Solar Energy, Battery Systems, and Grid Investments

During the past 10 years, solar systems, large and small, have been installed in many areas of the world, especially in southern Germany and southern California. With many small solar systems connected to a distribution system, the passing of clouds causes their output to become highly irregular. When there were few solar systems this was not a problem, but not so with many systems.

Increasingly, battery systems are added to such distribution grids for regulation, i.e., keeping voltage and frequency with prescribed ranges, and, if such battery systems have sufficient storage capacity, for shifting solar energy from midday hours to early evening hours. 

Solar/Battery Systems Combos: If a large capacity solar system, multi-megawatt, is directly connected, via a substation, to a high voltage grid, a multi-megawatt battery system is often required for regulation, before being allowed to connect to the grid.

If the battery system has sufficient storage capacity, solar energy can be charged around midday, at lower electric rates, and discharged during peak evening hours, at higher rates. Besides, regulation and energy shifting, the battery system has other economic advantages, such as reducing a utility’s monthly and annual peak demands and for participating in the real-time markets and forward capacity and reserve markets, as discussed in an ISO-NE report How Energy Storage Can Participate in New England’s Wholesale Electricity Markets and in below Parts 1 and 2.

Solar Systems Do Not Reduce High Voltage Grid Investments: The design of the capacity of a high voltage grid system is similar to the design of a highway system. The peak electric loads (equivalent to traffic during peak hours) are projected, for say 10 or 20 years, and the need for high voltage wires, poles and substations is determined and capital costs are estimated.

Historic electricity demand growth was 2 to 3 percent per year, but that growth has been near zero since about 2005, largely due to energy efficiency improvements, a.k.a., cost cutting. That means annual HV grid investments became significantly less than long-term projections. Some RE proponents claimed the reduction in HV grid investments was due to build outs of distributed solar and wind systems, but that is invalid, as shown under Part 3.

Part 1.

Solar/Battery System Combos: If a grid has many small capacity solar systems, or a multi-megawatt solar system, the minute-to-minute variations of their outputs, such as during variable cloudy conditions, would destabilize the grid, and quick-responding battery systems to perform regulation, a.k.a., firming, would be required.

– Battery systems draw variable AC energy from the grid, convert it to DC, absorb it, discharge steady DC energy, convert it to steady AC energy, and feed that back into the grid. 

– With enough storage capacity, MWh, battery systems can store midday solar energy and discharge it during peak evening hours, about 5 to 9 pm.

– Battery system losses: 7%, charging + 7%, discharging + 3%, DC to AC conversion + 1%, balance of plant* = 18%.

* Electricity for lighting, HVAC, electronics, etc., drawn from a nearby distribution grid.

Throughout the world, many utilities have installed multi-megawatt battery systems, especially in southern Germany and southern California, and solar/battery system combos, during the past 10 years.

An example is the Smart Grid Demonstration Project of the Public Service Company of New Mexico, which consists of a solar system plus battery system for simultaneous voltage smoothing and peak shifting. Please read the very complete report of the New Mexico combo. See URL for Report.

https://www.smartgrid.gov/files/FTR_FINAL_PNM_27May14.pdf

In Sterling, Massachusetts, NEC ES is providing a turnkey GSS® energy storage system which includes a single 53′ container housing 3.9 MWh of lithium ion batteries, a 2 MW power conversion system, and proprietary NEC ES AEROS® controls software suite. NEC ES will provide service and maintenance. The battery system has a turnkey cost of $2.7 million, or $692/kWh. With a $1.4 million grant from the Massachusetts Department of Energy Resources (DOER) and financial grants from the U.S. Department of Energy Office of Electricity (DOE-OE), and other grants, it pays for itself in about 2 years, without grants in about 7 years. The battery systems have a 10-year performance warranty, but are expected to function significantly longer than that.

In Kauai, Hawaii, there is a 52 MWh battery system, combined with a 13 MW field-mounted PV solar system, tied, via a substation, to the HV electric grid. The battery system is used for regulation and energy shifting. The system is designed to supply stored energy from 5 to 10 pm, which reduces the use of peaker plants. Tesla and its subsidiary, Solar City, own the entire installation.

Solar/Battery System Combo in Vermont: Green Mountain Power, a utility in Vermont with 77% of the electricity market, owns the Stafford Hill, 2.5 MW solar plant, 7722 panels, on about 20 acres. GMP added 2 MW (storage 2.4 MWh) of advanced lead acid batteries, 2 MW (storage 1 MWh), of lithium-ion batteries, total storage of 3.4 MWh, and (4) 500-kW inverters. The expected life of the batteries is about 10 – 15 years. The turnkey cost was about 7.155, solar system + 5.345, battery system = $12.5 million.

– $235,000 was a cash donation from the US-DOE

– $50,000 was a cash donation from the VT-DPS, via the Clean Energy Development Fund.

– About 30% of the project turnkey cost is a federal investment tax credit that can be used to offset state and federal taxes due on other GMP operations.

– GMP sells Renewable Energy Credits, RECs, to out of state entities to reduce the generating cost of solar electricity.

– The whole project can be written off, likely in less than 10 years, which also reduces state and federal taxes.

– GMP claims the project pays for itself in about 5 to 6 years.

GMP prefers to own solar/battery system combos, and to sell/lease to ratepayers heat pumps (made in Japan), solar systems (PV panels likely made in China) and Tesla wall-hung batteries (made in Nevada), because that adds to GMP’s asset base on which it is allowed to earn 9% per year, and helps GMP collect subsidies, and tax credits, and have fast write offs, to minimize paying taxes, and increase its net profits.

GMP prefers not utilizing about 77% of the 200 MW reserved for Vermont of the recently approved 1000 MW, HVDC line, because, that does nothing the increase GMP’s asset base.

NOTE: Without the cash grants and other subsidies, this project likely would not pay until after the batteries are used up. With plentiful, domestic, low-cost, natural gas available for at least the next several decades, the wholesale prices during peak hours would not be high enough compared to wholesale prices during midday hours to make energy shifting pay in New England. Lay people overlook the levelized cost of utility-scale solar is at least 10 c/kWh (subsidized) in New England. Without the renewable energy credits, RECs, to reduce that cost to about 5 c/kWh, there would be no price difference to make energy shifting pay. GMP may be hoping, with enough PR about islanding, and micro-grids, and batteries “being the future”, people would not notice the project’s poor economics.

NOTE: At a battery system cost of $5,345,000/3400 kWh = $1,572/kWh (2.27 times the $692/kWh cost of the Sterling, MA, battery system), the wholesale prices during peak hours, usually 5 – 9 pm, would have to be extremely high (which happens on rare occasions) compared with the normal, much lower, wholesale prices during midday hours (usually with high solar production) to make energy shifting pay in New England. In southern California, such large price differences are more frequent, because of time of day pricing. With large numbers of solar systems disturbing the grids, California had to mandate utilities install such expensive battery systems.

Operation During Summer Months: The DC output of the solar system varies from about 2 MW around noontime to about 0.5 MW when a cloud passes over the panels, for a ramp rate of 1.5 MW in 10 seconds, or 9 MW per minute, which can be dealt with by the quick-responding battery system, as it occurs. The variable solar DC energy is charged into the batteries at a levelized cost of, say 10 c/kWh (subsidized), 15 c/kWh (unsubsidized). The batteries are topped of with grid energy at midday wholesale prices of, say 6 c/kWh, which is converted from AC to DC (conversion and charging losses about 10%).

The battery system has (4) 500 kW inverters to convert the DC energy discharged from the batteries to AC. The battery system can discharge AC energy at a steady level of up to 2 MW. However, according to GMP, typically, the discharge is maintained at a steady level of about 1 MW, spread over up to 3.4 hours, during peak hours, about 5 to 9 pm, to ensure catching the peak demand for that period. The steady output, a.k.a., firm output, is required by ISO-NE, i.e., no firm output, no feeding in.

Operation During Winter Months: In NE, the ratio of the maximum summer/minimum winter monthly output is about 4; in Germany, it is about 6. Plus, there are days with minimal output, when panels are covered with snow and ice. Less solar energy will be charged into the batteries during midday and more energy is taken from the grid to top off the batteries.

NOTE: The battery discharge of 3.4 MWh implies the batteries would be completely drained. If done on a daily basis, the batteries likely would not last beyond 10 years.

Part 2.

The ISO-NE grid operator performs the accounting of the grid capacity costs and distribution costs. GMP pays the grid operator for its share of capacity costs (the money paid to power plants to make sure they are available during peak load periods). GMP’s share of the capacity cost is based on its demand during one annual peak hour. GMP’s share of the transmission cost is based on its demand during 12 monthly peak hours. GMP’s capacity payment is about $35 million/y and transmission about $55 million/y, for a total of about $90 million/year.

Reducing ISO-NE Capacity Charges: Capacity Payment = (Capacity Load Obligation) x (Net Regional Clearing Price)

– The CLO is based on the peak contribution value, e.g., the load on the peak day/hour each year identified by ISO-NE. 

– The NRCP = (Total Capacity Payments to Resources)/(Total Capacity Supply Obligation, MW – Self Supply, MW – Excess Real-Time Energy Generation, MW).

GMP’s peak demand is about 770 MW. GMP reduces its peak demand (as seen by ISO-NE), and thus its ISO-NE capacity and transmission charges, by using: 

– 100 MW of diesel-generators and gas turbine-generators, all, or some of which, are often used a few hours a day to reduce GMP’s peak demand. 

– 100 MW of hydro plants for which GMP gets a 43 MW capacity credit from ISO-NE.

GMP used its batteries to reduce the ISO-NE peak demand of 25,466 MW by about 2 MW from 3 to 4 pm, on August 12, 2016, and claims it reduced its annual demand charges by about $200,000. GMP did not provide any calculations of this very important number. While the battery system was discharging, it was being charged by the simultaneous DC energy production of the solar system, which reduced the net discharge from the battery; deep discharges shorten battery life. See graphs on page 4 and 13 of URLs.

Theoretically, GMP could reduce its entire demand, as seen by ISO-NE, to zero, by building 770 solar/battery system combos at a cost of about 770 x 12.5 = $9.625 billion. Any surplus electricity would be exported to the NE grid. Any shortage, daily and seasonal, would be supplied by other sources, such as bio, hydro, wind, and imported from the NE grid.

Open meadow area required: 770 x 20 = 15400 acres. Production 770 x 2.5 x 8766 x 0.14 = 2,362,437 MWh/y, less 18% battery system losses = 2,173,442 MWh/y, about 36.2% of total electricity supply to Vermont utilities.

Reducing ISO-NE Transmission Charges: Regional Network Service (RNS) transmission charge = (Pool RNS Rate) x (Monthly Network Load).

In a similar manner, described above, GMP can reduce its transmission charges during the 12 monthly peak hours. No monthly savings are yet available for Stafford Hill, but they are expected to be similar to those of Sterling, MA, about $17000/month, or about 204,000/y, if a 2 MW load reduction, for one peak hour, in each of 12 months. See Table 2 and See URL

NOTE: The Sandia National Laboratories performed a detailed study of the battery system at Sterling, MA, and determined the ISO-NE capacity charge reduction due to 2 MW would be as shown in the below Table 1, and determined the ISO-NE transmission charge reduction due to 2 MW would be as shown in Table 2.

ISO-NE has a 2015-16 capacity liability of 32,968 MW x 3.129 $/kW-month x 1000 kW/MW x 12 months = $1,237,888,464

The Sterling Municipal Light Department share, no storage = (9.631, Sterling MW)/(24039, ISO-NE peak MW hour) x $1,237,888,464 = $495,946

The SMLD share, with 2 MW feed-in from storage for a maximum of 1.95 hours = 7.631/24039 x $1,237,888,464 = $392,956/y, for an ISO-NE capacity charge reduction of $102,990, close to the Sandia value in the below Table 1.

Table 1; ISO-NE Capacity Charge Reduction

For the ISO-NE transmission charge, the price is from June 1 to May 31, thus for a 2 MW feed-in, the ISO-NE charge reduction = (5 months x $7.27889/kW-month + 7 months x 8.22512/kW-month) x 12 months x 2 MW = $187,941; Sandia used a slightly greater amount. See below Table 2.

Table 2; ISO-NE Transmission Charge Reduction

Cost Reduction by Energy Shifting (arbitrage), Normal Peak Demand Day: The shifting operation reduces GMPs electricity purchases during peak hours, when summer wholesale prices are higher, typically about 8 to 10 c/kWh, and even higher on some days.

Solar production is 2.5 x 8766 x 0.14 – 18%, losses = 2519 MWh/y, of which about 1 MW x 3.4 h x 365 d = 1241 MWh/y is shifted, about 50%.

GMP produces solar electricity at 10 c/kWh (subsidized). GMP can buy electricity at wholesale prices for 6 c/kWh during midday hours and 7 – 8 c/kWh during peak hours, around 5 – 9 pm. GMP reduces the cost of solar electricity by 4 – 5 c/kWh by selling RECs to out-of-state entities. That means that electricity cannot be counted to any Vermont “90% RE by 2050” goals. Vermont’s meadows and open land are used as energy production plantations for Connecticut entities.

On a normal peak demand day, the cost reduction from energy shifting would be 3400 kWh x (8, wholesale price – 10, levelized solar + 5, REC) = $102

Cost Reduction by Energy Shifting (arbitrage), Very High Peak Demand Day: Cost reductions by energy shifting are greatly increased on very hot summer days, when wholesale prices likely would be very high, as happened at 4 pm, on August 12, 2016, according to the ISO-NE, Hourly Wholesale Load Report. These are very rare occurrences.

The feeding from the solar system and from the battery system to the grid has to be apportioned and timed to maximize revenues when prices are highest. Prices were 7.7 c/kWh at 10 am, reached a maximum of 36.8 c/kWh at 4 pm, and were 7 c/kWh at 10 pm.

On a very high peak demand day, the cost reduction from energy shifting would be 3400 kWh x (36.8, wholesale price – 10, levelized solar + 5, REC) = $1081.

Revenues from Selling Solar Electricity: GMP receives revenues from selling the portion of production not used for shifting, about 50% of the annual solar production.

Revenues from Real-Time and Forward Markets: The steady output of the solar/battery system combo enables GMP to participate in auctions and obtain revenues in the Real-Time Capacity Market, and Forward Capacity Market, which ensures there is capacity sufficiency in future years and Real-Time Reserve Market, and Forward Reserve Market, which ensures there is reserve sufficiency in future years. The auctions are managed by ISO-NE.

Another Revenue Option for GMP: GMP has other options to gain revenues by leasing or selling 7 kWh, wall-mounted, Tesla battery units. GMP sells turnkey units for $6500. GMP charges a ratepayer (with a solar system or not) $37/month x 12 month/y x 10 years = $4440 for leasing a unit for a 10-year period.

GMP automatically drains a percentage of the energy, say 7 kWh, from those batteries in late afternoon to reduce its peak demand charges, for arbitrage.

It would take 1000 batteries (turnkey cost $6.5 million) x 7 kWh to provide 7 MWh, or 2.06 MW spread over up to 3.4 hours, which is a drop in the bucket to reduce GMP’s peak demand of about 770 MW.

NOTE:

– Enhanced demand management and enhanced energy efficiency would significantly reduce peak demands at significantly less cost per MW, than solar/battery system combos. However, that likely would add very little, or nothing, to GMP’s asset base.

– From the above it is clear, large solar plants must have battery systems, or other means of firming, such as a diesel-generator set, for safely connecting to the grid. 

– The peak demand, MW, is in the late afternoon, when solar energy output has become much less than at noon, even more so in winter. 

– It would be much less costly for GMP and ratepayers to turn on a quick-starting diesel-generator set for that period, as GMP has been doing for decades. A standard 1 MW, D-G set has a turnkey cost of less than $1.0 million.

Part 3.

Solar Systems Do Not Reduce High Voltage Grid Investments: Below it is shown solar systems do not reduce investments in high voltage systems, because peak demands occur when solar output is minimal, about 5 to 6 pm.

Below is a table with the hourly NE demand on 13 July 2016 from the real-time ISO-NE grid status website. The NE peak demand was 22173 MW at 5 pm. The Vermont demand values were obtained by pro-rating, based on an assumed 1000 MW peak demand. See page 6 of URL.

The NREL pvwatts program was used to obtain the hourly output in July for a 5 kW system in Woodstock, VT, using Concord, NH, weather data. That output was prorated upwards, based on an assumed Vermont installed solar system capacity of 125 MW.

If Vermont’s maximum solar output is assumed to occur on 13 July 2016, the output of the 125 MW would have been 15.8 MW during hour 17 (5 pm), for a demand reduction of 1.58%. If Vermont’s solar system capacity were doubled, that demand reduction percentage would also double.

Conclusions:

– The NE grid capacity is designed to accommodate at least the maximum demand that could occur in the foreseeable future. 

– The small demand reduction percent due to solar output would have minimal impact on grid capacity design and investments.

– Solar output (and wind output) cannot be counted on to reduce grid design capacity, because in New England:

During winter, solar output could be near zero, if panels were covered with snow and ice when peak demand occurs. 

Solar output is near zero, or zero, about 75% of the hours of the year. 

Wind output is near zero, or zero, about 40% of the hours of the year.

Total solar + wind output is near zero during many hours of the year, including peak demand hours, per the real-time, ISO-NE grid status website.

 Photo Credit: Michael Coghlan via Flickr