Senthil Kumar V – Research Overview B Tech Chem Engg AC Tech - Anna U 1992-1996 M Tech Chem Engg IIT - Madras 1996-1998 Chemical Reaction Network Theory Graph theoretic method to detect multiple steady states and oscillations Process calculations National Org Chem Ind Ltd 1998-1999 Engineer – Technical Services Trouble shooting of HDPE plant – ZN catalyst based slurry reactors Operation Research Cell – Optimizing product portfolio Process development BARC 1999-2000 Scientific Officer - C Separation of Uranium from phosphatic ores through multi-stage liquid extraction Spreadsheet tool development 1 PhD Chem Engg IISc 2000-2005 Statistical geometric analysis of hard-disk and hard-sphere microstructures Molecular simulations for rapid granular flows Voronoi characterization of micro-structures Energy Storage & Thermal Modeling General Motors R&D 2005-2012, Senior Researcher Fuel cell vehicle - Hydrogen storage - HSE CoE of US DoE 2008 - 2011 Lithium battery - Nonlinear equivalent circuit model Lithium battery – Physics based Reduced Order Model Photo-thermal Nano-composites Desalination & Battery Modeling Samsung Adv Inst Tech - India 2012-Current, Principal Engineer Battery Management System – Physics based Reduced Order Models Nano-filtration membranes for sea water desalination Patents: 5 & Publications: 12 (+3) Conference Proceedings: 9 This presentation does not contain any confidential or proprietary information
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
Senthil Kumar V – Research Overview B Tech
Chem Engg AC Tech - Anna U
1992-1996
M Tech Chem Engg IIT - Madras 1996-1998
Chemical Reaction Network Theory
Graph theoretic method to
detect multiple steady states and oscillations
Process calculations National Org Chem Ind Ltd
1998-1999 Engineer – Technical Services
Trouble shooting of HDPE plant – ZN catalyst based
slurry reactors
Operation Research Cell – Optimizing product portfolio
Process development BARC
1999-2000 Scientific Officer - C
Separation of Uranium from phosphatic ores through
multi-stage liquid extraction
Spreadsheet tool development
1
PhD Chem Engg
IISc 2000-2005
Statistical geometric analysis of hard-disk and hard-sphere
microstructures
Molecular simulations for rapid granular flows
Voronoi characterization of
micro-structures
Energy Storage & Thermal Modeling General Motors R&D
2005-2012, Senior Researcher
Fuel cell vehicle - Hydrogen storage - HSE CoE of US DoE 2008 - 2011
Lithium battery
- Nonlinear equivalent circuit model
Lithium battery – Physics based Reduced Order Model
Photo-thermal Nano-composites
Desalination & Battery Modeling Samsung Adv Inst Tech - India
2012-Current, Principal Engineer
Battery Management System – Physics based
Reduced Order Models
Nano-filtration membranes for sea water desalination
• As packing fraction increases Voronoi tessellation becomes regular. • Voronoi tessellation is a geometric framework which can describe disorder to order seamlessly. • It offers exact definitions of local volume and geometric neighbors. • Can we derive statistical measures using Voronoi tessellation which can describe structure and property of materials at different states of aggregation or disorder?
4
Cell volume distribution & Configurational entropy
J Chemical Physics 2005; 123: 114501
5
Computation of hard disk / hard sphere excess entropy
6
Voronoi Neighbor Statistics
• A topological instability has a thermodynamic connection!
J Chemical Physics 2005; 123: 074502
• Vertex A – formed by intersection of 3 planes – stable • Vertex B – formed by intersection of 4 planes – unstable • Small perturbations – Vertex B breaks down to form a tiny
quadrilateral face – Topological instability. • Perturbed FCC lattice has 14 neighbors instead of 12 (for
rhombic dodecahedron)
7
Sheared Voronoi Neighbor Statistics
• Higher inelasticity (lower coefficient of restitution) disorder to order transition occurs at a higher volume fraction.
• Ordered layers sliding past each other offer lesser resistance to flow than a disordered structure.
• Velocity dependent coefficient of restitution and shear amorphization lead to a non-hydrodynamic pathway to shear thickening.
Physical Review E 2006; 73: 051305
8
Bond-orientational analysis of sheared structures
J Chemical Physics 2006; 124: 204508
• Sheared granular flows showed high degree of crystallinity, but has only about 40% FCC structures.
• Further investigation showed that sheared structures are a mixture of FCC, HCP and BCT structures.
• Flow analogues of martensitic transformations, occurring in rapidly quenched metals.
Li Battery: Detailed Model & Its Reduction
e- e-
Separator Lithium Metal
Oxide Graphite
- +
LiC6 x Li++Li1-xC6+x e- x Li++ LiMO2 +x e- Li1+xMO2
Li+
Discharge of a Lithium Metal Oxide cell
“Current”
Loss of electrons Oxidation Anode Gain of electrons Reduction Cathode
Load
9
Solid
Electrolyte Electrolyte
Solid
Electrolyte
5 phases in 3 regions (n, s & p)
Isothermal 1D model: Mass and Charge balances for each phase.
5 phases x 2 balances =10 PDEs
The detailed electrochemical model is difficult to use for packs, on-board control , calendar / cycle life predictions, detailed parameters estimation, inclusion of complex degradation mechanisms etc.
Nonlinear PDEs
Linear ODEs
Nonlinear algebraic expressions
Approximations
Nonlinearity in a system is not bothersome if it can be relegated to
algebraic evaluations.
Ideal Reduction
Model reduction achieved : 10 coupled PDEs to 5 linear ODEs.
10
Li Battery: Model reduction methodology & Voltage predictions
nnninin
n
nninnn
in
n
ninn
n
nn
inin
nn
n
inn
nnninn
nn
n
inn
nnnn
jlatqdt
dq
D
l
dt
dcl
tqD
ltctc
xlDl
tqtctxc
tjtxjl
q
x
cD
x
jlatqdt
cdl
l
q
x
cD
x
jtax
cD
xt
c
13
3
2,
,&
1
1
22
2
2
2
22
2
2
22
22
2
222
222
2
2
2
222
22
22
Nominal current pulse Max % error 0.2%
High current pulse Max % error 2%
ROM: J of Power Sources, 2013; 222: 426
• Volume averaging reduces PDEs to ODEs. • But profile or gradient information is lost. It is
recovered using Profile based approximations, in terms of relevant internal variables.
• All approximations accurate on volume averaging.
n s p
Hydrogen Storage: Hierarchical modeling approach
Refueling << Discharge << Dormancy << Venting
minutes << hours << days << weeks
Slower processes Fast process
11
0-D bed model
System model
Transport resistances & Process design
Slow processes
Drive cycle response & System integration
1-D bed model Refueling
2-D bed model Flow maldistribution & Tank design
Engineering calculations
2-D pellet model Optimum pellet shape & size
MS Excel
Matlab
Simulink
Comsol
Comsol
Comsol
CGH2 supply
LN2 HX Ti = 80 K,
Pi = 153 bar
350 bar
Expander
JT valve
Tf = 68 K, Pf= 20 bar A
Engineering calculations – Process design
US Patent: 2009/0107155, Granted
0-D model results: Dormancy, Venting & Discharge
12 Int J of Hydrogen Energy 2009; 34: 5466
5 kg 300 miles 0.5 kg 30 miles
Tm,
hQ
Tm f ,
om
ff Tm ,
A
System simulation results: FTP-75 drive cycle, 0.23 kW heating rate
The adsorbed phase responds to the slowly
varying demand
The gas phase responds to the
fluctuating demand
13
hQ
Hot gas recirculation
Electrical heating
Int J of Hydrogen Energy 2012; 37: 2862
1D model results: Refueling & Optimal tank design
K 1380 T
K 80fT
14
Single axial flow bed Single radial flow bed 7-cartridge radial flow bed
Pa 9.455P Pa 6.99P Pa 7.4P
min. 24.4 Isobaric fillt min. 27.4 Isobaric fillt min. 46.4 Isobaric fillt
Int J of Hydrogen Energy 2010; 35: 3598
.
2-D model results: Flow maldistribution & Pellet design
sT
15
Int J of Hydrogen Energy 2011; 36: 15239
Under review in Int J of Hydrogen Energy, 2013
16
Nano-composites for Windshield Deicing
Incident beam
L1
Composite
x 0
L2
Ice
Tamb Tamb
x = L2 x = -L1
To be submitted to Int J of Heat and Mass Transfer, 2013
• 3D simulation showed that, the heat source could be homogenized. • An analytical Fourier solution was derived for the bilayer system with heat source in the composite layer.
Graphene based nano-composites, with electrical heating
US Patent: 2011/0297661, Filed
17
Nano-filtration membrane design for Sea-water Desalination
Accepted in Desalination (Elsevier), April 2013
Na+
Cl-
-
-
Cl- Cl-
Cl- Cl- Na+
Na+
Na+
Na+
Na+
Na+
Na+
Na+
-
-
• Donnan Steric Pore Model (DSPM) has Nernst-Planck equation for transport of charged species, along with electro-neutrality, dielectric exclusion etc. • A Reduced Order Model (ROM) was derived using constant potential gradient approximation, leading to an algebraic algorithm amenable for spreadsheet implementation, for rapid process evaluation. • With experimental data on reduced water flux at higher salinity, the mass transfer coefficient in the concentration boundary layer can be back-calculated – Concentration Polarization. • Using these mass transfer coefficients along with DSPM – ROM, the experimental salt rejections were well described, with membrane charge density and dielectric constant of ordered water within pores as the two model parameters.
Experimental data for properties / kinetics estimation & model validation. Applied math techniques for model solution methods.
Personal picks for potential future research
Meso scale: Voronoi based mesh-free methods, Continuum scale: Spectral methods, Model order reduction: Volume averaging & Profile based approximations.
Direct application of meso scale methods when length / time scales are not too large.