Institute for Chemical Process and Environmental Technology TOWARDS A VIRTUAL REALITY PROTOTYPE FOR FUEL CELLS Steven Beale Ron Jerome Steven Beale Ron.

Institute for Chemical Process and Environmental Technology

TOWARDS A VIRTUAL REALITY PROTOTYPE

FOR FUEL CELLS Steven Beale Ron Jerome Steven Beale Ron Jerome National Research CouncilNational Research Council

Anne Ginolin Anne Ginolin Institut Catholique d’Arts et Métiers

Martin Perry, Dave Ghosh Martin Perry, Dave Ghosh Global Thermoelectric Inc.Global Thermoelectric Inc.


Introduction Fuel Cells convert chemical energy (hydrogen and Fuel Cells convert chemical energy (hydrogen and

oxygen) to electrical energyoxygen) to electrical energy

Potential replacement for IC enginesPotential replacement for IC engines

National Fuel Cells Initiative: Kyoto summitNational Fuel Cells Initiative: Kyoto summit

Three main parts: Anode, cathode, electrolyteThree main parts: Anode, cathode, electrolyte

Solid Oxide Fuel Cells (SOFC’s) can use methane/natural Solid Oxide Fuel Cells (SOFC’s) can use methane/natural gas in place of hydrogengas in place of hydrogen

Built in stacks of 10-50 cells: Connected in parallel Built in stacks of 10-50 cells: Connected in parallel hydraulically; electrically in serieshydraulically; electrically in series


Introduction SOFC’s operate at up to 1000 ºC SOFC’s operate at up to 1000 ºC

If supply of fuel and air non-uniform, reaction rates and If supply of fuel and air non-uniform, reaction rates and hence temperatures will varyhence temperatures will vary

Temperature uniformity important: If too cold, cell reaction Temperature uniformity important: If too cold, cell reaction shuts down; Too hot, mechanical failureshuts down; Too hot, mechanical failure

Current work to model fluid mechanics (mechanical Current work to model fluid mechanics (mechanical design). Goal: Uniform delivery of air and fuel to the design). Goal: Uniform delivery of air and fuel to the membrane-electrode assemblymembrane-electrode assembly

Chemistry not considered at present timeChemistry not considered at present time


Introduction Single cell modelSingle cell model

Stack model: 10-50 cells. Two approachesStack model: 10-50 cells. Two approaches–Direct Numerical Simulation (DNS). Direct Numerical Simulation (DNS). –Distributed Resistance Analogy (DRA)Distributed Resistance Analogy (DRA)

For the DNS require large amounts of storage and memoryFor the DNS require large amounts of storage and memory

Certain details lost with the DRACertain details lost with the DRA

Several different DRA implementations are possibleSeveral different DRA implementations are possible


DRA approach

F F is a ‘distributed resistance’ obtained from theory, is a ‘distributed resistance’ obtained from theory, experiments, or detailed numerical simulations. experiments, or detailed numerical simulations. rri i areare

volume fractions of air and fuel. volume fractions of air and fuel.

iiiiiiiii uFrpruur

tur

2;

0

iii urtr

UFp

urU

velocitylSuperficiaU

velocityalInterstitiu


Determination of resistance term Many internal flows are correlated in the form;Many internal flows are correlated in the form;

where the Reynolds number is written in terms of a where the Reynolds number is written in terms of a “hydraulic diameter”“hydraulic diameter”

The distributed resistance is just,The distributed resistance is just,

Reaf

uDhRe

2

2

hDra

F


Example: Plane duct

For a plane duct:For a plane duct:

For more complex geometries must use numerical For more complex geometries must use numerical integration empirical correlationsintegration empirical correlations

Huuf w

12

221

PHr

3

12HP

F

32

1212

nBH

QL

H

uLp


PHOENICS settings

Used PHOENICS VR to construct SOFC stack modelUsed PHOENICS VR to construct SOFC stack model

Diffusion terms turned off using Group 12 patches Diffusion terms turned off using Group 12 patches GP12DFE etc. for DRAGP12DFE etc. for DRA

Source term with PATCH type PHASEM in momentum Source term with PATCH type PHASEM in momentum equations, and Coefficient equations, and Coefficient CC = = FF//rr

2-D flow imposed in core (2-D flow imposed in core (w w = 0)= 0)


Results

Source: http://www.globeinvestor.com/

Phase 1 CFD modelling begins

Phase 1 endsPhase 2 CFD modelling begins


Results Results of flow calculations for 24 designs were displayed Results of flow calculations for 24 designs were displayed

as 3-D VRML files using a secure web site to the client in as 3-D VRML files using a secure web site to the client in Calgary across internet.Calgary across internet.

This allowed us to work together “at a distance”This allowed us to work together “at a distance”

Images were also displayed locally to client in Ottawa Images were also displayed locally to client in Ottawa using NRC Virtual Reality (VR) wallusing NRC Virtual Reality (VR) wall

SOFC stack completely redesigned as a result of this workSOFC stack completely redesigned as a result of this work


Results

FLOW IN

Exit manifoldInlet manifold

FLOW OUT

Fuel cell stack


NRC Virtual Reality Wall


Pressure in SOFC manifolds and stack


Velocity vectors in manifolds and stack

NB: Vector scale different in stack from in manifolds


Discussion Geometry is quite simple, but flow in inlet manifold

complex; Pressure maximum at front of step. Due to horizontal inlet

Flow within the core of this SOFC stack is uniform, i.e., design is good. Little variation in vectors, in spite of inlet design

Core flow is a low Reynolds number (creeping) flow, driven primarily by the pressure gradient

Pressure drops consistent with values based on theory

Flow in outlet manifold is less complex than inlet: Size and form less critical.


Discussion Gradient across stack is uniform horizontal.

In manifolds gradient is relatively small and decreases with height - due to injection/suction

Manifold losses are in many cases quite significant, with substantial variations observed, depend on particular configuration under consideration.

For uniform flow the ratio of Pstack/Pmanifolds should be large.

Parametric studies identified which parameters important - allowed for the stack design to be optimised.


Comparison of DRA and DNS


Comparison of DRA and DNS approaches DRA and DNS results are similar with minor systematic

deviations.

Details of velocity profile lost with the DRA.

If core resistance is small, inertial effects become significant: Pressure and velocity less uniform.

For tall stacks need large pressure gradient within core, so inertial effects due injection/suction of working fluid from manifolds does not lead to starvation at top of core.

Various DRA approaches are possible. Minor differences occur due to convection terms.


Conclusions DRA model may be used as an engineering tool to design

SOFC's with a measure of confidence: Certain details of flow field are lost. However combines computational speed with accuracy

Certain SOFC models superior.

Back pressure across stack should be large to maintain uniformity of pressure and velocity across core.

Geometric features by which this may be achieved were identified using parametric studies

SOFC design re-configured as a result of CFD.


Future (current) work SOFC’s with more complex passages.

Flow of two working fluids, combined with inter-fluid heat transfer and Ohmic heating.

Initial anlysis suggests the conventional DRA for heat transfer may need to be modified due to low Reynolds number effects.

Concurrent display and manipulation of graphics data, locally on VR walls, and across the country via CA*net 3.

Experimental facilities to gather empirical data and conduct flow visualisation studies for model validation.

Institute for Chemical Process and Environmental Technology TOWARDS A VIRTUAL REALITY PROTOTYPE FOR FUEL CELLS Steven Beale Ron Jerome Steven Beale Ron.

Documents

chemical process

fuel cell stack slide

work slide

chemical energy hydrogen

electrical energy fuel

supply of fuel

possible slide

approaches stack model