slide 1 the renewables challenge: keeping the lights on while managing variability and uncertainty...
TRANSCRIPT
Slide 1
The Renewables Challenge:Keeping the lights on while managing
variability and uncertainty
Princeton University Academic Mini-Reunion
October 2, 2015
Warren B. PowellPENSA Laboratory
Dept. of Operations Research andFinancial Engineering
http://energysystems.princeton.edu
Outline
The renewables challenge Combining wind and solar The uncertainties of energy Three energy problems
» An energy storage problem» An energy portfolio policy model» Simulating the PJM grid using SMART-ISO
Concluding remarks
Outline
The renewables challenge Combining wind and solar The uncertainties of energy Three energy problems
» An energy storage problem» An energy portfolio policy model» Simulating the PJM grid using SMART-ISO
Concluding remarks
Wind in the U.S.
99.9 percent from renewables!
Fossil Backup
BatteryStorage
Wind &Solar
20 GW
750 GWhr battery!
Outline
The renewables challenge Combining wind and solar The uncertainties of energy Modeling sequential decision problems under
uncertainty Three energy problems
» An energy storage problem» An energy portfolio policy model» Simulating the PJM grid using SMART-ISO
Concluding remarks
Wind energy in PJM
Total PJM load plus actual wind (July)
53 wind farms
Wind energy in PJM
Total PJM load plus actual wind (July)
100GW
101,000 MWhr battery$50 billion!!
Wind ~ 37 percent of total load
Solar energy
Solar from all PSE&G solar farms
Solar energy
Total PJM load plus factored solar (July)
Solar ~ 15 percent of total load
Combining wind and solar
Mixture of wind and solar to meet July load
815,000 MWhr battery $989 billion!!
Combining wind and solar
Mixture of wind and solar to meet July load
100GW
260,000 MWhr battery $130 billion!!
Capacity factor analysis
Computing the “capacity factor”
Capacity
Actual
Generated windCapacity factor = 39%
Maximum capacity
Capacity factor analysis
Classical analysis of renewables» Multiply installed capacity by the capacity factor
• 1 MW solar panel• Capacity factor of .25• Translates to 0.25 MW of generation
» Now, treat the .25 MW as if it is a form of conventional generation.
» This makes it possible to scale up renewables without regard to the challenges of variability and uncertainty.
» Let’s try to do better.
Outline
The renewables challenge Combining wind and solar The uncertainties of energy Three energy problems
» An energy storage problem» An energy portfolio policy model» Simulating the PJM grid using SMART-ISO
Concluding remarks
Northeast Reliability Councils and Interconnects
Energy from wind
1 year
Wind power from all PJM wind farms
Jan Feb March April May June July Aug Sept Oct Nov Dec
Energy from wind
30 days
Wind from all PJM wind farms
Modeling wind
Forecast vs. actual for a single wind farm
Actual
Forecasted
Solar energy
Princeton solar array
Solar energy
Princeton solar array
PSE&G solar farms
Solar output over entire year (all farms)
Sept Oct Nov Dec Jan Feb March April May June July Aug
Solar from PSE&G solar farms
Solar from a single solar farm
Solar from PSE&G solar farms
Within-day sample trajectories
Brazil
Rainfall
Foz do Iguaçu (Brazil) – 2011 through 2013
2011 2012 2013
Commodity prices
The price of natural gas» Reflects global and local economies, competing global
commodities (primarily oil), policies (e.g. toward CO2), and technology (e.g. fracking).
$4 /mmBTU
$120 /mmBTU!
Locational marginal prices on the gridLMPs – Locational marginal prices
$58.47/MW
Locational marginal prices on the gridLMPs – Locational marginal prices
$977/MW !!!
Locational marginal prices on the gridLMPs – Locational marginal prices
$328/MW !
Locational marginal prices on the grid
$52/MWhr
Uncertainty
It is important to separate:» Predictable variability
• Diurnal cycles• Large weather patterns• Major human events (Super
bowl)
» Stochastic uncertainty • Temperature deviations from
forecast• Late/early arrival of a storm• Generator failures• Wind shifts
PJM load
Aggregate solar
Dealing with uncertainty
Available at energysystems.princeton.edu
Outline
The renewables challenge Combining wind and solar The uncertainties of energy Three energy problems
» An energy storage problem» An energy portfolio policy model» Simulating the PJM grid using SMART-ISO
Concluding remarks
Policy function approximations
Battery arbitrage – When to charge, when to discharge, given volatile LMPs
1 4 7 10 13 16 19 22 25 28 31 34 37 40 43 46 49 52 55 58 61 64 67 700.00
20.00
40.00
60.00
80.00
100.00
120.00
140.00
Grid operators require that batteries bid charge and discharge prices, an hour in advance.
We have to search for the best values for the policy parameters
DischargeCharge
Charge Discharge and .
Policy function approximations
Policy function approximations
Our policy function might be the parametric model (this is nonlinear in the parameters):
charge
charge discharge
charge
1 if
( | ) 0 if
1 if
t
t t
t
p
X S p
p
Energy in storage:
Price of electricity:
Policy function approximations
Finding the best policy» We need to maximize
» We cannot compute the expectation, so we run simulations:
DischargeCharge
0
max ( ) , ( | )T
tt t t
t
F C S X S
E
Outline
The renewables challenge Combining wind and solar The uncertainties of energy Three energy problems
» An energy storage problem» An energy portfolio policy model» Simulating the PJM grid using SMART-ISO
Concluding remarks
99.9 percent from renewables!
» What answer do we get if we model this problem more carefully?
SMART-Invest
Features:» Finds the optimal mix of wind, solar and storage, in the
presence of two types of fossil generation:• Slow (steam) generation, which is planned 24 hours in
advance• Fast (turbine) generation, which is planned 1 hour in advance• Real-time ramping of all fossils within ramping limits
» Simulates entire year in hourly increments, to capture all forms of variability (except subhourly)
» Minimizes investment and operating costs, possibly including SRECs and carbon tax.
» Able to directly specify the cost of fossil generation (anticipating dramatic reduction in fossils).
» Properly accounts for the marginal cost of each unit of investment.
8760
,1
min ( ) ( , ( | ))Invi
inv inv opr opr invt t t tx i I
t
C x C S X S x
SMART-Invest
The investment problem:
Investment cost in wind, solarand storage.
Capital investment cost inwind, solar and storage
( )oprtX S is the operating policy.
Operating costs of fossil generators,energy losses from storage, misc. operating costs of renewables.
Wind
Solar
SMART-Invest
Stage 1:Investment
Hour 1 Hour 2 Hour 3 Hr 8758 Hr 8759 Hr 8760…
Wind CapacitySolar Capacity
Battery CapacityFossil Capacity
Wind
Solar
Find search direction
Update investments
SMART-Invest
Operational planning
» Meet demand while minimizing operating costs» Observe day-ahead notification requirements for
generators» Includes reserve constraints to manage uncertainty» Meet aggregate ramping constraints (but does not
schedule individual generators)
24 hour notification of steam
1 hour notification of gas
Real-time storage and ramping decisions (in hourly increments)
36 hour planning horizon using forecasts of wind and solar
SMART-Invest
Operational planning model
» Model plans using rolling 36 hour horizon» Steam plants are locked in 24 hours in advance» Gas turbines are decided 1 hour in advance
Slow fossil running units1 24 36
1 24 36Slow fossil running units
1 24 36Slow fossil running units
1 24 36Slow fossil running units
1 24 36Slow fossil running units
1 24 36Slow fossil running units
1 24 36Slow fossil running units
1 24 36Slow fossil running units
…
…
…1 2 3 4 5 8760
1 year
Lock in steam generation decisions 24 hours in advanceThe tentative plan is discarded
SMART-Invest
Lookahead model with adjustment» Objective function
» Reserve constraint:
» Other constraints:• Ramping• Capacity constraints• Conservation of flow in storage• ….
Tunable policy parameter
Robust policies
Policy search – Optimizing reserve parameter
Low carbon tax, increased usage of slow fossil, requires higher reserve margin ~19 percent
High carbon tax, shift from slow to fast fossil, requires minimal reserve margin ~1 pct
The value of storage
The marginal value of storage» On the margin, value of storage can be expensive!
Ene
rgy
in s
tora
ge
Time
This investment in batteries is only used a small fraction of the time.
Policy studies
Renewables as a function of cost of fossil fuels
Solar
Wind
$/MWhr cost of fossil fuels
100
80
60
40
20
0Perc
ent f
rom
ren
ewab
le Total renewables
0 20 50 60 70 80 90 100 150 300 400 500 1000 3000
» Study assumes unconstrained access to wind at lowest cost (this is not available in the eastern U.S.)
» Note the reluctance to introduce solar…
» … and even greater reluctance to use storage.
Battery
Policy studies
Sensitivity to CO2 tax.
Slow fossil Fast fossil
Nuclear
Wind
Solar
“Other” fast fossil
Outline
The renewables challenge Combining wind and solar The uncertainties of energy Three energy problems
» An energy storage problem» An energy portfolio policy model» Simulating the PJM grid using SMART-ISO
Concluding remarks
© 2010 Warren B. Powell Slide 59
Lecture outline
Simulating the PJM grid with SMART-ISO
The PJM planning process Model validation Modeling wind Integrating offshore wind
The timing of decisions
Day-ahead planning (slow – predominantly steam)
Intermediate-term planning (fast – gas turbines)
Real-time planning (economic dispatch)
The timing of decisions
The day-ahead unit commitment problemMidnight
Noon
Midnight Midnight Midnight
Noon Noon Noon
The timing of decisions
Intermediate-term unit commitment problem
1:15 pm 1:45 pm
1:30
2:15 pm
1:00 pm 2:00 pm 3:00 pm
2:30 pm
The timing of decisions
Intermediate-term unit commitment problem
1:15 pm 1:45 pm
1:30
2:15 pm
1:00 pm 2:00 pm 3:00 pm
2:30 pm
The timing of decisions
Intermediate-term unit commitment problem
Turbine 3
Turbine 22
Turbine 1
Notification time
1:15 pm
1:30
2:00 pm 3:00 pm
Ramping, but no on/off decisions.
Commitment
decisions
The timing of decisions
Real-time economic dispatch problem
1pm 2pm
1:05 1:10 1:15 1:20 1:25 1:30
The timing of decisions
Real-time economic dispatch problem
1pm 2pm
1:05 1:10 1:15 1:20 1:25 1:30
Run economic dispatch to perform 5 minute ramping
Run AC power flow model
The timing of decisions
Real-time economic dispatch problem
1pm 2pm
1:05 1:10 1:15 1:20 1:25 1:30
Run economic dispatch to perform 5 minute ramping
Run AC power flow model
The timing of decisions
Real-time economic dispatch problem
1pm 2pm
1:05 1:10 1:15 1:20 1:25 1:30
Run economic dispatch to perform 5 minute ramping
Run AC power flow model
© 2010 Warren B. Powell Slide 70
Lecture outline
Simulating the PJM grid with SMART-ISO
The PJM planning process Model validation Modeling wind Integrating offshore wind
SMART-ISO: Calibration
Historical generation mix during 22-28 Jul 2010
Nuclear
Steam
Comb. cycle+gasPumped hydro
SMART-ISO: Calibration
Simulated generation mix during 22-28 Jul 2010
Steam
Nuclear
Comb. cycle+gasPumped hydro
Actual vs. simulated LMPs
January April
July October
© 2010 Warren B. Powell Slide 74
Lecture outline
Simulating the PJM grid with SMART-ISO
The PJM planning process Model validation Modeling wind Integrating offshore wind
Energy from wind
30 days
Wind power from all PJM wind farms
Energy from wind
Illustration of forecasted wind power and actual» The forecast (black line) is deterministic (at time t, when the forecast
was made). The actuals are stochastic.
This is our forecast of the wind power at time t’, made at time t.
'ttf
This is the actual energy from wind, showingthe deviations from forecast.
Energy from wind
Two types of uncertainty arise in forecasting:» At time t, we have forecasts for different times into the
future:
• The forecast is an imperfect estimate of the actual load at time t’:
» As new information arrives, the forecasts themselves change from time t to t+1:
'ttf
1, ' , ' 1, 'f
t t t t t tf f This change in the forecast is “stochastic” at time t.
' , ' 'L
t t t tL f The actual load at time t’ is “stochastic” at time t.
Forecasting wind
Rolling 24-hour forecast of PJM wind farms
Actual
Hour of day1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24
Meg
awat
ts
Onshore & offshore wind farms
We were given access to data on the wind power generated by onshore wind farms within PJM
Proposal: Use onshore data to calibrate a stochastic model of forecasting errors. Then use this model to create a simulated “actual” for offshore.
Distribution of forecast errors» Uses adjusted spatial correlations to improve fit.
Simulating onshore wind
Observed
Simulated
Error in forecasted wind speed
Simulating onshore wind Cumulative histogram of the # of consecutive time intervals the
observed/simulated time series is above the forecasted one (chosen farm only):
Time actual is above forecast
ObservedSimulated
Simulating onshore wind Cumulative histogram of the # of consecutive time intervals the
observed/simulated time series is below the forecasted one (chosen farm only):
Time actual is below forecast
ObservedSimulated
Wind forecast samples80
70
60
50
40
30
20
10
0
Wind forecast samples80
70
60
50
40
30
20
10
0
Wind forecast samples80
70
60
50
40
30
20
10
0
Wind forecast samples80
70
60
50
40
30
20
10
0
Simulating offshore wind
Offshore wind – Buildout level 5
Forecasted wind
© 2010 Warren B. Powell Slide 88
Lecture outline
Simulating the PJM grid with SMART-ISO
The PJM planning process Model validation Modeling wind Integrating offshore wind
SMART-ISO: Offshore wind study
Mid-Atlantic Offshore Wind Integration and Transmission Study (U. Delaware & partners, funded by DOE)
29 offshore sub-blocks in 5 build-out scenarios:» 1: 8 GW» 2: 28 GW» 3: 40 GW» 4: 55 GW» 5: 78 GW
Modeling wind
» Steadier than onshore? Where???
GW
Modeling wind
The power from wind:
» The cubic relationship means small changes in speed translate to large changes in power.
31
2P B Av
Wind speed (in m/sec)v
3Area of rotor blades in mA
3Density of air ( 1.225kg/m )
Power coefficient
fraction of wind converted to mechanical energy
.593 (the Betz limit)
B
Unit commitment under uncertainty
Actual wind
Hour ahead
forecast
How forecasting uncertainty causes outages.
SMART-ISO: Offshore wind study
SMART-ISO: Offshore wind study
Less steamUniform increase in gas
SMART-ISO: Offshore wind study
Outage probabilities over 21 scenarios for January, April and October:
Per
cent
of
sam
ples
ther
e is
an
outa
ge
Bas
e P
JM r
eser
veO
ptim
ized
res
erve
sP
erfe
ct in
form
atio
n
SMART-ISO: Offshore wind study
Outage probabilities over 21 scenarios for July
Per
cent
of
sam
ples
ther
e is
an
outa
ge
SMART-ISO: Offshore wind study
Ramping reserves, July, 2010
Perfect forecast
Imperfect forecast
16 14 12 10 8
6
4
2
0
1 2 3 4 5
Buildout levels
Ram
ping
res
erve
s (G
W)
SMART-ISO: Offshore wind study
Observations» The requirement of reserves at 20 percent of capacity is only for
July, and we believe this over-estimates the reserves needed.» Other months require reserves at 10 percent of capacity, which is
still substantial, considering that renewables generate energy at roughly 20 percent of capacity.• 10 percent reserves means that 100 MW of wind generation
requires 10 MW of spinning reserve.• 100 MW of generating capacity translates to around 25 MW of
power (on average). So 25 MW of power requires 10 MW of spinning reserve from a fossil plant.
» This is for an “as is” network – we are using existing generation technologies and existing planning procedures.
» A richer portfolio including demand response, battery storage and more sophisticated planning should help.
Outline
The renewables challenge Combining wind and solar The uncertainties of energy Three energy problems
» An energy storage problem» An energy portfolio policy model» Simulating the PJM grid using SMART-ISO
Concluding remarks
The challenge of renewables
What did we learn?» Wind and solar are variable, and uncertain, with very
different characteristics.» Back-of-the-envelope analysis (e.g. capacity factor
analysis) completely ignores the challenges of dealing with variability and uncertainty.
» It is important to design effective policies for dealing with variability, forecasts, and uncertainty.
» Reserves to handle uncertainty can be significant, and are easily overlooked.
» Renewables are a powerful alternative to reduce our CO2 footprint, but they need to be planned properly to avoid unexpected costs.