Early version (not final)

1. •Why is Storage Important
•How Much Storage is Needed
•Why is Storage so Hard
•Possible Futures

2.  National Renewable Energy Laboratories
 Website is http://www.nrel.gov
 Ludington Power Storage information is
directly from Consumers Energy website -
http://www.consumersenergy.com/content.aspx?id=1830
 Energy for Future Presidents –
Richard Muller
 Simple web searchs for the rest

3.  Study about impact of different renewable levels
 Looked at 20% to 90% renewable levels in the
electrical grid
 Study got lots of media attention
 Reporting was mostly superficial
 Study is very impressive and well done
 May be a bit optimistic on two points
It does not model the effect of day to day changes
Loss of base load efficiency may reduce CO2 reduction

4.  Variable generation – covers offshore/onshore
wind, solar photovoltic (PV), and solar thermal
 Overnight capital cost – cost of building a plant if
you could do it in a single day
 Capital Recovery Factor – multiplier against
overnight capital cost to get yearly cost
 Dispatchable – generator runs when requested
 Curtailment – when a generator produces energy
that the grid must refuse

5.  Several reasons
 Geography – grid cannot move the energy to
where it could be used
 Temporal – grid cannot store the energy to use at
a better time
 Quality – voltage or wave pattern are wrong
 Curtailment is a very bad thing
 Economic value not who gets paid

7. • Electrical usage is about 40% of total US
energy consumption
• Only 34% of CO2 emissions
• US uses 2300 trillion watt hours in a year

8.  Ludington is about 1.8
GWatts
 27 Billion gallons
 More than 8 hours of storage
 1000 acres
 9% of US storage
 Opened in 1973
 Largest storage site in the
world when new
 Now the second largest in US,
fifth in the world
 Bath County VA is largest at 3
GWatts

9.  Many countries adding wind and solar PV
 Many states have renewable energy mandates
 Emphasis is on wind, solar, and biomass
 Local production is also given special treatment
 Wind and solar are variable driven by Mother
Nature
 If you don't need the power when it comes, too
bad
 Storage lets you use the power when needed

10.  GE recently announced a battery option for
their 1.6 Megawatt Wind Turbine
 Standard version stores 50 KWHours
 How much is that in hours?

11.  GE recently announced a battery option for
their 1.6 Megawatt Wind Turbine
 Standard version stores 50 KWHours
 How much is that in hours?
 About 2 minutes

12.  Minute by Minute
 Solar PV most sensitive
 Hour by Hour
 Solar PV and Wind both
 Day by Day
 All Solar and Wind
 Seasonal
 Mostly Solar but Wind to a limited extent

13.  The grid distributes power at light speed
 Speed of light in copper not in a vacuum
 Sudden changes in load or in demand cause
variations in voltage
 Tight limits on what is allowed
Minor variations cause local issues
Major variations can cause blackouts
 Today's grid has little need for large scale
storage
 Built to match yesterday's technology

14.  Baseload Power – Plants run 24/7/365
 Nukes, most hydro and coal plants
 Load Following – Many plants run some part
of most days
 Small coal, some hydro, gas combined cycle,
biomass
 Peakers – Plants run during the biggest loads
 Open cycle gas turbines
 Even some oil-fired plants for the hottest days

15.  Load on the grid changes with day and time
 Biggest change is day to night
Night load is 50 to 60 percent of weekday load
 Next is workday to weekend day
Weekend day is about 75 percent of workday
 Last is normal day to hot day
Depends strongly on region – Not sure how much

18.  Another problem is sudden changes
 From NREL, ERCOT load with simulated wind

19.  Four ways to handle variable generation
 Over-engineer and accept curtailment at times
 Load management to reduce peaks and valleys
 Move electricity between regions
 Store energy at valleys and release at peaks

21.  5.5% curtailment at 80% renewable load
 100 terawatt hours lost per year
 At 5 cents per kilowatt hour, $5 Billion dollars

22.  5.5% curtailment at 80% renewable load
 100 terawatt hours lost per year
 At 5 cents per kilowatt hour, $5 Billion dollars
 100% increase in load management

23.  5.5% curtailment at 80% renewable load
 100 terawatt hours lost per year
 At 5 cents per kilowatt hour, $5 Billion dollars
 100% increase in load management
 100% increase in high tension lines

24.  5.5% curtailment at 80% renewable load
 100 terawatt hours lost per year
 At 5 cents per kilowatt hour, $5 Billion dollars
 100% increase in load management
 100% increase in high tension lines
 400% to 650% increase in storage
 From 20 GWatts to 100-150 Gwatts
 Store/release 200 to 300 terawatt hours per year

26. As a result of the modeling assumptions, most of the new storage is CAES;
however, the tradeoff between CAES and PSH is largely due to the data
limitations in the ReEDS model, where the vast majority of potential PSH in
much of the United States was not evaluated. In addition, the relative risk
associated with CAES versus PSH was not considered. PSH is a proven tech-
nology, while CAES has yet to be deployed in either bedded salt or in porous
rock formations, which represents a large fraction of assumed deployments.
The limited deployment of batteries estimated by ReEDS is due to their high
cost and assumed minimal projected cost reduction over time as well as to a
lack of full valuation of their benefits to bulk power and distribution systems.
Overall, ReEDS results demonstrate an obvious discrepancy with relative histor-
ical and proposed deployment of these technologies, where PSH dominates.
The analysis of energy storage technologies for RE Futures demonstrates the
need for more comprehensive estimates of the cost and resource availability for
both CAES and PSH, as well as a more complete assessment of the various
benefits of energy storage as renewable energy penetrations increase.

27.  Storage needs lots of usage to be practical
 Outside edges of the curve are rare
 But the right side of the curve has more power
than it might appear

28.  Assuming a fixed life and KWh dominates
 (OCC*CRF + FOM)/(8.760*UR) + CEI/EFF
 OCC – Overnight Capital Cost per KWh
 CRF – Capital Recovery Factor
 FOM – Fixed Operations and Maintenance
 8.760 – Hours in a year/100
 CEI – Cost of energy coming in
 EFF – Efficiency of the storage
 UR – Utilization Rate in Percent

29.  (OCC*CRF + FOM)/(8.760*UR) + CEI/EFF
 Ludington is about $100 per KWh
 CRF is based on useful years and interest rate -
.054 for 50 years, .36 for 3 years
 FOM is usually between 1 and 10 percent of
OCC – pumped hydro is the lowest
 (100 * .054 + 1)/(8.76 * .25) = 2.9 ¢/KWh
 (4.0 / .85) = 4.7 ¢/KWh
 Call it 7.6 cents per Kwh

30.  Pumped hydro runs around $100 per KWh but
$2000 per KW
 For every 6 KWHours, we need $2000 of KW
capacity
 2000/6 is 333
 (333 * .054 + 1)/(8.76 * .25) = 8.7 ¢/KWh
 (4.0 / .85) = 4.7 ¢/KWh
 Call it 13.5 cents per Kwh

31.  Ludington is already cost effective
 Even before Wind and Solar were common
Ludington was built
 Getting an $800 Million upgrade to be more
valuable in the age of Variable Generation

32.  Ludington is marginal, nearly not worth it
 Pumped hydro has scaling issues
 No such thing as a small pumped hydro station
 Years to build
 Limited locations
Many locations have environmental issues
 They don't store as much as it sounds
Ludington stores 3 percent of what Michigan uses
 All types of storage have issues

33.  (OCC*CRF + FOM)/(8.760*UR) + CEI/EFF
 (333 * .054 + 1)/(8.76 * .25) = 8.7 ¢/KWh
 (17.94 + 1 )/(8.76 * .25)
 (4.0 / .85) = 4.7 ¢/KWh
 First component most important
 Need to keep OCC and CRF down
 Nice to keep FOM down as well
 Need to keep UR up
 Second component less critical
 Curtailment implies a very low value for CEI

34.  Time of day shifting – output runs 6 hours a day
 6/24 is .25
 Day to day shifting – output runs 10 hours but only
every third day
 10/24 * .333 is about .14
 Seasonal shifting – output runs 10 hours a day but
only 25 percent of the year
 10/24 * .25 is about .10 but not full power
 Closer to .05 or .06

35.  Hardest area to estimate
 NREL report suggests 100 to 150 GWHours
 I think they are optimistic
 Their daily wind and solar numbers too steady
 I think we will need twice as much
 100 to 150 GWHours for time shifting
 100 to 150 GWHours for day to day smoothing
Fossil fuel usage will be a bit higher without this
 Today we only have about 20 GWHours

36.  Energy density
 Power density
 Environmental impact
 Scale issues
 Use of rare/sensitive
materials
 Efficiency
 Responsiveness
 Energy loss over time
 Capital cost
 Operating and
maintenance cost
 Useful lifetime
 Energy Stored Over
Investment - ESOI

37.  Capital Cost
 Very important – running $333 for grid storage
 Operating and Maintenance Cost
 Also important – pumped hydro is very low
 Useful Lifetime
 Long life time is critical – pumped hydro is 30+ years
 Environmental Impact
 Needs to be low as possible – Pumped hydro is bad

38.  Energy Density
 Not very important – Pumped hydro is really bad
 Power Density
 Somewhat important – Pumped hydro is fair here
 Scale Issues
 Important to get off the ground
 Efficiency
 Important but less than you might think

39.  Use of rare/sensitive materials
 May be critical as grid storage is huge
 Responsiveness
 Pumped hydro is 2 or 3 minutes so not a big issue
 Energy loss over time
 Low levels are ok but high levels are not
 Energy Stored Over Investment (ESOI)
 Need to beat lead-acid batteries at 3- but don't
need to beat pumped hydro at 100+

40.  Potential energy has low energy density
 1 foot-pound is 0.000376616097 watt hours
 1 gallon 300 feet up is .9 watt hours
 Chemical bonds are higher energy density
 Battery powered cars go up hills lifting
The batteries
The car
Passengers
 Fuels like gasoline are even higher
6 gallons will push many cars further than a Tesla S

42.  Stores electrical energy directly with minimal
conversion cost
 Very fast charging and energy release
 Does not store much power
 Does not scale well
 Leaks power over time
 Good choice for second by second grid
stability

43.  Graphene based supercapacitor announced
 Very early days
 Improved energy density – 60 KWhrs per liter
 Very good for a supercapacitor
 Very low for a battery
 Low leakage
 No idea about price yet

44.  Store energy in a large rotating high density
wheel
 Very fast response
 Does not scale very well
 Leaks energy but slowly
 Good for both S/S and M/M grid stability
storage
 In use as grid storage in some locations

45.  Store energy as heat – liquid salt
 Good energy density
 Leak rate depends on quality of
insulation
 Day to Day is possible
 Week long is hard but doable
 Longer is probably out of reach
 Integrates especially well with Concentrating
Solar Power (CSP)

46.  Recovery of Power takes time
 Use molten salt to boil water and spin turbines
 Needs some amount of water
 Does not scale down well
 Square-Cube impact on doubling size
 Looks good for day level and week level
storage
 Cost structure is unclear

47.  Store energy by lifting something up
 Comes in several forms
 Costs are low but so is energy density

48.  Ludington is a perfect example
 Big plant requiring lots of area and volume
 Needs lots of elevation near large body of water
 Takes about one minute to spin up
 Subhydro is pumped hydro upside down
 Large storage tanks at the bottom of the ocean
Really big tanks mostly made out of concrete
 All moving parts underwater
Maintenance costs could be a killer

49.  Slate (www.slate.com) had a posting about
using rail tracks
 Sounds good where pumped hydro is out
 But find a train that weighs as much as Ludington
 Find a location where rock comes to the
surface
 Cut away a big cube and build motors underneath
 Lift it up to store energy and lower to retrieve

50.  Abbreviates to CAES
 Many designs out there
 Because temperature, pressure, volume are
related CAES is an engineering challenge
 Some grid storage already
 Large scale usually means pressurizing a cavern
 Finding good locations is hard
 Only been done a few times, cost is unclear

51.  Store into high pressure cylinders
 Split the heat away and store separately
 One example is Lightsail
Energy density is 13 gallons per KWh
Better than pumped hydro, worse than most batteries
 Store into tanks underwater
 Hydrostor is trying this out near Toronto
 Not clear what the advantage would be

52.  There is more than one battery chemistry
 Batteries have been researched for years
 No silver bullets but some possibilities
 Good points
 Good energy density
 Mostly good power density and responsiveness
 Good efficiency
 Pretty scalable

53.  Bad points
 Some types use rare or dangerous metals
 Most don't last many cycles
 Some need high temperatures

54.  Well researched
 Lead can be dangerous
 Poor and uncertain cycle lifetime
 Bad ESOI – 2 to 3

55.  High quality lithium ores are rare
 Cost is pretty high
 Fire risk not entirely understood
 Energy density is pretty good
 Power density is worse than lead acid but
pretty good

56.  High operating temperature
 Good energy density and power density
 Common materials
 No long term risk to ecology
 Expensive but might improve with scale
 Current models have fire hazard

57.  EOS Energy Storage is a new player
 Impressive but unproven claims
 Cost - $1000/KW and $160/KWh
 30 year lifetime with one full cycle per day
 High energy density
 Fast response
 At least fair power density
 Main materials are cheap
and non-toxic

58.  Storing Excess Variable Generation as Fuel
 Not biomass!
 Direct conversion of wind, solar PV or solar
thermal into fuel
 Done in the laboratory
 Nothing scales today

59.  Electrolysis is well understood
 Currently requires rare materials
 Hydrogen gets lots of press
 Poor energy density when a gas even pressurized
 Liquified hydrogen is dangerous and inefficient
 Many metals turn brittle with long exposure
 Hydrogen can be mixed with methane
 Works well with our current infrastructure
 May be limited to 10-20%

60.  Starts with Methane – CH4
 Works well with current infrastructure
 Not the best energy density
 Ethane – C2H6
 More changes needed in infrastructure
 Energy density improves
 Propane – C3H8
 Changes needed same as ethane
 Energy density still improving

61.  Methanol – CH3OH
 Good energy density
 Moderately toxic but by-products are not an issue
 Ethanol - CH5OH
 Better energy density
 Not toxic in low concentrations
 Propanol CH7OH
 Don't know much about this

62.  Ammonia – NH4
 Seriously toxic in moderate to high concentrations
 Good energy density
 Important precursor in agricultural usage

63.  Old Model
 Storage lets utilities have a higher percentage of
base load
 Storage needed to be efficient since the input had
marginal value
 Few surprises so cheap KWHours was more
important than cheap KW

64.  New Model
 Storage lets utilities avoid curtailment
 Less important to be efficient because input
would be wasted without storage
 Sudden increases and decreases in Variable
Generation are common
 Cheap KW more important than cheap KWHours

65.  We don't see any big advances in grid level
storage
 Pumped hydro and underground CAES are the
only workable solutions
 Both are expensive and take years to build
 Hard to see how NREL 80% can happen in their
timeframe
Getting 400% to 650% increase in storage is necessary
but not sufficient in their vision
Need to build multiple plants at the same time

66.  At least one technology improves grid storage
economics substantially
 Supercapacitor, zinc air, CAES, or something else
 NREL 80% target much easier to make
 Still need to redo the electrical grid
 Reduces US CO2 emissions about 29%