against deactivation times at operating temperature of 850 °C with OLS permeability fit of the modelling approach.
Fig. 4. Piping and instrumentation diagram of the heat pipe test rig and sketch of the nickel hydrogen window; DH : Diameter of heat pipe housing; DW : Diameter of hydrogen window.
the modelling described in the previous part. The second part will be our results on a nickel hydrogen window, which was placed inside a sodium heat pipe and tested for more than 1000 h without failure. As a last topic we want to show the impact of sodium hydride formation on the hydrogen-alkali metal heat pipe system.
4.1. Hydrogen inactivation In the first measurement campaign, the hydrogen impact without hydrogen window was studied. For this reason the double wall of the heat pipe was flushed with pure hydrogen at 1 bar which resulted in complete inactivation. The temperature of the thermocouples was recorded, Fig. 5 shows a typical result with normalised temperatures. For the comparison of different experiments the deactivation time was determined as shown in Fig. 5. In a second step the corresponding inactive length was plotted against the deactivation time. The modelling approach was then used to fit a curve over the generated deactivation points using OLS approximation. The result is shown in Fig. 6. It appears that the experimental data can be described using the developed model. The model can be adjusted to the measured points using
Fig. 5. Exemplary temperature data from inactivation measurements at operation temperature of 850 °C with determination of inactivation time.
only the housing permeability. The increasing deactivation time of the test runs is probably tied to the growth of an oxide layer on the steel housing reducing the permeability of the housing by a factor of up to 4.5. According to various sources, the growth rate and density of the layer is based on the gas atmosphere and especially on the oxygen content [11,15,17]. Before the fourth test run the heat pipe was reactivated using 5% H2 in N2 instead of technical N2 , which lead to an increase in deactivation time. It is assumed that the lower O2 partial pressure caused a formation of very dense oxide layers as described in [15] which slowed down inactivation.
4.2. Hydrogen window In a second measurement campaign a 3 0.3 mm nickel tube with a length of 500 mm was mounted inside the heat pipe. It acts as a hydrogen window in the cold finger, which can be modelled using the described approach. The tube is flushed with nitrogen to remove the hydrogen permeating out of the heat pipe. The active length of the heat pipe can be measured directly in this setup with a thermocouple inserted into the hydrogen window. The active length of the heat pipe is measured without hydrogen and after reaching steady state operation in hydrogen atmosphere. The active length is the difference of these two values. Any errors occurring in the temperature measurement due to thermal conduction are avoided as these errors will occur at both measurements. Fig. 7 shows the results from the hydrogen window experiments. The hydrogen pressure on the outer side of the heat pipe
Fig. 7. Inactive length of a sodium heat pipe in hydrogen atmosphere with a 3 0.3 mm nickel hydrogen window; Antoine parameters from [1].
926
J.M. Leimert et al. / International Journal of Heat and Mass Transfer 92 (2016) 920–928
was kept constant at 1 bar. The inner pressure of the heat pipe which equals the vapour pressure of the working fluid was recorded directly during the measurements. For a comparison the saturated liquid line of pure sodium was calculated using Antoine parameters given in [1]. The measured values are in good agreement with the calculated inactivation process. Small deviations can be explained with impurities in the used sodium. The displayed permeability of the housing was taken from the deactivation measurements. The inactive length is rising slightly with the operating temperature as the higher housing permeability overcompensates the higher operating pressure. The heat pipe has a relative inactive length of 25–31% depending on the operating temperature. In order to reduce the inactive length, the geometry of the hydrogen window can be changed using higher diameters or lower wall thicknesses. However, the wall thickness is limited by the corrosion rate of nickel in liquid sodium which lies in the range of 2,5– 25 lm/a and the weldability of the materials [4]. The housing permeability has a high impact on the inactive length, it can be influenced by targeted oxidation of the steel using different gas atmospheres with low oxygen partial pressures. The permeability can be lowered by up to three orders of magnitude resulting in a much lower inactive length [15,16]. The hydrogen window was tested for approx. 1500 h inside the sodium heat pipe. After completion of the tests, the Nickel tube was demounted and cleaned of residual sodium. The tube diameter was measured and compared to a new nickel tube, which showed a decrease in diameter of 10–15 lm which corresponds to a corrosion rate of 9 106 cmmg 2 h. 4.3. Formation of alkali hydrides Sodium hydride is an ionic product of the reaction of molecular hydrogen with (liquid) sodium. It is only stable in its solid phase, up to an upper temperature limit depending on ambient pressure (e.g. 425 °C at ambient pressure for Na).
NaðlÞ þ 1=2 H2 NaHðsÞ
ð17Þ
Formation of other alkali metal hydrides is similar. Enthalpies of hydride formation of relevant alkali metals Li, Na and K are DHf ;LiðsÞ ¼ 77; 71 kJ/mol, DHf ;Na ¼ 56; 4 kJ/mol and DHf ;K ¼ 57; 82 kJ/mol respectively [3]. For lithium also a liquid LiH phase exists due to the more elevated temperatures. Detailed phase diagrams for binary alkali metal – hydrogen systems have been computed based on data from [29] and are displayed in Fig. 8. The temperature of the active heat pipe zone determines the heat pipe internal pressure. In case that the temperature in the inactive zone of the heat pipe drops below the formation temperature of the alkali hydride, the conversion of the liquid alkali metal
Fig. 8. Phase diagrams of binary alkali metal – hydrogen system in heat pipes; calculated with data from [29]. Exemplary determination of NaH formation limit (=approx. 400 °C) in cold finger for HP operated at 800 °C in active zone.
Fig. 9. Basic mechanism of metal hydride induced deactivation of high temperature heat pipes.
starts and hydrogen of the hydrogen buffer is consumed. This initiates an opposing trend to heat pipe deactivation and thus may lead to a quasi equilibrium situation as long as elementary alkali metals are still available. Ongoing consumption of the alkali metal leads to a slow blocking of the wick and to a complete deactivation of the heat pipe. Fig. 9 illustrates this phenomenon. A decomposition of the hydrides i.e. a complete reactivation of the alkali metal heat pipe is possible by heating the heat pipe over the formation limit. However, heating the hydrates above the decomposition temperature suddenly releases large amounts of hydrogen. The released hydrogen instantly floods the heat pipe and deactivates it nearly completely. This can result in material failures due to overheating of the metal container as the evaporating of the working fluid might be suppressed by the flooding with hydrogen. For continuous operation of heat pipes in hydrogen atmospheres this hydride formation limit has to be avoided. System design has to avoid cooling below the limit, even if some degree of hydrogen deactivation of the heat pipe is permitted. In experimental studies hydride formation leads to misinterpretation of the inactivation time and length. It is important to assure that the critical temperature is avoided by using supplemental heating of heat pipes in the inactive zone. The sodium hydride formation was investigated in a measurement campaign using planar heat pipes shaped as a flat rectangular plate. The detailed design and setup is described in [5], the heat pipe dimensions were 270 120 4 mm (L W H), the capillary structure was composed of two layers of mesh 80 on the upper as well as lower side of the flat heat pipe. One layer of stainless steel mesh 8 provides the open space for the vapour backflow within the heatpipe. Fig. 10 shows the temperature profile of the cold finger of a planar heat pipe in hydrogen atmosphere at 1 bar. The deactivation progresses very fast until the coldest spot of the heat pipe reaches the NaH formation temperature of 400 °C. After that the
Fig. 10. Temperature profiles during initial heat pipe deactivation and starting of NaH formation.
J.M. Leimert et al. / International Journal of Heat and Mass Transfer 92 (2016) 920–928
927
Acknowledgements This research is carried out in the framework of the Bavarian Hydrogen Center (BHC) joint research program. The authors would like to acknowledge the support provided by the Bavarian State Ministry of Science, Research and the Arts.
References
Fig. 11. Dynamics of hydrogen deactivation with NaH formation, solid black line shows cold end temperature of heat pipe, data points indicated inactive length of heat pipe in relation to entire heat pipe length.
deactivation nearly remains constant for up to 56 h. In this time the hydrogen is consumed by the working fluid. Fig. 11 also illustrates this result: The relative inactive length rises to up to 0.4 in the first three hours of deactivation, after that NaH is formed and deactivation decelerates. The instationary behaviour of the cold end temperature seen from 4 to 10 h can also be traced back to NaH formation: As the formation temperature is reached a large amount of hydrogen is consumed by NaH which leads to reactivation of a part of the heat pipe. This raises the cold end temperature over the decomposition temperature and results in a release of the hydrogen. The experiments described before were conducted in cylindrical heat pipes avoiding NaH formation by heating the cold end constantly to 500 °C. 5. Conclusion Hydrogen inactivation is a serious problem whenever heat pipes are used in atmospheres containing hydrogen. In this paper the underlying mechanisms of the inactivation were explained and measures to limit or avoid inactivation were discussed. The deactivation time of a heat pipe made of stainless steel which is typically used in the Heatpipe Reformer, where hydrogen inactivation is a main design challenge according to [15], is in the range of 100–200 min. This could result in a mandatory cyclic operation of the heat pipe and thus the entire system, where the operation of the reactor or heat exchanger is stopped and the apparatus is flushed with steam to remove hydrogen from the heat pipe. A hydrogen window made of nickel was tested for over 1500 h without failure and resulted in a relative inactive length of 25–31% rising with the operating temperature. The inactive length could be reduced by rising the temperature of the inactive zone or changing the geometry of the hydrogen window. Further significant influence parameters to deactivation are the formation of alkali hydrides and oxide layer growth on the steel housing depending on external gas atmospheres. Ongoing research at the Institute of Energy Process Engineering focuses on advanced degassing setups that allow continuous operation of liquid metal heat pipes in hydrogen atmosphere with high active lengths.
Conflict of interest None declared.
[1] P. Browning, P.E. Potter, An assessment of the experimentally determined vapour pressures of the liquid alkali metals, in: Handbook of Thermodynamic and Transport Properties of Alkali Metals, 1985, 349–358. [2] P. Calderoni, M. Ebner, R. Pawelko, High-temperature hydrogen permeation in nickel alloys, in: Proceedings of HTR, 2010. [3] M.W. Chase, NIST-JANAF Thermochemical Tables, 1998, Part I Al-Co. [4] C.F. Cheng, W.E. Ruther, Corrosion of nickel in liquid sodium, Corrosion 28 (1) (1972) 20–22.. [5] M. Dillig, J. Leimert, J. Karl, Planar high temperature heat pipes for SOFC/SOEC stack applications, Fuel Cells 14 (2014) 479–488. [6] P.D. Dunn, D.A. Reay, Heat Pipes, 4th ed., Elsevier Ltd, Oxford, 1994. ISBN 00804190387. [7] El-Genk, S. Mohamed, Tournier, P. Jean-Michel, Uses of liquid–metal and water heat pipes in space reactor power systems, Front. Heat Pipes 2 (1) (2011), http://dx.doi.org/10.5098/fhp.v2.1.3002 (März, ISSN 2155-658X). [8] A.V. Golubeva, A.V. Spitsyn, M. Mayer, D.I. Cherkez, T. Schwarz-Selinger, F. Koch, S. Lindig, A.A. Skovoroda, Gas-driven hydrogen permeation through tungsten-coated graphite, J. Nucl. Mater. 415 (1) (2011) S688–S691, http://dx. doi.org/10.1016/j.jnucmat.2011.01.106. ISSN 00223115. [9] J.K. Gorman, W.R. Nardella, Hydrogen permeation through metals, Vacuum 12 (1961) (1962) 19–24. [10] M. Groll, BioHPR report summary: hydrogen diffusion data for different coatings. Version: 2005, 2005, Forschungsbericht. . [11] E. Heimes, Untersuchung verschiedener Voroxidationsverfahren und ihre Wirkung als Wasserstoff Permeations-Barriere, 1986. [12] C. Hsieh, W. Wu, Overview of intermetallic sigma () phase precipitation in stainless steels, Int. Scholarly Res. Not. Metall. (2012) 16. . [20] D. Quataert, C.A. Busse, Investigation of the corrosion mechanism in tantalum– lithium high temperature heat pipes by ion analysis, J. Nucl. Mater. 46 (1973) 329–340. . [21] P. Quicker, V. Höllein, R. Dittmeyer, Catalytic dehydrogenation of hydrocarbons in palladium composite membrane reactors, Catal. Today 56 (2000) 21–34. [22] K. Richard, G. Petrich, Die Bedeutung der Sigmaphasenversprödung von Chrom und Chrom-Nickel-Stählen für den Chemie-Apparatebau, Chem. Ing. Tech. 1 (1963) 29–36, http://dx.doi.org/10.1002/cite.330350107. [23] J.T. Richardson, A gas fired heat-pipe reformer for small-scale hydrogen production, Stud. Surf. Sci. Catal. 107 (1997) 567–571 (, ISBN 0444823522). [24] J.T. Richardson, S.A. Paripatyadar, J.C. Shen, Dynamics of a sodium heat pipe reforming reactor, AIChE J. 34 (5) (1988) 743–752 (ISBN 1547-590). [25] V.G. Rivlin, G.V. Raynor, Critical evaluation of constitution of chromium–iron– nickel system, Int. Metals Rev. 1 (1980) 21–40. [26] C. Schäfer, Diffusionseigenschaften bestimmter Metalle bei der Hochtemperatur-Wasserstoffabtrennung, Technische Universität München, Diss., 2010.
928
J.M. Leimert et al. / International Journal of Heat and Mass Transfer 92 (2016) 920–928
[27] A.S. Schmidt, F. Verfuss, E. Wicke, Studies on the permeation of hydrogen and tritium through heat resistant alloys, J. Nucl. Mater. 131 (1985) 247–260., . [28] S.A. Steward, Review of hydrogen isotope permeability through materials, 1983. .
[29] D.R. Stull, H. Prophet, JANAF thermochemical tables, 1971.. [30] E.H. Van Deventer, V.A. Maroni, Hydrogen permeation characteristics of some austenitic and nickel-base alloys, J. Nucl. Mater. 92 (1) (1980) 103–111, http:// dx.doi.org/10.1016/0022-3115(80)90146-4 (ISSN 00223115). [31] A.C. Whittingham, An equilibrium and kinetic study of the liquid sodium– hydrogen reaction and its relevance to sodium-water leak detection in LMFBR systems, J. Nucl. Mater. 60 (1976) 119–131. .
Contents lists available at ScienceDirect
International Journal of Heat and Mass Transfer journal homepage: www.elsevier.com/locate/ijhmt
Hydrogen inactivation of liquid metal heat pipes Jonas M. Leimert ⇑, Marius Dillig, Jürgen Karl Friedrich-Alexander-University of Erlangen-Nuremberg, Institute of Energy Process Engineering, Fürther Str. 244f, 90429 Nürnberg, Germany
a r t i c l e
i n f o
Article history: Received 16 February 2015 Received in revised form 15 September 2015 Accepted 18 September 2015
Keywords: Heat pipe Hydrogen Gasification Heatpipe Reformer Hydrogen permeation
a b s t r a c t When heat pipes are applied in atmospheres containing hydrogen, e.g. in presence of syngas, a small amount of hydrogen permeates through the wall of the containment into the heat pipe and accumulates as non-condensable at the end of the condenser. If the hydrogen is not removed during operation, this mechanism causes a rapid deactivation of the heat pipe starting from the condensation zone. This paper demonstrates the impacts of this deactivation on heat pipe operation and heat transfer rates and presents a formal description of the mechanism. Possible countermeasures to avoid this problem are presented and will be discussed, in particular the choice of safe operation conditions and the use of hydrogen windows based on metal membranes. The hydrogen deactivation was investigated both experimentally and theoretically. We will focus on the kinetics of the inactivation process as well as the equilibrium state when applying hydrogen windows for hydrogen removal. The results will be discussed in comparison with modelling approaches. The deactivation time from beginning of hydrogen sweep to complete inactivation of the heat pipe is in the range of 1–3 h. In the experiments a nickel hydrogen window was applied in a sodium heat pipe. With this measure the inactive length could be limited to 25–31% of the heat pipe length in hydrogen atmosphere. Ó 2015 Elsevier Ltd. All rights reserved.
1. Introduction Heat pipes are used for efficient and isothermal heating of processes in energy process engineering and in chemical industry applications requiring high heat fluxes. They consist of a metal housing containing a working fluid which can be evaporated in the so-called evaporation zone by applying heat. The vapour flows through the housing to the condensation zone where heat is removed causing the working fluid to condense. The liquid working fluid flows back to the evaporation zone driven by capillary or gravitational force. The system works isothermally as the heat transfer is only driven by evaporation. Ongoing research topics for the application of high temperature heat pipes at FAU-EVT are the carbonate looping process, biomass gasification using the Heatpipe Reformer and SOFC stacks with integrated heat pipes. For all of these applications sodium heat pipes with high-temperature steel containers are used. In processes involving gaseous hydrogen the heat pipes deactivate due to hydrogen permeation to the inner side as shown in Fig. 1. A small amount of the hydrogen from the process permeates through the container wall into the heat pipe, accumulates at the cold end of the heat pipe and the heat pipe deactivates starting from the condensation zone. Due to the constant working fluid ⇑ Corresponding author. http://dx.doi.org/10.1016/j.ijheatmasstransfer.2015.09.058 0017-9310/Ó 2015 Elsevier Ltd. All rights reserved.
circulation the hydrogen partial pressure in the active part of the heat pipe remains zero and equals the vapour pressure of the working fluid in the inactive part. When applying a cooling duty, the inactive part has a much lower temperature than the rest of the heat pipe as no working fluid can condense in this region. Heat transfer only happens through conduction and diffusion with thermal resistances several degrees of magnitude higher than in the operating heat pipe. This phenomenon is called a ‘‘cold finger” [18]. We already described the heat pipe inactivation phenomenon in previous papers on the Heatpipe Reformer in [13] and thermal management of SOFC stacks using heat pipes in [5]. During the EU-project ‘‘BioHPR” (Ref. ENK5-CT-2000-00311) the inactivation was investigated by Groll et al., who examined different coatings like ZrO2 ; Y2 O3 and oxide layers to reduce hydrogen permeability of the housing [10]. Richardson et al. proposed a methane steam reforming reactor using heat pipes for heat transfer for isothermal operation and fast load changes [24,23]. However, in bench-scale experiments he encountered a complete inactivation of the heat pipe after 19 h of operation. The re-activation of the heat pipe took 28 days due to low the low temperature in the hydrogen-filled cold finger [24]. North and Anderson discussed heat pipe inactivation as a problem for the operation of heat pipes in bimodal space nuclear systems and proposed passive systems to remove the hydrogen from the heat pipe during operation [18,19]: For the construction of a heat pipe which can work in hydrogen containing atmospheres
J.M. Leimert et al. / International Journal of Heat and Mass Transfer 92 (2016) 920–928
921
Nomenclature Latin letters A surface area, m2 c concentration, – D diffusivity, mol m1 s1 D diameter, m 1 EA activation energy, kJ mol H reduction factor, – j molar flux, mol m2 s1 l length, m n partial pressure exponent, – n_ molar flow, mol s1 P permeability, mol m1 s1 Pa0:5 P0 permeability constant, mol m1 s1 Pa0:5 p pressure, Pa vapour pressure, Pa pLV 1 R gas constant, kJ mol K1 S solubility, Pa0:5 t time, s T temperature, K x X-coordinate, m Dx wall thickness, m
Subscripts az active zone g gas H housing HP heat pipe H2 hydrogen iz inactive zone l liquid MeOx metal and oxide Me metal W hydrogen window Abbreviations HP heat pipe HPR Heatpipe Reformer MFC mass flow controller OLS ordinary least squares estimation SOFC solid oxide fuel cell SS stainless steel
Greek letters time coordinate, s
s
it is possible to either reduce the hydrogen flow into the heat pipe or enhance the flow leaving the heat pipe using a so-called ‘‘Hydrogen Window”. For the first approach the container should comprise of a material with a very low hydrogen permeability. Also, the operating temperature of the heat pipe plays an important role as the hydrogen permeation is strongly temperature dependent. Unfortunately, in most applications the operating temperature can not be chosen freely. It is also possible to use coatings which reduce the hydrogen solubility on the surface [8,18]. The only part where the partial pressure of the hydrogen inside the heat pipe is high enough for an efficient removal is the inactive part. The hydrogen flow out of the heat pipe is greatly diminished because of the lower temperature of the container due to the deactivation. The construction of the hydrogen window should therefore keep the temperature of the material as high as possible and use a material with a high hydrogen permeability, e.g. nickel alloys or palladium.
2. Materials and methods It is important to investigate the mechanisms of hydrogen transportation through dense materials as well as the working principles of heat pipes in order to design a hydrogen tolerant heat pipe. Heat pipes consist of the metal container for the heat transfer fluid, made of a steel pipe, a mesh structure on the inner side of the tube for the transport of the condensed fluid to the evaporation zone and the heat transfer fluid itself.
2.1. Fundamentals of heat pipes Heat pipes are passive heat exchangers which rely on a evaporation–condensation mechanism: When heat is supplied to the one side the working fluid evaporates and flows to the cold end where it condenses. The condensate is transferred back to the evaporation zone by gravitational or capillary force. The operating pressure of the heat pipe equals therefore the vapour pressure of the working fluid pLV at the respective operating temperature T HP . The heat transfer of a heat pipe can be limited by the properties of the working fluid, heat pipe geometry and capillary structure, these limitations are discussed thoroughly in [6]. For the following considerations it is assumed that the heat transfer is only limited by radiation and convection on the heat pipe outer surface and therefore the active length of the heat pipe.
2.2. Hydrogen permeation through metals
Fig. 1. Scheme of the inactivation of heat pipes due to hydrogen permeation.
The permeation of diatomic gases like hydrogen through a dense body is divided into several steps: The adsorption of the hydrogen at the surface, the dissociation into protons and dissolution into the metal body, which is sometimes referred to as solution of the hydrogen. Then the diffusion through the metal in form of protons, which is in most cases the rate-limiting step of
922
J.M. Leimert et al. / International Journal of Heat and Mass Transfer 92 (2016) 920–928
the whole process and finally the recombination and desorption from the solid [16,21,26]. The dissolution of a gas in a solid body can be described using Sievert’s law [21]:
ci ¼ S pni
ð1Þ
In this equation ci is the concentration of the dissolved component, S the solubility and pi the partial pressure of the component in the gas phase. For a diatomic gas like hydrogen the exponent is n ¼ 0:5 as the gas dissociates after the solution in the metal. The resulting molar flux jH2 can be calculated using Fick’s first law [21] assuming a linear concentration profile:
jH2 ¼ D
dc Dc ¼ D dx Dx
ð2Þ
Eq. (1) can be applied to this equation for the concentration of the dissolved component. The resulting equation only depends on the partial pressure of the hydrogen on both sides of the solid wall p1 resp. p2 , the wall thickness of the membrane Dx and the coefficients of solubility and diffusion S and D:
jH 2 ¼
DS ðp n pn2 Þ Dx 1
ð3Þ
The permeability P, which is the product of the coefficients of solubility and diffusion is used more often in literature. To calculate the molar flow of hydrogen n_ H2 the resulting equation has to be multiplied with the area of the solid A:
n_ H2 ¼
PA n ðp pn2 Þ Dx 1
ð4Þ
As mentioned above, the partial pressure exponent n is assumed to be 0:5. However, in some cases also values ranging from 0.5 to 1 were found experimentally due to limitations in the dissolution. This can be caused by surface contaminations or oxide layers. For the following calculations the exponent was assumed to be 0.5 as this value is commonly used in literature, which allows easy comparability of the results. The calculation of the hydrogen molar flow in the presence of dense oxide layers is discussed in [16]. The reduced permeability can be described by using the reduction factor H.
H¼
n_ Me _nMeOx
ð5Þ
Möllenhoff reports about oxide layers on austenitic steel 1.4876 which reduced the permeation of hydrogen by a factor of up to 520 [16,17]. For the operation of heat pipes in hydrogen atmosphere such a dense oxide layer would influence heat pipe behaviour significantly [15]. The permeability itself is a temperature activated property and can be calculated with the Arrhenius equation using the permeability constant P 0 , permeability activation energy EA and ideal gas constant R:
EA
P ¼ P0 e RT
ð6Þ
Table 1 shows a collection of permeability constants and activation energies for possible heat pipe housing and hydrogen window materials. 2.3. Material considerations The hydrogen tolerant heat pipe comprises of the housing, the working fluid and the hydrogen window. In the following section the main features of these components are discussed in regards of their suitability in a heat pipe working in hydrogen atmosphere. Additionally, a model was developed to describe the time dependence and equilibrium state of the inactivation procedure. 2.3.1. Working fluid As discussed above, for the temperature regime from 600–1000 °C the working fluids of choice are mostly liquid alkali metals like lithium, sodium or potassium. As the vapour pressure of these metals rises with their position in the periodic table, lithium is suitable for higher temperatures than sodium or potassium. For the removal of hydrogen from the heat pipe the internal pressure of the hydrogen buffer, which equals the vapour pressure of the working fluid, is of importance as it enhances the hydrogen permeation out of the heat pipe. 2.3.2. Housing The housing of the heat pipe is normally made of high temperature steel, sometimes also refractory metals like Tantalum, Titanium or Tungsten are discussed because of their higher strength which allows lower wall thicknesses [7,18]. Because of the higher price compared to steel housings these materials are not considered for an application in process engineering. A serious problem for steel housings in the desired temperature regime is sigma phase embrittlement. The buildup of the sigma phase leads to a decrease of the notched impact strength of the affected steels of up to 80% [22]. The formation of the sigma phase depends mainly on the composition of the steel regarding the three main components, which are chromium, iron and nickel. Nickel base alloys with nickel contents of more than 40% are not affected by the sigma phase, while nickel contents below 11% can also lead to sigma phase embrittlement due to nickel rich clusters in the steel [12,25]. The hydrogen permeability is of course also affected by the main components and especially by the nickel content as nickel has a comparatively high permeability for hydrogen: As a result, nickel base alloys show a higher permeability than steels with iron as main component [2]. Fig. 2 supports this finding as the permeability of most literature values are directly proportional to their nickel content. For the application in hydrogen rich atmosphere heat pipe housings should therefore not comprise steels with a high nickel content. This poses an optimisation problem for the
Table 1 Permeation rates of different materials for high temperature heat pipe or hydrogen applications. P 0 [mol=m s Pa0:5 ]
P at 800 °C [mol=m s Pa0:5 ]
Refs.
Material
Temperature range [K]
303 SS
773–1173
3:6010
7
67,000
1:9710
10
[9]
304 SS
823–1173
5:10107
71,000
1:781010
[9]
316 SS
500–930
7
2:3610
63,496
2:021010
[30]
Inconel
500–930
9:30107
69,000
4:071010
[9]
Palladium
370–900
8:11107
15,464
1:43107
[28]
Tantalum
253–944
5:80109
20,203
5:58108
[28]
Nickel
673–1123
4:65107
55,204
9:551010
[9]
Tungsten
1100–2400
7:60107
1,321,923
1:071013
[28]
EA [J=mol]
J.M. Leimert et al. / International Journal of Heat and Mass Transfer 92 (2016) 920–928
923
The easiest approach for the hydrogen window is to apply nickel tubes directly into the gas channel of the heat pipe at the cold end of the condenser as shown in Fig. 4. The hydrogen will be collected at this position of the heat pipe due to the vapour flow of the working fluid. The disadvantage of this concept is that the heat pipe and the hydrogen window will cool down partially resulting in a ‘‘cold finger” and lower heat transfer rates due to the formation of an inactive zone. This leads to a decrease in hydrogen permeability of the hydrogen window in the inactive zone as the permeation is a temperature activated property. The length of the inactive zone can be influenced by the geometry of the hydrogen window. 2.4. Mathematical model for hydrogen inactivation Fig. 2. Permeability of stainless and nickel base steels as a function of the nickel content (temperature: 850 °C; permeability measurements from [9,14,27,30]).
container material as steels with low nickel content tend to sigma phase embrittlement. For the experiments shown in this work the steel Sandvik 253 MA (material number: 1.4835) was used. It combines low formation of sigma phase as it also shows 1% sigma phase after 2000 working hours at 800°C with a low nickel content of only 11%. The exact composition of the steel is shown in Table 2. 2.3.3. The hydrogen window The hydrogen permeating into the heat pipe during operation has to be removed continuously. One possibility is the construction of an open system where the hydrogen can be sucked out of the heat pipe as proposed by North and Anderson in [19]. This solution ensures the operation of the whole heat pipe without the buildup of a cold finger. The disadvantage is the increased effort in control and safety engineering as the potentially hazardous working fluid must not be sucked out of the heat pipe. Additionally alkali metals tend to plug small tubes when cooled down. Furthermore, alkali metals form solid hydrides in the presence of gaseous hydrogen in the temperature regime of about 400 °C, which causes severe plugging problems. Alkali hydride formation in heat pipe operation is also discussed in the experimental section. For these reasons the focus in this paper lies on passive systems which remove the hydrogen by permeation. For an efficient removal of the hydrogen the hydrogen window must consist of a material with high hydrogen permeability. The materials of choice are palladium, tantalum or nickel because, as shown in Table 1, their permeability is at least one order of magnitude higher than that of steels which are used for the housing. There are also some restrictions regarding corrosion and material compatibility: Palladium is not compatible with alkali metals in this temperature regime [18], Tantalum can withstand the alkali metals but is attacked by traces of oxygen which leads to material failure [20]. In contrast to those materials, the attack of sodium on nickel is moderate in the range of 2,5–25 lm/a [4]. Because of its low price and high durability, nickel will be used in the following experiments. Table 2 Composition of the housing material 253MA (1.4835) according to inspection certificate. Element
Mass fraction [%]
Carbon Silicium Manganese Phosphorus Chromium Nickel Cerium
0.075 1.50 0.59 0.022 20.94 10.86 0.04
In the following part a model for the inactive length of a heat pipe exposed to hydrogen is developed. This can be used in further works as a layout tool for the dimensioning of a hydrogen window. Additionally a model of the inactive length of a heat pipe as a function of time will be shown. To describe the deactivation of the heat pipe, a few assumptions have to be made, which are also illustrated in Fig. 1: The hydrogen will only enter the heat pipe in the active zone as the temperature of the housing is significantly higher in this part of the heat pipe. Furthermore the partial pressure of the hydrogen is zero as the hydrogen will be transported to the inactive zone immediately after permeation. The heat pipe is isothermal with one discrete temperature each in the active and inactive zone. The hydrogen in the inactive zone has the same temperature as the heat pipe working fluid, i.e. the working temperature of the heat pipe. This condition holds for heat pipes with a small inactive zone. In a real system the hydrogen temperature will be in between the heat pipe operating temperature and the temperature of the housing in the inactive zone depending on the heat transfer conditions. The hydrogen leaves the heat pipe only via the hydrogen window as the permeability of the window is at least one order of magnitude higher than that of the steel housing. Hydrogen solubility in liquid sodium has been neglected. According to Wittingham [31] hydrogen solubility in sodium at approx. 0.56 bar, is only 110 ppm and therefore not considered in dynamic considerations. Formation of alkali hydrides is discussed in the corresponding section. Assuming molar fluxes of hydrogen entering through the steel housing and leaving through the hydrogen window, n_ H respectively n_ W and an accumulated amount of hydrogen inside the heat pipe nacc , the hydrogen balance of the heat pipe can be developed as follows:
n_ H n_ W ¼
dnacc dt
ð7Þ
The molar flow of hydrogen through the steel housing and the hydrogen window can be calculated using (4):
n_ H ¼ P H
AH 0:5 pH2 p0:5 H2 ;HP D xH
n_ W ¼ PW
AW 0:5 pH2 ;iz p0:5 H2 ;permeate DxW
ð8Þ ð9Þ
These equations can be simplified by assuming that the partial pressure of hydrogen in the active zone of the heat pipe pH2 ;HP is zero as the hydrogen will be transported to the inactive zone immediately after permeation. This simplification is also valid for
924
J.M. Leimert et al. / International Journal of Heat and Mass Transfer 92 (2016) 920–928
the pressure on the outer side of the hydrogen window pH2 ;permeate as it is flushed with inert gas to remove the permeated hydrogen. The hydrogen pressure in the inactive zone pH2 ;iz equals the vapour pressure of the working fluid pLV at the operating temperature of the heat pipe. The area of the housing responsible for the hydrogen permeation AH;az depends on the length of the inactive zone liz , the diameter of the housing DH and the overall length of the heat pipe lHP . It is assumed that hydrogen only permeates in the active zone of the heat pipe as the permeation is much lower in the colder inactive zone and the driving pressure is also lower due to the hydrogen on the inside.
AH;az ¼ pDH ðlHP liz Þ
ð10Þ
The area of the hydrogen window where hydrogen permeates out of the heat pipe AW;iz can be calculated accordingly. It is assumed that the hydrogen only permeates through the hydrogen window:
AW;iz ¼ pDW liz
ð11Þ
The accumulated hydrogen in the heat pipe corresponds to an inactive volume, which can be calculated using the ideal gas equation. The hydrogen temperature and pressure are assumed as the heat pipe working temperature and the working fluid vapour pressure, respectively:
nacc
VH p ¼ 2 LV RT HP
ð12Þ
Assuming this hydrogen volume is always located at the top end of the condenser where it forms the inactive zone, the inactive length of the heat pipe can be calculated using the cylindrical geometry:
liz ¼
V H2 p ðD 4
H
2xH Þ2
ð13Þ
2.4.1. Model for time-dependence of inactivation The inactivation of the heat pipe with respect to the time can be calculated by integrating (7) over time by numerical integration:
Z nacc ðtÞ ¼ 0
t
n_ H ðsÞ n_ W ðsÞdðsÞ
ð14Þ
2.4.2. Model for inactive length The equilibrium state at t ! 1 of the inactive length can also be estimated. In equilibrium state the change in accumulated hydrogen becomes zero, resulting in a simple equation:
n_ H n_ W ¼
dnacc ¼0 dt
ð15Þ
The inactive length of the heat pipe can then be expressed as:
DxW PH DH p0;5 liz H2 ¼ 0;5 lHP DxH PW DW p0;5 LV þ DxW P H DH pH2
ð16Þ
If no hydrogen window is used, the hydrogen window permeability may be set to zero or to the permeability of the steel housing if it is used as hydrogen window. The inactive length was calculated for two cases: A hydrogen window tube 3 0.3 mm using Nickel and degassing only over the heat pipe housing made of stainless steel SS304 (material number: 1.4301) with permeability values from Gorman and Nardella [9]. The length of the hydrogen window was 400 mm. Fig. 3 shows the modelling results with DT being the temperature difference between active and inactive zone. The inactive length is decreasing with the heat pipe operating temperature as the heat pipe inner
Fig. 3. Equilibrium calculation of the inactive length of sodium heat pipes with degassing via housing or with a nickel hydrogen window at 1 bar and 10 bar hydrogen pressure; DT: temperature difference between active and inactive zone; DH = 33.7 mm, xH = 3.2 mm, DW = 3 mm, xW = 0.3 mm, LW = 400 mm.
pressure is rising. The temperature of the cold end has a high impact on the inactive length which stresses the importance of a good isolation on the cold end. The Nickel hydrogen window decreases the inactive length by a factor of about two. However, with a low temperature drop in the inactive zone a high active length is also possible without hydrogen window. This could be achieved by insulation or supplementary heating of the inactive zone. 3. Experimental For the study of the deactivation phenomena a heat pipe test rig was erected at FAU-EVT. It can be used for heat pipes with an outer diameter of up to 50 mm and a length of up to 1500 mm. The heating is supplied by a Sandvik furnace with a heat duty of 5 kW. The heat pipe is fixed by four steel clamps. It is possible to change the operating angle of the heat pipe, e.g. for measurements of maximum heat transfer. The heat pipe is provided with an enclosure of 500 mm to allow different gas atmospheres for the testing procedure, see Fig. 4. The inactivation is measured by six thermocouples which detect the wall temperature of the heat pipe. The heat pipe housing is made of Sandvik 253 MA with three layers of steel mesh made of 1.4841. The mesh has a wire thickness of 0.160 mm and a wire space of 0.200 mm. The upper ending of the heat pipe has a compression fitting so that different materials for degassing can be tested. The heat pipe can be evacuated by a steel pipe leading to a ball valve and a vacuum pump. The heat pipe has an overall length of 1200 mm with an evaporation zone of 500 mm, an adiabatic zone of 200 mm and a condensation zone of 500 mm. At the beginning of the experiments the heat pipe was heated to the desired operation temperature. When stationary state was reached, the double wall of the heat pipe was flushed with hydrogen. The deactivation of the heat pipe was recorded using six thermocouples type K. After the whole condensation zone of the heat pipe was deactivated the reactivation procedure was started by flushing with nitrogen. 4. Results and discussion The results are divided into three parts: First the results from the deactivation measurements are shown with a comparison to
J.M. Leimert et al. / International Journal of Heat and Mass Transfer 92 (2016) 920–928
925
Fig. 6. Measured inactive length plotted against deactivation times at operating temperature of 850 °C with OLS permeability fit of the modelling approach.
Fig. 4. Piping and instrumentation diagram of the heat pipe test rig and sketch of the nickel hydrogen window; DH : Diameter of heat pipe housing; DW : Diameter of hydrogen window.
the modelling described in the previous part. The second part will be our results on a nickel hydrogen window, which was placed inside a sodium heat pipe and tested for more than 1000 h without failure. As a last topic we want to show the impact of sodium hydride formation on the hydrogen-alkali metal heat pipe system.
4.1. Hydrogen inactivation In the first measurement campaign, the hydrogen impact without hydrogen window was studied. For this reason the double wall of the heat pipe was flushed with pure hydrogen at 1 bar which resulted in complete inactivation. The temperature of the thermocouples was recorded, Fig. 5 shows a typical result with normalised temperatures. For the comparison of different experiments the deactivation time was determined as shown in Fig. 5. In a second step the corresponding inactive length was plotted against the deactivation time. The modelling approach was then used to fit a curve over the generated deactivation points using OLS approximation. The result is shown in Fig. 6. It appears that the experimental data can be described using the developed model. The model can be adjusted to the measured points using
Fig. 5. Exemplary temperature data from inactivation measurements at operation temperature of 850 °C with determination of inactivation time.
only the housing permeability. The increasing deactivation time of the test runs is probably tied to the growth of an oxide layer on the steel housing reducing the permeability of the housing by a factor of up to 4.5. According to various sources, the growth rate and density of the layer is based on the gas atmosphere and especially on the oxygen content [11,15,17]. Before the fourth test run the heat pipe was reactivated using 5% H2 in N2 instead of technical N2 , which lead to an increase in deactivation time. It is assumed that the lower O2 partial pressure caused a formation of very dense oxide layers as described in [15] which slowed down inactivation.
4.2. Hydrogen window In a second measurement campaign a 3 0.3 mm nickel tube with a length of 500 mm was mounted inside the heat pipe. It acts as a hydrogen window in the cold finger, which can be modelled using the described approach. The tube is flushed with nitrogen to remove the hydrogen permeating out of the heat pipe. The active length of the heat pipe can be measured directly in this setup with a thermocouple inserted into the hydrogen window. The active length of the heat pipe is measured without hydrogen and after reaching steady state operation in hydrogen atmosphere. The active length is the difference of these two values. Any errors occurring in the temperature measurement due to thermal conduction are avoided as these errors will occur at both measurements. Fig. 7 shows the results from the hydrogen window experiments. The hydrogen pressure on the outer side of the heat pipe
Fig. 7. Inactive length of a sodium heat pipe in hydrogen atmosphere with a 3 0.3 mm nickel hydrogen window; Antoine parameters from [1].
926
J.M. Leimert et al. / International Journal of Heat and Mass Transfer 92 (2016) 920–928
was kept constant at 1 bar. The inner pressure of the heat pipe which equals the vapour pressure of the working fluid was recorded directly during the measurements. For a comparison the saturated liquid line of pure sodium was calculated using Antoine parameters given in [1]. The measured values are in good agreement with the calculated inactivation process. Small deviations can be explained with impurities in the used sodium. The displayed permeability of the housing was taken from the deactivation measurements. The inactive length is rising slightly with the operating temperature as the higher housing permeability overcompensates the higher operating pressure. The heat pipe has a relative inactive length of 25–31% depending on the operating temperature. In order to reduce the inactive length, the geometry of the hydrogen window can be changed using higher diameters or lower wall thicknesses. However, the wall thickness is limited by the corrosion rate of nickel in liquid sodium which lies in the range of 2,5– 25 lm/a and the weldability of the materials [4]. The housing permeability has a high impact on the inactive length, it can be influenced by targeted oxidation of the steel using different gas atmospheres with low oxygen partial pressures. The permeability can be lowered by up to three orders of magnitude resulting in a much lower inactive length [15,16]. The hydrogen window was tested for approx. 1500 h inside the sodium heat pipe. After completion of the tests, the Nickel tube was demounted and cleaned of residual sodium. The tube diameter was measured and compared to a new nickel tube, which showed a decrease in diameter of 10–15 lm which corresponds to a corrosion rate of 9 106 cmmg 2 h. 4.3. Formation of alkali hydrides Sodium hydride is an ionic product of the reaction of molecular hydrogen with (liquid) sodium. It is only stable in its solid phase, up to an upper temperature limit depending on ambient pressure (e.g. 425 °C at ambient pressure for Na).
NaðlÞ þ 1=2 H2 NaHðsÞ
ð17Þ
Formation of other alkali metal hydrides is similar. Enthalpies of hydride formation of relevant alkali metals Li, Na and K are DHf ;LiðsÞ ¼ 77; 71 kJ/mol, DHf ;Na ¼ 56; 4 kJ/mol and DHf ;K ¼ 57; 82 kJ/mol respectively [3]. For lithium also a liquid LiH phase exists due to the more elevated temperatures. Detailed phase diagrams for binary alkali metal – hydrogen systems have been computed based on data from [29] and are displayed in Fig. 8. The temperature of the active heat pipe zone determines the heat pipe internal pressure. In case that the temperature in the inactive zone of the heat pipe drops below the formation temperature of the alkali hydride, the conversion of the liquid alkali metal
Fig. 8. Phase diagrams of binary alkali metal – hydrogen system in heat pipes; calculated with data from [29]. Exemplary determination of NaH formation limit (=approx. 400 °C) in cold finger for HP operated at 800 °C in active zone.
Fig. 9. Basic mechanism of metal hydride induced deactivation of high temperature heat pipes.
starts and hydrogen of the hydrogen buffer is consumed. This initiates an opposing trend to heat pipe deactivation and thus may lead to a quasi equilibrium situation as long as elementary alkali metals are still available. Ongoing consumption of the alkali metal leads to a slow blocking of the wick and to a complete deactivation of the heat pipe. Fig. 9 illustrates this phenomenon. A decomposition of the hydrides i.e. a complete reactivation of the alkali metal heat pipe is possible by heating the heat pipe over the formation limit. However, heating the hydrates above the decomposition temperature suddenly releases large amounts of hydrogen. The released hydrogen instantly floods the heat pipe and deactivates it nearly completely. This can result in material failures due to overheating of the metal container as the evaporating of the working fluid might be suppressed by the flooding with hydrogen. For continuous operation of heat pipes in hydrogen atmospheres this hydride formation limit has to be avoided. System design has to avoid cooling below the limit, even if some degree of hydrogen deactivation of the heat pipe is permitted. In experimental studies hydride formation leads to misinterpretation of the inactivation time and length. It is important to assure that the critical temperature is avoided by using supplemental heating of heat pipes in the inactive zone. The sodium hydride formation was investigated in a measurement campaign using planar heat pipes shaped as a flat rectangular plate. The detailed design and setup is described in [5], the heat pipe dimensions were 270 120 4 mm (L W H), the capillary structure was composed of two layers of mesh 80 on the upper as well as lower side of the flat heat pipe. One layer of stainless steel mesh 8 provides the open space for the vapour backflow within the heatpipe. Fig. 10 shows the temperature profile of the cold finger of a planar heat pipe in hydrogen atmosphere at 1 bar. The deactivation progresses very fast until the coldest spot of the heat pipe reaches the NaH formation temperature of 400 °C. After that the
Fig. 10. Temperature profiles during initial heat pipe deactivation and starting of NaH formation.
J.M. Leimert et al. / International Journal of Heat and Mass Transfer 92 (2016) 920–928
927
Acknowledgements This research is carried out in the framework of the Bavarian Hydrogen Center (BHC) joint research program. The authors would like to acknowledge the support provided by the Bavarian State Ministry of Science, Research and the Arts.
References
Fig. 11. Dynamics of hydrogen deactivation with NaH formation, solid black line shows cold end temperature of heat pipe, data points indicated inactive length of heat pipe in relation to entire heat pipe length.
deactivation nearly remains constant for up to 56 h. In this time the hydrogen is consumed by the working fluid. Fig. 11 also illustrates this result: The relative inactive length rises to up to 0.4 in the first three hours of deactivation, after that NaH is formed and deactivation decelerates. The instationary behaviour of the cold end temperature seen from 4 to 10 h can also be traced back to NaH formation: As the formation temperature is reached a large amount of hydrogen is consumed by NaH which leads to reactivation of a part of the heat pipe. This raises the cold end temperature over the decomposition temperature and results in a release of the hydrogen. The experiments described before were conducted in cylindrical heat pipes avoiding NaH formation by heating the cold end constantly to 500 °C. 5. Conclusion Hydrogen inactivation is a serious problem whenever heat pipes are used in atmospheres containing hydrogen. In this paper the underlying mechanisms of the inactivation were explained and measures to limit or avoid inactivation were discussed. The deactivation time of a heat pipe made of stainless steel which is typically used in the Heatpipe Reformer, where hydrogen inactivation is a main design challenge according to [15], is in the range of 100–200 min. This could result in a mandatory cyclic operation of the heat pipe and thus the entire system, where the operation of the reactor or heat exchanger is stopped and the apparatus is flushed with steam to remove hydrogen from the heat pipe. A hydrogen window made of nickel was tested for over 1500 h without failure and resulted in a relative inactive length of 25–31% rising with the operating temperature. The inactive length could be reduced by rising the temperature of the inactive zone or changing the geometry of the hydrogen window. Further significant influence parameters to deactivation are the formation of alkali hydrides and oxide layer growth on the steel housing depending on external gas atmospheres. Ongoing research at the Institute of Energy Process Engineering focuses on advanced degassing setups that allow continuous operation of liquid metal heat pipes in hydrogen atmosphere with high active lengths.
Conflict of interest None declared.
[1] P. Browning, P.E. Potter, An assessment of the experimentally determined vapour pressures of the liquid alkali metals, in: Handbook of Thermodynamic and Transport Properties of Alkali Metals, 1985, 349–358. [2] P. Calderoni, M. Ebner, R. Pawelko, High-temperature hydrogen permeation in nickel alloys, in: Proceedings of HTR, 2010. [3] M.W. Chase, NIST-JANAF Thermochemical Tables, 1998, Part I Al-Co. [4] C.F. Cheng, W.E. Ruther, Corrosion of nickel in liquid sodium, Corrosion 28 (1) (1972) 20–22.
928
J.M. Leimert et al. / International Journal of Heat and Mass Transfer 92 (2016) 920–928
[27] A.S. Schmidt, F. Verfuss, E. Wicke, Studies on the permeation of hydrogen and tritium through heat resistant alloys, J. Nucl. Mater. 131 (1985) 247–260.
[29] D.R. Stull, H. Prophet, JANAF thermochemical tables, 1971.