Adsorptive storage system model development and its validation

Adsorptive storage system model development and its validation

Résumé

  Cette thèse présente l’évaluation multiphysique du comportement thermique et de la capacité de stockage d’un réservoir d’hydrogène fonctionnant par cryo-sorption de grand volume rempli d’un matériel adsorbant de type métallo-organique, le MOF-5, pour les applications stationnaires qui requièrent un grand volume d’hydrogène et pour des systèmes de distribution d’hydrogène.Dans un premier temps, un modèle général du système de stockage d’adsorption basé sur la mécanique des fluides numérique (MFN) est développé et implémenté dans la plateforme MFN du logiciel multiphysique COMSOL. Afin de valider le modèle développé, des expériences de stockage de l’hydrogène sont menées: de l’hydrogène gazeux à 77 Kest chargé dans prototype de stockage à petite échelle (2,5 L) rempli de MOF-5 jusqu’à ce que la pression atteigne 9 MPa. La température et la pression mesurées expérimentalement sont comparées avec la température et la pression simulées, et nous observons un bon accord entre les résultats de simulation et les résultats expérimentaux. La vérification de la cohérence du modèle de stockage d’adsorption est également réalisée en comparant les températures obtenues lors de la simulation avec ceux obtenus à partir des mesures de banc d’essai à l’aide du charbon activé AX-2pM comme adsorbant. Le modèle est ensuite adapté à l’étude des effets du refroidissement par circulation continue sur les performances thermiques et la capacité de stockage du réservoir prototype à petite échelle rempli de MOF-5.

  Les résultats de simulation montrent une fois de plus un bon accord avec les résultats expérimentaux. Dans le processus de refroidissement par circulation continue, une partie de l’hydrogène rempli est adsorbée et le reste passe au travers le milieu poreux et est récupéré. Les résultats expérimentaux et les simulations montrent que la chaleur générée pendant le temps de charge est transportée hors du réservoir pendant le refroidissement par circulation continue, ce qui entraîne une diminution de la température du système de stockage. Le modèle est ensuite utilisé pour étudier l’effet de la conversion endothermique para-ortho de l’hydrogène sur le comportement thermique et la capacité de stockage du réservoir prototype à petite échelle. Les résultats montrent que la conversion endothermique permet de réduire la température du système et augmente ainsi la capacité de stockage nette. Le modèle CFD qui est utilisé pour simuler le refroidissement par circulation continue et la conversion para-ortho est étendu afin d’étudier la performance multiphysique d’un réservoir de stockage d’hydrogène de grande capacité (20 m3) rempli de MOF-5. La pression maximale du réservoir est réglée à 4 MPa. Dans un premier temps, la simulation est effectuée pour réduire la température initiale du réservoir de 300 K à 80 K. La réduction de la température initiale est obtenue par des cydes de charge-décharge. Une fois que la température du réservoir atteint 80 K, le réservoir est rempli d’hydrogène à 77 Kjusqu’à ce que la pression maximale de 4 MPa soit atteinte. Après, on utilise le refroidissement par circulation continue pour maintenir la température autour de 80 K. Après -5.5 heures de refroidissement en circulation continue, un seuil de viabilité de l’approche est atteint lorsque la masse totale d’hydrogène stocké dans le réservoir (-275 kg) devient égale à celle stockée dans un réservoir cryo-comprimé similaire avec de l’hydrogène maintenu à 77 K et 4 MPa. Après 5.5 heures, le réservoir rempli de MOF-5 utilisant le refroidissement par circulation continue contient nettement plus d’hydrogène qu’un réservoir cryo-comprimé.

Basics ofadsorption

  Adsorption is the preferential adhesion of a chemical species from agas, liquid or dissolved solid to a solid surface. This results in increased concentration of chemical species close to the surface of the solid relative to the bulk of the solid [1]. The solid surface that adsorbing the chemical species is called an adsorbent, while the adsorbed chemical species are referred to as the adsorbate. On the basis of the forces of attraction between adsorbent and adsorbate, adsorption is classified as physisorption or chemisorption. Physisorption occurs mainly due to the weak Van der Waals force of attraction between the adsorbent and the adsorbate, while chemisorption occurs when the adsorbate is held on an adsorbent surface by stronger chemical forces that are specifie for each adsorbent and adsorbate

Adsorption models

  The adsorption isotherm is a curve which shows the variation of adsorption with pressure at a given constant temperature. The adsorption model is a mathematical representation of excess or absolute adsorption isotherms and it is required for predicting the thermal and storage performance of the adsorptive hydrogen storage system. A series of adsorption models such as Langmuir, Freundlich, Brunauer-Emmett-Teller (BET), Toth, MPTA, Ono-Kondo, Dubinin-Radushkevich (D-R), Dubinin-Astakhov (D-A), modified Dubinin-Astakhov, etc. have been suggested and studied [4-6]. Among these, the modified Dubinin-Astakhov (D-A) model is one of the most widely used adsorption models for adsorptive hydrogen storage system model [7-9].

Test bench

   The test bench consists of a charging manifold, a discharging manifold and a flowthrough manifold aIl tethered to a subscale-prototype (2.5 L) stainless steel tank filled with MOF-5 adsorbent. For charging experiments, pressure regulator is adjusted and then hydrogen gas (99.999 %, Praxair) is admitted into the tank by means of a thermal mass flow controller (TMFC, Brooks 0-100 SLPM) and a back pressure regulator (BPR, TESCOM) installed on the charging manifold. A bypass manifold and a manual valve connected across the TMFC help to maintain the required 50 psia pressure difference between the inlet and outlet ofthe TMFC at the start of flow control. Once admitted into the tank, the pressure of the hydrogen gas is monitored using a pressure transducer, PT (Dresser DXD, 0-3000 psia). For cryogenjc charging tests, hydrogen gas is pre-cooled to -77 Kbefore it enters the storage tank by means of a helical cooling loop immersed inside a liquid nitrogen (LN2)-filled 40 L cryogenic vessel, while the storage tank is vertically immersed in a 125 L cryogenic vessel (CryoFAB) filled with LN2. Nine k-type thermocouples are attached along the axial (Tl-T6 along axial from top to bottom) and radial (T7-T9 along radial from axis to tank wall) directions within the tank to monitor the spatial distribution of tank temperature, while a single k-type thermocouple inserted at the entrance ofstorage tank monitors the inlet hydrogen temperature. To isolate the hydrogen manifolds from the storage tank, two aIlmetal valves (Swagelok) are attached at the entrance and exit of the tank. To measure the flow of the gas during discharging and flowthrough, two thermal mass flow meters are used (TMFR, Brooks 0-30 SLPM, Bronkhorst TMFC (0-150 SLPM). A rotary vane pump is attached to the charging manifold to evacuate the whole system prior to aIl experiments. The data acquisition and control of valves, mass flow controllers and temperature sensors are made possible using National Instruments Compact Field Point communication system which is interfaced to a PC using NI LabVIEW professional development software suite. Pressure data are acquired by interfacing the transducer to PC using RS-232 SeriaI communication protocol

Sample loading and void volume measurement

  The MOF-5 sample required in our tests is obtained from BASF. Since this adsorbent is sensitive to moisture and air, it was stored and handled within a dry argon-filled glove box workstation. After embedding the thermocouples in the storage tank, the tank is transferred into the glove box and 358 g ofMOF-5 is added into it. After filling, all-metal valves on both sides of the tank are closed until it is re-inserted back into the test bench setup. Once attached, the argon gas is outgassed by means.of the mechanical pump. The empty tank volume, the manifold volume (the volume between MFC and the all-metal valve at the tank inlet) and the void volume of MOF-5 tank are estimated by admitting known amounts of ultrapure helium (99.999 %, Praxair) into the system. Void volume measurements with MOF-5 are carried out at room temperature and at a final equilibrium pressures ofless than 2 MPa, so as to minimize any potential He adsorption.

Parameters used in the simulation

  The modified Benedict-Webb-Rubin real gas equation of state, as implemented in REFPROP was used to ca1culate the thermodynamic properties such as specifie heat capacity, density, viscosity, enthalpy and thermal conductivity of hydrogen gas. Initially, the NIST database was interfaced with COMSOL Multiphysies®using Matlab (R2010b) [14, 17]. While this is a highly accurate method of importing thermodynamie data, it compromises the computational speed. Therefore, we obtained data from NIST, which is converted into a grid form in Matlab and it is implemented in COMSOL Multiphysics®platform [18]. The properties of normal hydrogen are used in the simulation when the para-ortho conversion is not taken into account. The correlations of temperature dependent thermal conductivity and the specifie heat capacity of stainless steel tanks are obtained from elsewhere [19]:
log k (W m-1K- 1 ) = -1.4087 + 1.3982 logT + 0.2543 (logT)2 –
0.6260 (logT)3 + 0.2334 (logT)4 + 0.4256 (logT)5 – 0.4658 (logT)6 +
0.1650(logT)1 – 0.0199 (logT)8 (2.24)
22log CpO kg- 1K-1 ) = 22.0061-127.5528 [log (T (K)] + 303.647 [log (T (K)]2 –
381.0098 [log (T (K)P + 274.0328 [log (T (K)]4 – 112.9212 [log (T (K)]S +
24.7593 [log (T (K)]6 – 2.239153[log (T (K)F (2.25)
Density of stainless steel used is 7830 (kg m-3) [18]. The fitted modified D-A parameters are taken from the literature and are given in table (Table 2.2) [20]. Figure 2.3 shows fitted modified D-A parameters using experimental data [5, 20].

Initial and boundary conditions

  The initial and boundary conditions used for model validation are based on our experimental test bench. Due to the temperature gradient, the initial temperature ofthe tank in the MOF-5 test bench is not uniform, but is rather in the range of 77- 79 K. To implement this temperature range in the model as an initial condition, we divided the whole bed domain into different subdomains and applied different initial temperature (corresponds to experimental initial condition) to each subdomain. The initial pressure and mass flow rate are set to 0.822 MPa and 15 SLPM. In the model, the mass flow rate is implemented in the form of mass flux of 0.0335 (kg m-2 S-l). A heat flux boundary condition with heat transfer coefficient, 200 (W m-3) is applied to the outer wall of the tank.

Result and discussion

  The validation of the CFD model, whieh compares the temperature and pressure of the tank during hydrogen charging obtained from simulation with those measured experimentally is presented here. Figure 2.7 shows the temperature measured along the central axial points in the tank and those obtained from the simulation. While the overall trend of the temperature evolution is clearly reproduced, the temperatures at the points Tz and T3 are slightly overestimated by -2 K (Figure 2.7). We attribute the observed small difference in temperatures to differences in the positions of monitoring points in the simulation and the experiments. In the test bench storage experiment, thermocouples are implanted prior to filling MOF-S. The effects caused by filling of MOF-S can shift the absolute positions of the thermocouples randomly by ± 0.3 cm. In addition to this, a temperature difference can be caused by the non-uniform bulk density of MOF-S
[7]. The variation of the powder bulk density within the vessel would affect the local heat generation rates, volumetrie heat capacity and bed thermal conductivity
[7]. The use of a temperature-invariant thermal conductivity in Eq. (2.20) and lack  LDF (Linear Driving Force) model in the mass balance equation may also contributes to the observed difference between model and experimental results.

Table des matières

RÉSUMÉ
ABSTRACT.
FOREWORD
ACKNOWLEDGEMENT
LIST OF FIGURES
LIST OF TABLES
ACRONYMS
NOMENCLATURES
SECTION A
CHAPTER 1 INTRODUCTION
1.1 CONTEXT
1.2 OBJECTIVE OF THE THESIS
1.3 STRUCTURE OF THE THESIS
REFERENCES
CHAPTER 2 ADSORPTIVE HYDROGEN STORAGE SYSTEM MODEL DEVELOPMENT AND ITS VALIDATION
2.1 Basics ofadsorption
2.2 Development ofadsorptive system modelfor predicting the multiphysics performance ofa hydrogen storage system
2.3 Validation ofmodel with experimental data obtained[rom the test bench using MOF-S as an adsorbent material
2.4 Validation ofthe model with experimental data obtainedfrom the test bench using Maxsorb TM activated carbon as an adsorbent material
2.5 Conclusion
References
CHAPTER 3 EFFECTS OF FLOWTHROUGH COOLING AND PARA-ORTHO CONVERSION ON THE PERFORMANCE OF A SUBSCALE-PROTOTYPE HYDROGEN STORAGE TANK FILLED WlTH MOF-5 
3.1 Background
3.2 Effect offlowthrough cooling on the performance ofa subscale-prototype tankfilled with MOF-5
3.3 Effect ofpara-ortho conversion heat on the performance ofa subscale-prototype storage system filled with MOF-5
3.4 Conclusion
References
CHAPTER 4 MULTIPHYSICS PERFORMANCE OF A BULK CRYO-ADSORPTIVE HYDROGEN STORAGE RESERVOIR FILLED WlTHMOF-5 
4.1 Thermal and storage performance ofthe bulk tank.
4.2 Effect offlowthrough cooling together with para-ortho conversion on the storage
tank performance
4.3 Effect ofthe massflow rate on the cooling time reduction during the flowthrough cooling stage
4.4 Performance ofthe bulk tank during the dormancy stage
4.5 Conclusion
Reference
CHAPTER 5 SUMMARY OF THE THESIS AND FUTURE WORK
5.1 Adsorptive storage system model development and its validation 
5.2 Effects offlowthrough cooling and para-ortho conversion on the performance ofa subscale-prototype hydrogen storage tank filled with MOF-5
5.3 Multiphysics performance ofa cryo-adsorptive bulk hydrogen storage reservoir filled with MOF-5.
5.4 Future work
References
SECTION B
ARTICLES
Article -1 
Article – 2
Article – 3
SECTION C
ANNEXURE

Télécharger le rapport complet