Have you ever had a surplus solar generation problem? Have you ever wished for a flatter load curve? Do you sometimes feel like your job performance relies a little too much on the vagaries of the real-time wholesale electricity market? Do program managers avoid eye contact with you? If you answered yes, then you are probably an electricity procurement manager.

Do you think globally while acting locally? Do you get evangelical on the subject of electric vehicle chargers? Do you know how to get a sweet deal on a heat pump? Do your procurement colleagues call you ‘the hippie’? In all likelihood, you are a customer program manager.

It turns out that procurement managers and program managers at CCAs can solve some of the most vexing challenges facing our electric grid as we push toward our renewable energy future – if they work together. It all has to do with Distributed Energy Resources (DERs).

DER Economic Benefit

Over the last eighteen months, TerraVerde Energy has teamed up with the Center for Climate Protection, Lancaster Choice Energy, Peninsula Clean Energy, and the National Renewable Energy Lab in a Solar Energy Innovation Network (SEIN) grant project to develop an approach to designing tariffs and rates that incentivize the deployment of DERs.

At the heart of our approach to DER rate design, we analyzed the economic benefit that procurement managers can unlock by operating a coordinated aggregation of DER resources to provide grid benefit. If we can understand the economic benefit of operating a DER aggregation, then we can assess the value available for sharing with customers who participate in customer programs for DER deployment.

Although the techniques demonstrated in this post can be applied to any DER type for any Load Serving Entity (LSE), our specific analysis focused on understanding the economic benefit that operating a Virtual Power Plant (VPP) composed of centrally-controlled, customer-sited solar + storage resources can provide to a California CCA.

What’s a Virtual Power Plant?

A VPP is a collection of customer-sited storage resources that can be centrally dispatched by an operator to provide grid benefit. Much like a real power plant, a VPP can participate in California’s wholesale electricity markets. In California, a VPP operator can:

• Dispatch power as an alternative to buying in CAISO’s day-ahead market.
• Dispatch clean energy as an alternative to buying dirtier energy in wholesale markets.
• Store surplus solar generation when wholesale prices are low.
• Sell power into CAISO’s real-time market when prices are high.
• Reduce peak demand to lower Resource Adequacy (RA) requirements.
• Participate in utility demand reduction programs.
• Sell into CAISO’s ancillary service markets.

DER Valuation Model

As a first step toward analyzing DER valuation, we acquired a year’s worth of hour-by-hour procurement cost data from our CCA partners. This data includes:

• Hourly scheduled load.
• Hourly actual load.
• CAISO day-ahead prices.
• CAISO real-time prices.
• Volume and pricing commitments in any bi-lateral power purchase agreements.
• Resource adequacy costs.
• Service fees.

With this data, we can calculate the CCA’s total procurement cost for the year.

Next, we simulate the operation of a VPP according to some strategy. The strategy we chose is one that maximally flattens the CCA’s load curve by drawing power during times of minimum load and discharging during times of maximum load. You can find out more about our load flattening algorithm, which will become publicly available as part of this grant project, at FlattenMyLoad.com.

In our simulation, we model VPP operation by adjusting the simulation’s hourly scheduled load and actual load to mirror the effect of the VPP charging and discharging.  We then analyze any value streams that we can derive from operating the VPP. And finally we add the new value streams to the the CCA’s total procurement cost and compare to the total cost without VPP.

We can use this strategy to analyze any number of different value streams under a VPP operation strategy. Some of these value streams stack, which means they can be pursued simultaneously, while some value stream operate to the exclusion of others. For example, a CCA can simultaneously find value from both selling into real-time markets and reducing resource adequacy requirements by shaving peak demand. However, a CCA can’t sell the same power into both the real-time and ancillary service markets.

Value Stream: Day-Ahead Peak Avoidance

The first value stream that we analyzed is the avoidance of scheduled, peak-period energy purchases.

Each day, a CCA’s scheduling coordinator must report the following day’s expected hour-by-hour energy needs to CAISO. Any portion of this ‘scheduled load’ that is not already covered in bilateral power purchase agreements will be purchased by CAISO on behalf of the CCA in CAISO’s day-ahead market. The usual practice among scheduling coordinators is to provide their very best estimate of the following day’s load, because any difference between scheduled and actual load will have to be purchased or sold on the CCA’s behalf in CAISO’s more volatile and unpredictable real-time market.

In our day-ahead analysis, we gave the scheduling coordinator a few extra tasks. First, the coordinator determines the CCA’s best estimate for the next day’s hourly load curve. Then, the scheduling coordinator runs the scheduled load estimate through the load flattening algorithm to determine when the VPP should charge and discharge to maximally flatten load. (It’s fun! Give it a try at FlattenMyLoad.com) The scheduling coordinator then increases the load schedule reported to CAISO during the hours when the VPP is charging, and it decreases the load schedule during the hours when the VPP discharges.

Figure 1: Flattening of a CCA’s actual load curve on a critical peak day, achieved by applying FlattenMyLoad to the CCA’s scheduled load curve.

The net effect is that the CCA increases day-ahead energy purchases during hours when cheap solar power is available and it decreases day-ahead energy purchases during hours of scarcity when power is both expensive and relatively dirty.

In Figure 2, we’ve modeled this day-ahead peak price avoidance strategy for 200 VPPs of different sizes. The horizontal axis shows VPP size, as a percentage of the CCA’s peak annual load. And the vertical axis shows cost savings per kW of 4-hour battery storage.

Figure 2: Annual savings per kW of 4 hour VPP due to avoidance of peak-period purchases on the day-ahead market.

We see that initially, the value of storage for this value stream increases, and then begins to decrease. We expect this behavior. With a very small VPP we aren’t fully utilizing the best hours of the day for charging and discharging, so as we increase VPP size we use more and more of those most valuable hours. Once our VPP is large enough to exceed those most valuable hours of the day, the VPPs capacity must be applied to less valuable hours for charging and discharging, which leads to a declining marginal value. This can help guide us toward optimal VPP size. The CCA should stop acquiring VPP when the additional value of acquiring more VPP exceeds the additional cost of acquiring more VPP.

For the CCA that we studied, the day-ahead peak price avoidance value stream yields about $30 per kw ($30,000 per MW) per year of avoided cost in the day ahead market. This may not be quite enough value to yield a positive return on a VPP investment, but happily there are other value streams to consider!

Value Stream: Selling into the Real-Time Market

Next, we’ll analyze the economic value of selling our discharged power into CAISO’s real-time market. We will use exactly the same VPP strategy that we used in the first example, which allows us to stack this new value stream on top of the first.

CAISO operates uniform price auctions, which means that in a CAISO auction all winning bidders receive the price that clears the auction, even if they bid below the clearing price.

In our first example, we already made a plan to discharge power at peak times, which allows us to avoid purchases in the day-ahead market. For the real-time value stream, we simply bid all discharged power into the market at a price of $.01, which will win all real-time auctions that have a positive price and which will pay off at the real-time market clearing price.

Figure 3 shows the value stream we can expect to yield from this strategy, which has the same characteristic shape that we saw in the first example. Our new strategy yields more than twice the value of the first!

Figure 3: Annual savings per kW of 4-hour VPP from selling discharge into CAISO’s real-time market.

Value Stream: Resource Adequacy Requirement Reduction

A CCA’s Resource Adequacy (RA) requirement is based on the CCA’s contribution to CAISO’s monthly peak system demand. For individual CCAs, which make up a relatively small share of total CAISO system demand, it is fair to approximate the CCA’s reduction in RA requirement as tracking linearly with the CCA’s reduction in peak monthly load. In other words, a 5% reduction in a CCA’s peak load will lead to a 5% reduction in RA costs.

Figure 4 shows the value stream we can expect to achieve from this strategy. One item of note is that this new value stream has a concave, rather than convex shape. This is because of the geometry of the load curve. It takes more and more stored energy to achieve each incremental reduction in peak demand.

Figure 4: Annual savings per kW of 4 hour VPP due to reduction in resource adequacy requirement.

Value Stacking

Figure 5 shows the result of stacking the three value streams outlined above. Because all three strategies can be achieved simultaneously, it’s perfectly fair to stack these particular value streams.

Figure 5: Stacked cost savings from three value streams under maximal load-flattening strategy.

Other Value Streams

There are other unanalyzed value streams that are more challenging to evaluate, but that exist never-the-less.

First, customers often operate batteries to reduce utility demand charges. Although we are not explicitly chasing customer demand charges with our VPP strategy, any customer demand that is coincident with system peak will be reduced as we reduce system peak via our VPP operation. This is how it should be. Customers should be compensated for reducing their contribution to system peak load, and they should not be compensated for reducing demand that doesn’t impact system peak. To analyze this particular value stream, we need to know both the utility’s demand charge structure and we need the customer’s load curve.

Next, there is another value stream from reducing greenhouse gas (GHG) emissions. CAISO’s system-wide mix of wholesale power varies during the day and year, becoming cleaner during times when renewable power is plentiful and becoming dirtier when renewable power is scarce. By storing clean power for discharge during times when system power is dirtiest, we achieve a GHG reduction benefit. (Visit our Terrablog post, ‘How batteries can reduce CCAs GHG emissions‘ to learn more on this subject.)

This value stream is also difficult to evaluate because California does not yet have a mechanism for exchanging time-based renewable energy credits. Under present GHG reduction calculations, all GHG reduction is treated the same. But under the new Clean Net Short method, Load Serving Entities must reduce GHGs hour-by-hour. Therefore, an hour of clean energy supplied during the night should be more valuable than an hour of clean energy supplied during the day. But there is currently no market mechanism to help us quantify the difference in economic value.

Finally, battery systems can provide resiliency benefit to the customer that sites the battery. Although the portion of battery that is reserved for resiliency can not be used to provide the previously analyzed value streams, any portion of the battery that is set aside to be used for grid operations can help offset the installation costs of a resiliency system.

The value of a resiliency system to a customer depends very heavily on the costs that occur if a customer loses power, so it will vary from customer to customer. However if a customer requires a resiliency system, the customer can reduce the cost of that system by allocating a portion of the system for grid operations.

Unanalyzed Costs

Not included in this analysis are two costs that are currently unknown.  First, the scheduling coordinator typically works for a fee that would likely be adjusted if the scheduling coordinator’s scope of service is increased.  Second, access to CAISO markets and communication to the DER aggregation must be supplied by a company that supplies VPP services.  The VPP operator’s fees are unknown to us presently, and may be significant.

Conclusion

By analyzing the procurement cost impact of operating an aggregation of DER resources, we can understand the economic value of a DER aggregation, which allows us to design cost-effective customer programs for acquiring access to DER resources.

For CCA program managers, this an opportunity to acquire a valuable resource for their procurement manager counterparts while still acting locally to make a global impact.  For procurement managers, this is an opportunity to thank your hippies for a job well done.

Keep an eye out for future posts related to the SEIN Advanced Rate Design project on the subjects of:

• The FlattenMyLoad algorithm for maximally flattening any load curve.
• Creating customer programs for building VPPs and other DER aggregations.
• Final SEIN Advanced Rate Design report and codebase.