Carnegie Mellon
Potential of Hydro Power and Storage
for the Integration of Wind Generation
CMU Electricity Conference
March 9th, 2011
Gabriela Hug
Assistant Professor
ghug@ece.cmu.edu
1
Carnegie Mellon
O tli
Outline
•
•
•
•
•
Introduction
Control Concept & Modeling
Case 1: Wind and Run-of River Power Plants
Case 2: Generation/Storage Dispatch
Conclusions
2
Carnegie Mellon
I t d ti
Introduction
•
Goal
– up to 20% wind penetration by 2030
•
Challenges:
– Intermittency and variability
– Missing infrastructure
•
Balancing Potential:
–
–
–
–
Storage
Demand response
Conventional generation
Curtailment
3
Carnegie Mellon
H d P
Hydro
Power
•
Types of Hydro Power
Run-of River Power Plant
Storage Power Plant
Pumped Hydro Power Plant
4
Carnegie Mellon
O tli
Outline
•
•
•
•
•
Introduction
Control Concept & Modeling
Case 1: Wind and Run-of River Power Plants
Case 2: Generation/Storage Dispatch
Conclusions
5
Carnegie Mellon
C t l Concept
Control
C
t
•
Predictive Control
– Use model of plant to be controlled to predict influence of input
– Choose input sequence which
gives best performance over
horizon
– Apply first step and measure
new state
6
Carnegie Mellon
M d li
Modeling:
Storage
St
•
Storage
Limits on
– Storage size
– Charging and
discharging rate
No simultaneous charging and discharging
7
Carnegie Mellon
M d li
Modeling:
Hydro
H d Power
P
•
Pumped Hydro Power Plant
•
Storage
g Power Plant
1
β
⋅ PSO (k )
QS I (k )
8
Carnegie Mellon
M d li
Modeling:
Hydro
H d Power
P
•
Run-of River Power Plant
– Retention differentiates a river from a tank
– Goal:
• Minimize discharge variations
• Minimize deviations of water levels from reference value (and keep within
limits)
9
Carnegie Mellon
M d li
Modeling:
Hydro
H d Power
P
•
Saint Venant Equations
=> dependency between water discharge and water level at each individual
point in the river
=> Linearization and discretization in time and space
•
Dependency between discharge and electric power
10
Carnegie Mellon
M d li
Modeling:
Generation
G
ti and
d Load
L d
•
Conventional Generation
– Capacity limit
– Ramp rate
•
Intermittent Generation
– Predictions of output
– Allow curtailment
•
available
Load
– Predictions of demand
– Allow demand control
available
11
Carnegie Mellon
O tli
Outline
•
•
•
•
•
Introduction
Control Concept & Modeling
Case 1: Wind and Run-of River Power Plants
Case 2: Generation/Storage Dispatch
Conclusions
12
Carnegie Mellon
C
Case
1
1: Wi
Wind
d and
d Run-of
R
f Ri
River Plants
Pl t
•
Objective Function
minimize discharge changes
minimize level deviations
•
Constraints
smoothen wind power
– River
Ri
flflow model
d l
– Constraints on water level and turbine/weir discharges
13
Carnegie Mellon
C
Case
1
1: T
Testt S
System
t
•
Cascade of four run of river power plants (20km apart)
•
•
•
•
•
Operating Point: 3000m3/s (1200m3/s through weirs)
Water Level Constraints: ±12cm
Weir discharge and inflow constant
10% rms prediction error
2 hours prediction horizon
horizon, 5 minute resolution
14
Carnegie Mellon
C
Case
1
1: Si
Simulation
l ti Results
R
lt
15
Carnegie Mellon
C
Case
1
1: Si
Simulation
l ti Results
R
lt
16
Carnegie Mellon
O tli
Outline
•
•
•
•
•
Introduction
Control Concept & Modeling
Case 1: Wind and Run-of River Power Plants
Case 2: Generation/Storage Dispatch
Conclusions
17
Carnegie Mellon
C
Case
2
2: G
Generation/Storage
ti /St
Dispatch
Di
t h
•
Economic Objectives
– Minimize generation costs
– Minimize conversion losses
•
Environmental Objectives
– Minimize CO2 emissions/cost (natural gas, coal)
– Minimize impact on water flow (hydro)
•
Quality of Service
– Minimize demand side management
– Minimize wind curtailment
18
Carnegie Mellon
C
Case
2
2: Si
Simulation
l ti Setup
S t
•
Thermal Power Plants
Capacity
•
Ramp Rate
Economic Cost
Environmental Cost
4 PG + ΔPG
Coal
700 MW
25 MW / 0.5h
0.02 PG2 + 5PG + c
Natural Gas
500 MW
100 MW / 5min
0.06 PG2 + 20 PG + c
2 PG + 10ΔPG
Nuclear
450 MW
3 MW / h
0.01PG2 + 2 PG + c
PG
R
Renewable
bl
River Flow
Hydro
y
(4
( plants)
p
)
3000 m3/s
Capacity
Wind
1000 MW
Weir discharge
Economic Cost
0.5PG + c
1200 m3/s
Environmental Cost
Δh 2 + 250δq 2
Non-Usage
0.01 ⋅ (PGref − PG )
2
19
Carnegie Mellon
C
Case
2
2: Si
Simulation
l ti Setup
S t
•
Storage
Capacity
Storage
•
400 MWh
Economic Cost
5Ploss
Load
Maximum
Load
2500 MW
Critical
70%
Quality of Service
100 ⋅ (PLref − PL )
2
20
Carnegie Mellon
Case 2: Simulation Results
21
Carnegie Mellon
C
Case
2
2: Si
Simulation
l ti Results
R
lt
22
Carnegie Mellon
C
Case
2
2: R
Reference
f
23
Carnegie Mellon
C
Case
2
2: R
Reference
f
24
Carnegie Mellon
C
Case
2
2: Si
Simulation
l ti Results
R
lt
25
Carnegie Mellon
C
Case
2
2: Si
Simulation
l ti Results
R
lt
26
Carnegie Mellon
C
Conclusions
l i
•
Electric power systems field is currently in a major transition
⇒ Major challenges need to be resolved
•
Predictive control allows for full exploitation of device potentials
•
Storage reduces need for fast-ramping
fast ramping backup generation and required
ramp rates if optimally controlled
•
Existing hydro power provides storage capacity
•
Coordination achieves overall optimal performance
•
Integration of intermittent renewable generation asks for hybrid solution
27
Carnegie Mellon
28
Carnegie Mellon
C
Conventional
ti
l Hydro
H d Power
P
100%
90%
80%
70%
60%
50%
40%
30%
20%
10%
0%
29
Carnegie Mellon
M d li
Modeling:
Hydro
H d Power
P
•
Linear Model
– River Flow
– Discharge to Electric Power
30
Carnegie Mellon
C
Case
1
1: Si
Simulation
l ti Results
R
lt
•
Smoothness of Total Power Output
31
Carnegie Mellon
C
Case
2
2: G
Generation/Storage
ti /St
Dispatch
Di
t h
•
Objective Function
•
Constraints
–
–
–
–
Ramp rates
R
t
Capacities
Water level limits
G
Generation
ti = Demand
D
d
32