Supply Chain Planning and Design for Biodiesel Production Via Wastewater Sludge Sandra Duni...

Preview:

Citation preview

Supply Chain Planning andDesign for Biodiesel Production Via

Wastewater Sludge

Sandra Duni Eksioglu, PhDIndustrial Engineering Department

Clemson University

International Congress and Expo on

BIOFUELS & BIOENERGYAugust 25-27, 2015

Bioenergy: Opportunities Bioenergy production is expected to increase:

RFS: Production of biofuels by 2022: 36BGY o 2013: 14BG of corn-ethanol & 1BG of biodiesel

It is a clean and renewable source of energy.

It reduces the risk of oil embargos, price strikes, geopolitical dependence

It supports US farmers and local economy

www.wfpa.org

Bioenergy: Major Challenges Technological challenges

Food versus fuel debate

Biomass logistical/transportation challenges:Biomass supply is constrained by land availability Biomass is seasonalBiomass looses dry matter with time Production yields are uncertain Bulky & difficult to transport Widely dispersed geographically Small & medium-sized farms

US is committed to reduce GHG emissions by 17% below 2005 levels by 2020

Environmental policy focuses on physical processesEnergy efficient facility/vehicles: USDA Alternative Fuel & Fleet Efficiency programs Alternative fuels: Biodiesel Income Tax Credit, Excise Tax Credit, Alternative Fuel Excise Tax

Focus on physical processes overlooks the impact of business processes and operational practices on emissions

Inventory replenishment decisions impact emissions Inventory Inventory

Outsourcing, centralized warehousing, rapid-response logistics, just-in-time production, etc. impact emissions.

Bioenergy and Environment

BIODIESEL Supply Chain Structure

Municipal WWT Plants

Pulp & Paper

Meat Packing

Poultry Slaughtering &

Processing

Animal & Marine

Fresh/Frozen Fish

Sludge Supply Biocrude Plants

Diesel Plants

Customers

Research QuestionsEstimate supply-chain related costs for production of biodiesel.

Biomass transportation (sludge):What factors have a great impact to the transportation cost of sludge?Under what conditions pipeline becomes a viable transportation mode?

Supply chain design and management:Should a biocrude plant be co-located at a WWT facility? What factors have a great impact on the supply chain costs?

Provide insights about biodiesel supply chain related costs to potential investors.

BIODIESEL Supply Chain

Solution Approach

Sludge TransportationTechno-Economic

Analysis

Sludge Supply AnalysisRegression Analysis

GIS Tools

Supply Chain Design & Mgmt.:Bi-level stochastic Optim. Model

Model Validation & VerificationGIS Tools

Case StudyNumerical Analysis

Input Analysis

Transportation Cost Analysis

I. Facility-owned Single Trailer Truck, 30m3 capacity.

Similar analyses is conducted for rented Single Trailer Truck, facility-owned and rented Tandem Trucks of 30m3 and 40m3 capacity.

(a) Fixed costs ($/m3)

Fuel cost

Labor cost

Maintenance and repair costTire cost

Cost of ownershipAnnual sale taxesLicense fees and taxes

Management and overhead costInsurance cost

(b) Variable costs ($/m3/mile)

Transportation Cost Analysis

II. Pipeline Transportation of SludgeRaw Activated

SludgeDensity 1100 @ 22 0C 1200 @ 25 0C kgm-3

Viscosity 1.0 @ 22 0C 1.5 @ 25 0C cp

Total solids 0.50% 5.00%

ParameterEnhanced

Activated SludgeUnit

0

0.05

0.1

0.15

0.2

0.25

0.5 1.5 2.5 3.5 4.5 5.5

VC

Tra

nsp

orta

tion

($/

m3/

mile

)

Slurry Design Velocity (m/sec)

150 m3/day

320 m3/day

480 m3/day

700 m3/day

1,000 m3/day

2,000 m3/day

Smaller the capacity, higher transp. costs.

Transportation Cost Analysis Data: Mississippi Department of Environmental Quality. Pipeline vs. truck for volume of sludge shipped: 843.5 m3/day

Variable cost ($/mile/m3) is smallest for facility-owned tandem trailer truck.

As transportation distance increases, pipeline costs decrease.

0.04

0.05

0.06

0.07

0.08

0.09

0.1

0.11

0 200 400 600 800 1000

Pipeline Transportation

Tandem Trailer (own)

Single Trailer (own)

Tandem Trailer (rent)

Single Trailer (rent)

Distance (miles)

DV

C (

$/m

3/m

ile)

Summary: Sludge Transportation Costs

150 250 350 480 700 1000

Own Single Trailer Truck 0.101 0.103 0.103 0.101 0.103 0.103

Rent Single Trailer Truck 0.131 0.131 0.133 0.131 0.133 0.133Own Tandem Trailer Truck 0.098 0.098 0.098 0.096 0.098 0.096Rent Tandem Trailer Truck 0.125 0.125 0.125 0.123 0.125 0.123Pipeline (100 miles) 0.301 0.215 0.175 0.144 0.115 0.092

MODE OF TRANSPORTATION

Volume Shipped (m3/day)

$/gallon of biodiesel

25 50 75 100 125 150 175 200

Own Single Trailer Truck 0.077 0.096 0.114 0.133 0.152 0.171 0.191 0.208

Pipeline (150m3/day) 0.077 0.152 0.226 0.301 0.375 0.449 0.524 0.598

Pipeline (480m3/day) 0.037 0.073 0.109 0.144 0.180 0.216 0.251 0.287

Pipeline (1,000m3/day) 0.024 0.047 0.070 0.093 0.115 0.138 0.161 0.184

MODE OF TRANSPORTATION

Distance (miles)

$/gallon of biodiesel

A Two-Stage Stochastic Prog. Model

First-stage decisions x

Uncertainty()

Second-stage decisions y(, x)

Biomass supply uncertainty

Strategic decisions:-Plant locations/sizes-Pipeline location/size-Nr. of trucks purchased

Planning decisions:-Production-Transportation-Shortage

Model Formulation: Constraints

Supply Point k

SupplyPoint k+1

)(1 kjmy

)(1,1,1 mjky

Plant j

Plant j+1

)()(1 1

1 k

J

j

M

mkjm Sy

Sludge SupplyTrans. Mode m

)(2 jiyRefinery i

Refinery i+1

Plant j

Plant j+1)(2

1,1 ijy

L

lljl

I

iji xBCy

11

2 )(

i

J

jji BdCy

)(1

2 Production Capacity

F

ffkjfkjm zPCy

1

1 )(

jkjkjm trCapy 11

1 )(

K

k

M

m

I

ijikjm yy

1 1 1

21 )()(

)()(1 1

32

J

j

G

gigji yy

Pipeline & Truck Capacity

Flow Balance

)(1,1 jmky

Model Formulation: Constraints

I

iggig by

1

3 )()(

Demand is Satisfied

3igy

31,1 tgy

Customer g

Customer g+1

Marketg

1gRefinery i

Refinery i+1

}1,0{ljx

}1,0{fkjz

Binary Const.

ngigjikjm Ryyy )(),(),(),( 321

Zj

Non-Negativity Const.

Solving the Two-Stage SP Model

NO

Solve sub-problems(yn(), n) , Calc. UB

n = n +1 Add nth optimality

cut to MP.

Solve the Master Prob. (xn, zn, vn) & LB

L-shaped algorithmInitialize; n = 1

UB – LB < n > Nmax

Stop; Report solution

YES

1. Using CPLEX

2. Lagrangean Relax.

Solve Master Prob.

Solve Master Prob.

1. One aggregate cut2. Multiple cuts

Optimality Cut

Optimality Cut

Scenario Definitions & Probabilities:Supply Uncertainty

Scenario Explanation Probability

1-5 Historical data collected from previous five years 0.15*5 = 0.75

6 Sludge supply of every facility is increased by 20% 0.05

7 Sludge supply of every facility is decreased by 20% 0.05

8 Supply from WWT plants changes by as much average change of population in MS in the last 10 year

0.05

9 Sludge supply from pulp & paper industry decreases by 20% 0.05

10 Sludge supply from poultry increased by 20% 0.05

Computational Results

Val. of Stochastic Solution $3. 36M

Stochastic Solution Costs: $311.60 MCapacity: 80MGYProduction: 75.59MGY

Expected Val. Solution:Costs: $314.96 MCapacity: 75MGYProduction: 72.55MGY

Inv. Costs Prod. CostsTrans. Costs

Distribution of Unit Cost ($2.72/gal)

1. Large capacity plants centrally located.2. Co-locate with large WWT facilities.3. Truck transportation

Modeling Environmental Policies

Evaluate the impact of environmental policies on supply chain operations.

The following constraints is added to the mathematical model:

for each

Modeling Carbon Cap

capK

k

J

j

M

m

J

j

I

i

I

i

G

gigigjijikjmkjm Cyeyeye

1 1 1 1 1 1 1

332211 )()()(

The following term is added to the objective function:

Where, is the carbon tax (in $/kg).

Modeling Carbon Tax

I

i

G

gigig

K

k

J

j

M

m

J

j

I

ijijikjmkjm yeyeyep

1 1

33

1 1 1 1 1

2211 )(**)(**)(**

Modeling Carbon Cap-and-Trade The following constraints is added to the mathematical model:

The following term is added to the objective function:

cp is the market price of carbon (per ton).

each for 0)(

0)(

)()()()()(1 1

33

1 1 1 1 1

2211

ct

ct

Cctctyeyeye capI

i

G

gigig

K

k

J

j

M

m

J

j

I

ijijikjmkjm

))()(( ctctcp

Modeling Carbon Offset The following constraints is added to the mathematical model :

The following term is added to the objective function:

co is the offset price of carbon (per ton).

))((0 ctc

eachforct

Cctyeyeye capI

i

G

gigig

K

k

J

j

M

m

J

j

I

ijijikjmkjm

0)(

)()()()(1 1

33

1 1 1 1 1

2211

Computational Results

Problem parameters: (94/26/10/52/3/3/s) Stopping criteria: Error gap < 1% OR Nr. of iterations 1,000.

Problem size: 7,436 binary and integer variables; 8,242 continues variable (per scenario)

0

5,000

10,000

15,000

20,000

25,000

5 6 7 8 9 10

CP

U t

ime

(sec

)

Number of scenarios

BSCN-LBSCN-MLBSCN-LR-LBSCN-LR-ML

CPLEX could not find a solution

within 1% error gap in 36,000

CPU sec.

BSCN-L BSCN-LR-L

Stopped due to iteration with

(2.77-4.29)% & (3.51-5.68)%

opt. gapBSCN-ML

BSCN-LR-MLStopped due to

1% error

Transportation under Regulatory Policies

(a) Carbon Cap 2800 tons/year

(c) Carbon Cap 2000 tons/year

(b) Carbon Cap 2400 tons/year

Jackson POTW, Hinds County

Weyerhaeuser Co. Pulp & Paper

Complex, Lowndes County

Hattiesburg South Lagoon, Jones County

Forest POTW, Scott

County

Peco Foods, Madison County

Oxford POTW, Lafayette County

Summary & Conclusions Transportation activities in the supply chain will add on

average $0.16/gal to the cost of sludge-based biodiesel.

Investments in improving biocrude technology will have a great impact on biodiesel production level and costs.

Carbon regulatory policies will have an impact on supply chain operations. Shifting transportation modes from truck to pipeline.

Stochastic programing model provides better solutions to our problem.

Research Team Department of Industrial Eng., Clemson Univ.

Sandra D. Eksioglu, PhDE-mail: seksiog@clemson.edu

Mohammad MarufuzzamanE-mail: maruf237@gmail.com

Department of Chemical Eng. Mississippi State Univ.

Rafael Hernandez, PhD

Todd French, PhD

Andro Mondala, PhD

Department of Civil & Env. Eng. Mississippi State Univ.

Dennis D. Traux, PhD

Sandra D. Eksioglu, PhDClemson UniversityDepartment of Industrial Engineeringseksiog@clemson.edu

QUESTIONS

Theoretical Results