1 Hypothetical Accident Scenario Modeling for Condensed Hydrogen Storage Materials Charles W. James Jr, Matthew R. Kesterson, David A. Tamburello, Jose A. Cortes-Concepcion, and Donald L. Anton Savannah River National Laboratory September 14, 2011

2 The objective of this study are to understand the safety issues regarding solid state hydrogen storage systems through:  Development & implementation of internationally recognized standard testing techniques to quantitatively evaluate both materials and systems.  Determine the fundamental thermodynamics & chemical kinetics of environmental reactivity of hydrides.  Build a predictive capability to determine probable outcomes of hypothetical accident events.  Develop amelioration methods and systems to mitigate the risks of using these systems to acceptable levels. Objectives

3 Modeling and Risk Mitigation Punctured / Ruptured Tank Storage Vessel Penetration Possible Water Film Ambient Atmosphere at Temperature Contains O 2, N 2, CO 2 & H 2 O (l), H 2 O (g) Heat Generated by Chemical Reaction Volume Media Temperature Depends on T a, T i, dH/dt, k eff, c peff, … Surface Liquid Water y x t H2H2 Spilled Media Accident Scenario (from UTRC risk assessment): Storage system ruptured and media expelled to environment in either dry, humid or rain conditions. Risk: Under what conditions will there be an ignition event? What are the precursors to the ignition event? Temperature Humidity Water presence Media geometry

4 United Nations Groundwork - Ammonia Borane UN TestResult PyrophoricityPass Self-HeatFail Burn RateFail Water DropPass Surface Contact Fail Water Immersion Pass

5 NH 3 BH 3 TGA Experimental Results  TGA experiments were conducted in an Argon atmosphere.  First and second dehydrogenation reactions occurred

6 COMSOL model:  2-D, axisymmetric  Conduction, Convection, & Radiation Heat Transfer  Weakly Compressible Navier-Stokes Equations  Maxwell-Stefan Species Convection and Diffusion Reaction Kinetics:  Reaction 1-2:  E a =128 [kJ/mol]  A 0 =3.836x [1/s]  c= [1/K]  mol%=14% borazine*  Reaction 3-4:  E a =76 [kJ/mol]  A 0 =10 6 [1/s]  c=0  mol%=41% borazine * NH 3 BH 3 TGA Numerical Simulation 1 mm Sample Argon Gas Phase 1 mm 5 mm

7 NH 3 BH 3 TGA Comparison  Theoretical curve only takes into account H 2 reaction (no other products)  Additional 14 mol-% and 41 mol-% material loss during reaction (for simplicity, all losses assumed borazine)

8 NH 3 BH 3 Calorimetry Simulation Sample Air Phase Sample (5-20 mg) Not to scale Setaram C-80 Calorimeter options : -Dry Air/Argon -Air/Argon with water vapor -Temperature  Wall temperatures were ramped at 0.5 ºC/min  Atmosphere: Dry Air

9 NH 3 BH 3 Calorimetry in Dry Air  Furnace ramped to 150ºC  Additional exothermic heat flow during the temperature ramping  Endothermic dip due to foaming and melting of the material for T > 110 o C

10 Accident Scenarios  50 grams of NH 3 BH 3 was assumed to collect on the ground following a Gaussian distribution.  Mesh consisted of over 9,000 triangular elements  Scenario 1  A heat source (ex. Car muffler) sits 4 inches above the NH 3 BH 3.  Multiple iterations of Scenario 1 were simulated modifying the heat source temperature from 225ºC to 300ºC  Scenario 2  The NH 3 BH 3 falls onto a heated surface  Multiple iterations of Scenario 2 were simulated modifying the heat source temperature from 100ºC to 125ºC Bottom Surface 1.5cm 20 cm t Top Surface

11 Results – Overhead Heating  Reactions 1 and 2 went to completion  Reactions 3 and 4 started, but the reaction rate was slow.  Highest overhead temperature was 300ºC.  Simulations were initiated at higher temperatures, but the timestep needed by the solver was too small for the simulation to conclude in a reasonable timeframe.

12 Results – Overhead Heating Continued  Above 250ºC, the first reaction goes to completion under 1 hour.  At 300ºC, the first reaction is completed within 11 minutes  Below 250ºC, the second dehydrogenation does not start within the simulation time.  At 300ºC, the second dehydrogenation reaction is progressing (slowly).

13 Results – Ground Heating  Ground temperatures above 125ºC were not modeled due to the high rate of hydrogen release and the resulting decrease in simulation timestep.  Initial release of hydrogen occurs at the outer rim of the NH 3 BH 3 mound.  The maximum mound temperature progresses inward toward the center axis, at which point high pressure spikes due to hydrogen release were observed.

14 Results – Ground Heating  At 125ºC, the first dehydrogenation reaction proceeds quickly.  First reaction goes to completion within 2 minutes.  Second dehydrogenation reaction starts, but proceeds very slowly due to the ground temperature being held at 125ºC

15 Conclusions  COMSOL Multiphysics models successfully modeled dehydrogenation of Ammonia Borane as seen in the TGA and Calorimetry experimental comparisons.  Additional models were developed to simulate the release of hydrogen in postulated accident scenarios.  Temperatures above 125ºC (below heat) and 300ºC (above heat) yielded extremely fast hydrogen release rates.  High pressure spikes were observed during the hydrogen release which could be a precursor to the foaming seen experimentally.

16 Special Thanks to the following people:  SRNL  Bruce Hardy  Stephen Garrison  Josh Gray  Kyle Brinkman  Joe Wheeler  Department of Energy  Ned Stetson, Program Manager Acknowledgements THIS WORK WAS FUNDED UNDER THE U.S. DEPARTMENT OF ENERGY (DOE) HYDROGEN STORAGE PROGRAM MANAGED BY DR. NED STETSON