integrated hybrid life cycle optimization for multi-scale
TRANSCRIPT
Integrated Hybrid Life Cycle Optimization for Multi-Scale Sustainability Analytics of
Energy Systems
Fengqi YouRoxanne E. and Michael J. Zak Professor
Process-Energy-Environmental Systems Engineering (PEESE)School of Chemical and Biomolecular Engineering
Cornell University, Ithaca, New York
www.peese.org
3rd SEE SDEWES, Novi Sad, July 2018
Environmental Sustainability Issues
2
Life Cycle Analysis (LCA) – “Cradle to Grave”
3
Life Cycle Analysis (LCA)
4
• Quantifying environmental impacts of complex systems• Modeling the entire product/process life cycle• Holistic view of the system
Life Cycle Analysis
Life Cycle Optimization (LCO)
Integrating life cycle analysis approach with multi-objective optimization techniques
Life Cycle Analysis
5
Life Cycle Optimization: Theory and Methods
6
• Life Cycle Optimization▪ Life Cycle Analysis + Techno-economic Analysis + Design Optimization
Process Analysis
Process Design
Process Optimization
Life Cycle Analysis
Sustainable Design
Life Cycle Optimization
• Research Challenges▪ How to seamlessly integrate LCA into process systems optimization?▪ How to define the “optimal” systems boundary and functional unit?▪ How to incorporate state-of-the-art inventory analysis methods in LCO?▪ How to deal with uncertainty and solve large-scale LCO problems?
Process-based Life Cycle Optimization (LCO)
• Systems boundary must be defined in Phase I of LCA• Functional unit serves as the basis for calculation and comparison
Life Cycle Analysis
7
• Algae• Microalgae, cyanobacteria, & macroalgae• Non-food; high yield; rich in oil
• Algae-based biorefinery• Consume and utilize CO2; recycle nutrients & water• Produce fuels and value-added products• Process economics? Environmental sustainability?
Motivation
Chlorella Vulgaris
Biofuels
Bioproducts
8
• Optimal design and synthesis of algal biorefinery• Selection of technology, pathway, and processing methods• Determination of product portfolio under the given feed• Recycling nutrients, water and carbon dioxide• Mass balance, capacity, and equipment sizing• Energy and utility consumption• Process economics ? Techno-econmic analysis• Environmental sustainability ? Life cycle analysis• Cost-effective & sustainable design
Algal Biorefinery Process Design and Optimization
9
LCO for Sustainable Design of Energy Systems
10Gong & You (2015). Current Opinion in Chemical Engineering, 10, 77-86.
Superstructure of Algae Process
Harvesting
Lipid extraction
Biofuel production Bioproduct manufacturing
Remnant treatment
Cultivation
HydrogenPropylene glycol
Glycerol-tert-butyl etherPoly-3-hydroxybutyrate
BiodieselDiesel
Biogas utilization
11
Superstructure of Algae Process
<1,2> Flat plate photobioreator
<1,3> Bubble column photobioreator
<1,4> Tubular photobioreator
7,800+ processing pathways12
Optimization Model: Constraints for ONE Unit
{ }
,
,0
1, 0,1
4 defk de k
def DEde k d
de DE
de DE
e
DE DEde de
m m
m y UB
y y
∈
∈
=
≤ ≤ ⋅
= ∈
∑
∑( )
,/ /∈
+ ⋅ ⋅ = +
= ⋅ +
= ⋅∑
1 OP OP 1 2 OG1k k 2 k k
OG1 OP 2 OG1k k k k
steam1 SC C buf1 ms k
k Kk k
m SC X m m m
m SF m m
m MW R N m MW
• Technology selection • Mass balance and unit sizing
• Electricity
• Heating utility
• Cooling utility
( )∈
∈
=
= ⋅ ⋅ + ⋅
∑
∑
pce
turbine buf ng ng1
p
k co
ce PCE
k Km,k k
pct pc
ppt Y LHV m WF m
∈
= ⋅ ⋅ ⋅ ⋅
= ⋅
∑HE HChce1 hce1 k h
k Kce1,k
hce2 hce2 hce2
hc1 Y UC DTH CP mhc1
hc2 UHC mhc2
∈
= ⋅ ⋅ ⋅ ⋅
= ⋅
∑HE HCcce1 cce1 k c
k Kce1,k
cce2 cce2 cce2
cc1 Y UC DTC CP mcc1
cc2 UCC mcc2
= ⋅
sfequP CEPCI
equ equeccbequ equ mfb CEPCIB
equ equ
m Pecc P
P P
( )( )
1
1 1∈
⋅ + = ⋅ + ⋅
+ − ∑
LS
equequ
SE
LQU
IR IRaic K ecc lc
IR
= + + +aoc fsc utc omc wc
• Equipment capital cost
• Annualized investment cost
• Annual operating cost
= + +rma tran maghg ghg ghg ghg
• Greenhouse gas emissions
Mass and material balanceProcess network design specifications
Technology and pathway selectionEquipment sizing and capacity
Energy balanceUtility consumption
Techno-economic analysis
Life cycle environmental impact analysis
13
• Objectives:• Minimize: Unit cost of fuel product (techno-economic analysis)
• CAPEX + OPEX• Credit from selling by-products (glycerol, fertilizer, biogas, …)
• Minimize: Unit life cycle GHG emission (life cycle analysis)• Direct emissions: Cultivation, remnant treatment, & utility generation• Indirect emissions: External utility, e.g. electricity and steam, …
Optimization Model: Objectives
Systems boundary and life cycle stages of LCA
14
Pareto Optimal Curve
Minimum unit GWP
Minimum unit annualized cost
)
15
Pareto Optimal Curve
Minimum unit GWP
Minimum unit annualized cost
)
16
Superstructure of Algae Process
Harvesting
Lipid extraction Biofuel production Bioproduct manufacturing
Remnant treatment
Cultivation
Biogas utilization
17
2.71 – 3.78 14.43 Petroleum-derived diesel
Optimal Design of Minimum Unit Biofuel Cost
Unit cost of biofuel ($/GEG)
Unit GHG emissions (kg CO2-eq/GEG)
Biodiesel Throughput
(million GGE)Bioproduct
2.79 7.21 47 Propylene glycol
18
GWP of Algae-based H2, PHB, Propylene Glycol
Alternative bio-based propylene glycol is derived from soybean by ADM(R).
63%
51% 6%
19
ACS Sustainable Chem. Eng. 2015, 3, pp 82−96
20
Alternative approaches for Life Cycle Inventory (LCI) analysis• Process-based LCA• Economic Input-Output (EIO)-based LCA• Hybrid LCA
Hybrid Life Cycle Optimization (h-LCO)
21
(most widely used)(for macroscopic analysis)(state of the art)
Process-based LCA
22
Process-based
Resolution Process Sector Process (foreground)Sector (background)
Construction Bottom Up Top Down Hybrid
Scope Selected Processes Entire Economy Entire Economy
Process system boundary
Detailed process inventories
EIO-based LCA
23
Process-based EIO-based
Resolution Process Sector Process (foreground)Sector (background)
Construction Bottom Up Top Down Hybrid
Scope Selected Processes Entire Economy Entire Economy
Entire macroeconomy
Transactions among sectors
Sectors:Agriculture, mining, construction, manufacturing, wholesale trade, retail trade, transportation, etc.
Integrated Hybrid LCA
24
Process-based EIO-based Integrated Hybrid
Resolution Process Sector Process (foreground)Sector (background)
Construction Bottom Up Top Down Hybrid
Scope Selected Processes Entire Economy Entire Economy
Insights into Different LCA Approaches
25
Process-based LCA
26
Drawbacks:• System boundary truncation• Underestimation of the true impact
Advantage:• Specificity of process analysis
EIO-based LCA
27
Drawbacks:• Loss of precision at process level
Advantage:• Completeness of life cycle boundary
Integrated Hybrid LCA
28
Integrates process- and IO-based LCA
Advantages:• Completeness of life cycle boundary• Specificity of foreground processes
Toaster Example
29
(Functional unit: produce 1,000 pieces of bread)Comparing two toasters
Toaster Example
30
Processes SectorsPr
oces
ses
Sect
ors
AA ABB B
Toaster Example
31
18.7 17.7Direct emission
(process system)
Toaster Example
32
18.7 17.7
27.2 27.3
Direct emission (process system)
Full emission(process + IO systems)
Neglectedindirect emission
(IO system)34% 35%
Integrated Hybrid LCA
33
I1
I2
IM
…
Inputs
P_B2P_B1 P_BK
P_N2
…
P_N1 P_NK
…
…
……
P_A2P_A1 P_AK…
Processes
O1
O2
OP
…
Outputs
Processes Systems
S1
S2
S3
Sn
S1
S2
S3
Sn
Economic Input-output SystemsUpstream cutoffs
Downstream cutoffs
Sectors SectorsIntegrated Hybrid LCA:• Explicit process analysis – foreground process systems (precision of analysis)• EIO analysis – background macroeconomic systems (complement the truncated
system boundary)
Mathematical Foundation
34
1
Total environmental impact0
p dp io
u io
A C yE E
C I A
−− = − −
pE
ioE
Environmental extension factor (process systems)
Environmental extension factor (EIO systems)
pA Process matrix
Direct requirementsmatrixioA
uC Upstream cutoff matrix
dC Downstream cutoff matrix
• Unconventional natural gas from shale rocks
• Large-scale production due to hydraulic fracturing and horizontal drilling
• Half of the NG production in the U.S.
• Over 63,000 shale wells in the U.S.
Application to Shale Gas
35U.S. natural gas production
Hydraulic fracturing Horizontal drilling
Hybrid LCA of Shale Gas
36
• Climate change
• Water consumption
• Energy consumption
Goal and scope• UK shale gas• System boundary: well-to-wire• Functional unit: 1 MWh electricity generation from shale gas
Life cycle inventory• 40 basic processes in the process systems• Two-region IO model (UK-ROW) with 224 industrial sectors• Three cases from literature: best, balance, and worst cases
corresponding to the lowest, the medium, and the highest environmental impacts
Impact assessment• GHG emissions (100-year GWP factors;
CO2, CH4, N2O, HFCs, PFCs, and SF6)• Water consumption • Energy consumption
LCA of Shale Gas
37
Process Systems – 40 Basic Processes
38
Process ID Description Process ID Description
m1 Steel production, converter, chromium steel 18/8 m21Soda ash, dense, to generic market for neutralizing agent
m2Concrete production, for civil engineering, with cement CEM I m22 Sodium persulfate production
m3 Tap water production, direct filtration treatment m23 Sodium borates production
m4 Diesel production, low-sulfur m24 Citric acid production
m5 Diesel, burned in building machine m25 Pesticide production, unspecified
m6Diesel, burned in diesel-electric generating set, 18.5kW m26 N, N-dimethylformamide production
m7 Barite production m27 UK electricity generation, with mixed energy inputs
m8 Bentonite quarry operation m28Transport, freight, lorry, all sizes, EURO3 to generic market for transport, freight, lorry, unspecified
m9 Chemical production, inorganic m29 Injection in disposal well
m10 Chemical production, organic m30 Wastewater treatment by CWTm11 Lignite mine operation m31 Onsite treatment with MSFm12 Treatment of inert waste, inert material landfill m32 Onsite treatment with MEDm13 Treatment of drilling waste, landfarming m33 Onsite treatment- with ROm14 Silica sand production m34 Steam production, in chemical industrym15 Petroleum refinery operation m35 Tap water production, direct filtration treatmentm16 Isopropanol production m36 Transporting gas through pipelines
m17 Hydrochloric acid production, from the reaction of hydrogen with chlorine m37 Ethanolamine production
m18 Ethylene glycol production m38 Ethylene glycol productionm19 Potassium chloride production m39 Fugitive emissions of CO2
m20 Carboxymethyl cellulose production, powder m40 Fugitive emissions of CH4
Hybrid LCI Data Structure
39
IO System (896 × 896 matrix)
Agriculture
Mining
Energy
Transportation
Finance
Law
Communication
Real estate
Education
Recreation etc.
Sector groups:
• Multi-region: UK and ROW (rest of world)• Supply-Use Table (SUT): each containing 224 industrial sectors/products
LCA Results
40
• Electricity generation • Transportation
• Electricity generation • Processing
• Drilling • EIO system
Comparison with Existing Hybrid LCA Studies
41
• GHG emissions of shale gas are comparable to those of natural gas • Less GHG emissions than Coal and Oil
Activity – Linking SC Decisions with h-LCO
42
Definition: Activity is a flexible process that involves decision making.
Truck Rail Ship
Operation
Lorry
maintenance, lorry
disposal, lorry
Road
operation, maintenance, road
disposal, road
Operation, freight train
Locomotive
Goods wagon
Maintenance, goods wagon
Maintenance, locomotive
Disposal, locomotive
Railway track
Operation, maintenance, railway track
Disposal, railway track
Operation, transoceanic freight ship
Transoceanic freight ship
Maintenance, transoceanic freight ship
Port facilities
Operation, maintenance, port
?
Agriculture Mining Manufacturing Finance Law Energy
Unit priceUnit price
Unit price
Yue, Pandya, & You (2016). Environmental Science & Technology, 50, 1501–1509.
Hybrid LCO Model for Shale Gas
43
( )1min
opercap
tt T
TCTCdr
LCOETGE∈
++
=∑Economic objective:
Environmental objective: min pro IOTE TEUETGE+
=
s.t. Economic Constraints
Environmental Constraints
Mass Balance Constraints
Capacity Constraints
Composition Constraints
Bounding Constraints
Logic Constraints
IO IOn nssTE e P=
Total GHG emissions :
, ''
ns ns nsns
nNS
s nsP Paio UP∈
− ⋅ ≥∑
Total output of each industrial sector Pns
,nsn m mm
s mM
c price QUP∈
= ⋅ ⋅∑
Upstream input from industry sector ns to process systems
n
sfp
prpca
ocp
p P
PCC
ppr
rcpci
rpcii
∈
= ⋅
⋅∑
Nonlinear term:
Mixed-Integer Nonlinear Fractional Program
mpro pro
mTE e Q=
Case Study of UK Shale Gas Supply Chain
44
• 15 Shale sites (7 existing, 8 potential ones)
• 4 processing plants (2 existing, 2 potential)
• 6 CCGT power plants
• 10-year planning horizon (40 time periods)
MINLP problem: • 414 integer variables• 11,797 continuous variables • 15,370 constraints
Pareto-optimal Curve
45
Drilling Schedules and Production Profiles
46
Drilling Schedules Production Profiles
Supply Chain Design and Flow Information
47
ACS Sustainable Chem. Eng. 2016, 4, pp 3160-3173
48
LCO: Attributional v.s. Consequential
49
Multiobjective Optimization
Goal and Scope Definition
Life Cycle Inventory Analysis
Life Cycle Impact Assessment
Interpretation
Life Cycle Assessment
Automatic generation of system design decisions
• Attributional LCA
• Consequential LCA
Motivating Example
50
Environmental Impacts of producing A
Environmental Impacts of the conversion
Environmental Impacts of end of life of B+ +
(2)Environmental Impacts
of the new system -(1)
Environmental Impacts of the original system
Environmental Impacts of
conversion (2)
Environmental Impacts of
conversion (1)-Environmental
Impacts of end of life of B
Environmental Impacts of
end of life of C-+
Or
Attributional LCA: static and fact-based
Consequential LCA: dynamic and change-driven
Consequential Life Cycle Optimization
51
Consequential LCA
Techno-economic Analysis
• What upstream and downstream processes are influenced by the target process?
• How does the target process influence the upstream and downstream processes?
Consequential Life Cycle OptimizationMultiobjective Optimization
Process Models
How does it work?
An Analogy – Spot the Difference
52
Before After
Attributional LCA for Process Design Problems
53
Environmental impacts in the
use and end-of-life phases
Environmental impacts in the feedstock
production phase Environmental impacts from the process
Environmental impacts from transportation
• Not suitable for new systems• Overlook the power of markets and influences in other processes
• Applicable to existing systems
System Boundary of the Consequential LCO
54
Partial Equilibrium Model
55
Supply Demand
( )
1
1
1
,
,
1
,
,
,
,
l
l
l
l l,p l,p l,p l,p lp PPC
l,p l,pl,p
l,p l,p
l,p l,p l,p l,p
l,pp PPC
l,p lp PPC
l,p l,p l,p l,p l,p
AS asc XS bsc YS
ns nsasc
ms msbsc ns asc ms
PS l
l,
YS
XS PR
ms YS XS ms YS
p
l, p
l
l
l, p
∈
−
−
∈
∈
−
= ⋅ + ⋅ +
−=
−
= − ⋅
∀
∀
∀
∀=
=
⋅ ≤
∀
≤ ⋅ ∀
∑
∑
∑
( ) ,1l ll l l l l
l
edAD PR PDd leα αβ⋅
= ⋅ + − + ∀⋅
Price elasticity of demandAggregate supply function
Quantities by the target process
,l lAS AD l= ∀
Equilibrium
( )( )
,
,
Supplierl l llcustomerl l l l
LCI AS as0
L
PS l
PC DI D lA ad0
= − −
= − −
∀
∀Life Cycle Inventory
,l las0 ad0 l= ∀
Consequential LCO framework
56
Economic Objectivee.g. maximize net present value
Environmental Objectivee.g. minimizing ReCiPe points
Process ModelInteger variables for technology selection; Mass and energy balance
Market ModelPartial equilibrium models
s.t.
( ),,
max , , ,k l l l k kk l
h P Q X YC∑
( ), , , l k l k kk
Q f X YP l= ∀∑
( ), , , l l l l lAS m Q P YS l= ∀
, l lAS AD l= ∀
( ), , , ,, ,
min , ,l r s l r s l l ll r s
c v Q AS AD ⋅ ∑
( ), , , l l l l lAD n Q P YD l= ∀
Application to Algae-based Biofuel Production
57
Scenedesmus sp. with 27.4 wt % lipid
Renewable dieselFunctional Unit:1 GJ of renewable diesel
Detailed superstructure
586 markets in the U.S.
Fertilizer
Natural Gas
Electricity
Hexane
Hexane
Diesel
Diesel
Fertilizer
Natural Gas
Optimization Results for ReCiPe
59
Production Rate:3.5 MMGal/year
Environmental Impact Breakdown
60
Primary Markets Attributional Consequential
-7,000 -6,000 -5,000 -4,000 -3,000 -2,000 -1,000 0 1,000 2,000 3,000 4,000
Consequential
Attributional
Annual ReCiPe endpoint score (kPt./year)
Urea production DAP production Hexane production Electricity productionDiesel production Fertilizer consumption Fertilizer adjustiment Diesel consumptionDirect emissions Transportation
Diesel Combustion in end of life(1) Displace fossil diesel; (2) Transportation market
(diesel ↑; gasoline ↓)
Fertilizers Increase fertilizer production(1) Increase but not as much;
(2) Crops market (foreign ↑; domestic ↓)
Hexane Positive characterization factors*
Negative characterization factors*
*Data for “rest of the world” from Ecoinvent 3.3
Consequential Environmental Profile
61
ACS Sustainable Chem. Eng. 2017, 5, pp 5887-5911
62
• Life cycle analysis and life cycle optimization• Process-level LCA and life cycle design/optimization
• Systems boundary• Functional unit
• Integrated hybrid LCA and LCO • Process systems to supply chain, and to macroeconomics scales
• Consequential LCA and LCO• Dynamic and change-driven• Suitable for new product systems to account for influences of other
processes through the market
• Applications to energy systems• Algal biorefinery• Shale gas • ……
Conclusion
63
64
http://you.cbe.cornell.edu