An Operational or An for Hourly‐varying electricity...
Transcript of An Operational or An for Hourly‐varying electricity...
An Operational Model for Optimal Non‐Dispatchable Demand Response
for Continuous Power‐intensive Processes
oror
An Operational Model for Optimal Production Scheduling under Hourly varying electricity pricesunder Hourly‐varying electricity prices
for Continuous Power‐intensive Processes
Sumit Mitra, Ignacio E. Grossmann
EWO Meeting, 03/09/2011
1
Perspective of the Future Electric Power System
Renewable StFACTS,
Main challenge: balance supply and demand This is the motivation for “smart grids”!
Renewable StorageFACTS,
$
Demand Response
EnergyStorage
HVDC IndustrialCustomer
PEV
Renewable Energy
Source: U.S.-Canada Power System Outage Task Force
High Observabilityenabled by increased communication capabilities
2
enabled by increased communication capabilities
This slide is based on a presentation slide by Prof. Gabriela Hug, Department of Electrical Engineering, Carnegie Mellon University, ESI seminar, January 13 2011
Demand Response (DR): Overview and recent work for industrial processes
B i id T b l l d d d f tBasic idea: To balance supply and demand of a power system, one can manipulate both: supply and demand demand response (DR).Utility provider perspective: distinguish dispatchable and non‐dispatchable DR
ABB, CMU (2009, 2010)(Cement)
F BOC R t (2002)
Al O k Rid N i l L b (2009)
Former BOC, Rutgers (2002)UBuffalo (2007)(Air separation)
Goal of our current research: Develop a generic model for
Air Products, Lehigh (2010)(Air separation)
Alcoa, Oak Ridge National Lab (2009)(Aluminum)
Develop a generic model for non‐dispatchable DR
3
( p )
Other power‐intensive processes:‐ Chlor‐alkali synthesis‐ Paper production
Air Separation Plant
Industrial customer’s perspective: Problem Statement for an Air Separation Plant
Due to Air Separation Plant
Liquid Oxygen
Due to compressors!
Electricity costs
Liquid Nitrogen
Liquid Argon
Gaseous Oxygen
Gaseous Nitrogen
Strategic DecisionsOperational DecisionsPrice forecast
Demandforecast
InvestmentCosts
Additional equipment?
Additional tanks?
Equipment upgrade?
• When to produceand amounts?• Inventory levels?
4
q p pg
Overview: Research Methodology
Step 4a (projected)Model uncertainty in electricity prices
data
stochastic
St 1 St 2
stochastic
Step 1Obtain an efficientmodel for the short‐term scheduling
Step 2Model investment
decisions:additional equipment,upgrade equipment,
Step 3Develop a suitable
algorithm to solve thelarge‐scale optimization
problem
deterministic
add. storage tanks;(retrofit, multi‐period)
problem
decisionsoperational strategicSingle plant
Focus of this talk
Step 4b (projected) Multiple plants
decisions
solution methodmodeling
operational strategicSingle plant
5
p p
Operational Model
Objective function
Feasible region Mass balances
Logic constraints for transitions
6
Operational Model
Feasible region
ibl i j i i d
Feasible region
Mass balances
Feasible region: projection in product spaceModes: different ways of operating a plantEnergy consumption: requires correlation with production levels Mass balanceswith production levelsMass balances: differences for products with and without inventory
7
Modeling of transitions
Off Trans OnImmediate [2] After some hrs
Immediate [1]
St t di f t iti
Immediate [1]: Minimum uptime Immediate [2]: Minimum downtime Link between state and transitional variables
State diagram for transitions
Enforce minimum stay in a modeLogic constraints for transitions
Forbidden transitions
Coupling between transitions
Modes: different ways of operating a plantTransitions: between modes to enforce e.g. min. uptime/downtime
8
Result: Comparison of production schedules
Electricity consumption profiles for test case B Inventory profiles for test case B
9
Case studies: Air separation plantsProblem sizesProblem sizes
For the same demand, P2 can produce cheaper due to higher flexibility
cyclic is theharder problem
Savings compared toconstant operation
Demand Savings CPU time
Demand Savings CPU time
D i d d h d t l d t hi h ti l fl ibilitComputational results
10
Decreasing demand = harder to solve due to higher operational flexibilityComputational results
Synergies for integration of operational and strategic decisions
OperationalCosts
P1
P2
Regime of feasible operationOperational
Improvements(Scheduling)
StrategicImprovements
(Investment decisions)
( g)
Operational Flexibility
Take‐home message: Operational flexibility is key to achieve economic savings with demand response.
11
Operational flexibility is key to achieve economic savings with demand response.