EnterpriseWide Optimization in an Integrated Chemical Complex

1
Enterprise-Wide Optimization in an Integrated
Chemical Complex

• John Wassick
• The Dow Chemical Company

2
Acknowledgements
Naoko AkiyaScott BuryJeff FerrioRamkumar
Karuppiah Bikram Sharda Jim SturnfieldAdriana
Vazquez
3
Outline

• Dow an integrated chemical company
• Product integration and site integration
• Production Planning Across a Product Envelope
• Energy System Scheduling
• Site Design
• Waste Treatment Scheduling
• Concluding Remarks

4
Dows Business Groups
In early 2007, Dow's global business portfolio
was redefined. It is structured around eight
distinct groups each of which will be led by a
business president. Six of these groupings are
new, while two were already in place. Basic
Plastics Basic Chemicals Polyurethanes Designed
Polymers Latex Epoxy Intermediates /
Specialty Chemicals Specialty Plastics Dow
Agrosciences Hydrocarbons Energy
5
Dow products are used in many consumer markets
including
Building Maintenance and ConstructionTransportat
ionFurniture and FurnishingsPaper and
PublishingHome Care and ImprovementPersonal
and Household CareFoodHealth and
MedicineWater Purification
6
2006 Financial Performance
Sales by Operating Segment (dollars in millions)
Basics Portfolio
Performance Portfolio
Performance Plastics 13,044
Performance Chemicals 7,867
Basic Plastics 11,833
Basic Chemicals 5,560
Hydrocarbons and Energy 6,206
Unallocated and Other 316
Agricultural Sciences 3,399
Sales by Geographic Area (dollars in millions)
Rest of World
Europe (Includes Middle East and Africa)
United States
18,172
17,846
13,100
7
Accelerating Strategic Growth
Technology and RD
8
Product Integration

• Most Dow products are derived from basic
  chemicals feedstocks produced by Dow
• Product integration leads to facility integration
• Derivatives and their feedstock are called an
  envelope
• Example Chlorine Envelope

9
Chlorine Envelope
Polycarbonate
Epoxy Resins
Polypropylene Glycols
Phosgene
Polyurethanes
Epichlorohydrin
Glycerine
Propylene Oxide
Propylene Glycols
Ethylene Glycol
Allyl Chloride
Ethyl Chloride
Ethylcellulose
Ethylene Oxide
Ethylene Chlorohydrin
Methyl Chloride
Methylcellulose
Trichlorethane
Vinylidene Chloride
Polyvinylidene Chloride
Chloroform
Intermediates
Ethylene Dichloride
Carbon Tetrachloride
Vinyl Chloride
Hydrochloric Acid
Perchloroethylene
Methylene Chloride
Trichloroethylene
Cl2
10
Partial Network of an Integrated Site
Internal Users
Internal Source
Plant
Product
Export
Imported Material
Imported Material
Plant
Product
Plant
Product
Plant
Product
Internal Source
Plant
Plant
Product
Plant
Plant
Imported Material
Product
Internal Source
Plant
Imported Material
Product
Limited intermediate storage between plants
11
Envelope Integration
Envelope B
Envelope D
Envelope A
Envelope C
Utility Envelopes Waste Treatment Water
Energy
12
Network of Integrated Sites
Stade, Germany
Plaquemine, LA
China
Freeport, TX

• 27 billion lbs/yr
• 11.7 billion lbs internal use
• 44 of Dow US sales
• 21 Global sales
• 2,500 acres Hydrocarbons storage
• 13 dock facilities
• 5 deep-water berths

Saudi Arabia
Libya
Aratu, Brazil
13
Planning Across the Envelope

• Plan monthly production for chlorine and
  derivative plants
• Satisfy global customer demand across a wide
  variety of markets
• Achieve mass balance across the envelope
• Plan purchase of key intermediates
• Schedule internal shipments of intermediates
• Account for planned maintenance shutdowns
• Respond to unanticipated events

Traditional planning focused on material balance
with limited economics
14
Mathematical Approach

• Constraints
• Balancing each material inbound/outbound at each
  facility, site, region
• Satisfying demand for each material in each
  region
• Satisfying production capacity constraint
• Inventory mass balance
• Derivative demand contracts
• No chlorine shipped by marine or truck (safety)

15
Size of the Problem

• Sites (4)
• Production plants (26) not all plants at all
  sites
• Sales Regions (9)
• Materials (34)
• Products (14)
• Raw materials (6)
• Intermediates (14)
• Model
• 26195 constraints
• 28529 variables
• 60769 Non zeroes
• Solving time lt 1 minute

16
Components of the Energy Envelope
External Gas Market
Gas Storage
External Power Market
Cogeneration Unit Power Steam
Cogeneration Unit Power Steam
Cogeneration Unit Power Steam
Standby Boiler Steam
Standby Boiler Steam
PRV
Power Generating Steam Turbines
PRV
Internal Power Market
Internal Steam Market
17
Energy System Scheduling

• Goals
• Produce steam to satisfy site needs
• Purchase and store natural gas as needed
• Produce and purchase electrical power to satisfy
  site needs
• Produce electrical power for export when
  advantageous
• Operate system at economic optimum
• Participation in external markets
• Operation of system assets

18
Energy System Markets
19
Comparative Power Price Early Aug 2007
20
Monthly and Daily Gas Market Fuel Prices
21
Energy System Scheduling Challenges

• Energy system participates in several markets
• Site energy market as a producer
• External electrical energy market as a producer
  and consumer
• Fuel market as a consumer
• Problem has multiple time dimensions
• Months
• Days
• Minutes
• Many factors that have significant uncertainty
• Power and steam needs of the site
• Effects of weather
• Fuel and power price,
• Operations of the site energy system

22
Designing the Integration of a Site

• Design Goals
• Individual process units meet long term capacity
  targets in spite of
• Planned and unplanned plant outages
• Temporary degradation of plant performance
• Short term raw material shortages
• Planned maintenance turnarounds
• Synchronized maintenance turnarounds among plants
• Support shutdown and restart response to
  different failure modes
• Minimize overall capital and operational costs

23
Integration Design Decisions
Product A 1 trains? 2 trains?
Product B 3 trains
MTBF - 6 Yr. MTTR - 12HRS
MTBF - 5 Yr. MTTR - 16 HRS
Product A Tank
Raw Material A
MTBF - 6Yrs MTTR - 12 HRS
MTBF - 2 Yr. MTTR - 10 HRS
MTBF - 5 Yr. MTTR - 16 HRS
How much Storage?
MTBF 3 Yr. MTTR - 5 HRS
2 x 50 or 2 x 75?
3 x 50?
24
Mathematical Approach

• Constraints
• Material balance at each plant
• Certainty of satisfying customer/region demand
• Inventory mass balance
• Expected plant production capacity
• Expected plant online time

25
Reliability Optimization of an Integrate Site
Challenges

• Optimizing tank inventory levels with uncertain
  failure events
• Maintaining mass balance around storage tanks
  when events occur at random times and random
  durations
• Inventory may accumulate or be consumed due to
  failures these two failure modes occur
  separately and randomly
• Optimization model tends to optimize best case
  unless we specify worst case situations, it will
  not choose them
• How to take advantage of heuristics to simplify
  the optimization

26
Components of the Waste Treatment Envelope
27
Waste Treatment Scheduling

• Goals
• Operate system within environmental constraints
• Insure system does not constrain production
  plants
• Manage system inventory
• Operate system at lowest cost
• Schedule material transfers and waste treatment
• Account for planned maintenance shutdowns
• Respond to unanticipated events

28
Resource Task Network
Nk,t number of occurrences of task k at time t
?k,t continuous extent of task k at time t
(e.g. amount of transfer) ?k,t duration of task
k
Shared header
Tank to truck by shared header
Waste in tank
Waste in truck
Tank nozzle
?krt amount of resource r consumed/produced
relative to the ?k,t µkrt amount of resource r
consumed/produced per occurrence of task k
29
Resource Task Network Basic Constraints
30
Resource Task Network Supporting Constraints
31
Objective Function
32
Waste Treatment Resources
109 Resources in RTN
33
Waste Treatment Tasks
79 Tasks in RTN
34
Transfer to Treatment Unit Task
Treatment Chemicals
Shared header
Tank Nozzle
Pre-treated waste
Tank to burner by shared header
Waste in tank
Burner nozzle
Volume Capacity
Chemical Capacity
35
Waste Treatment Objective Function
36
Optimal Schedule versus Manual Schedule
Optimizer avoided off site processing
Optimizer reduced costs at the largest treatment
unit
37
Final Comments

• Additional EWO Opportunities
• Real time optimization of the envelopes
• Strategic site selection and design
• Optimization of the site as a single enterprise
• Unresolved Modeling Challenges
• Representing complicated operating policies
• Discrete decisions
• Complicated constraints
• Influence of decisions in other time frames
• Modeling uncertainty and risk