## .

Fairly Pricing Net Intervals While Keeping the Utility

Financially Healthy

Mark B. Lively
Consulting Engineer
Utility Economic Engineers
Gaithersburg, MD
MbeLively@aol.com

Recent events have made electricity economics more complicated.  Previously, input prices were predictable, as was the mix of generating fuels.  The cost of the utility could thus be well determined ahead of time.  Further, the load patterns of customers simplified rate design.  Prices were often invariant over long time periods.  The price invariance allowed utilities to read inexpensive watt-hour meters once a month, instead of installing more expensive interval meters.

The passage of the Public Utility Regulatory Policy Act (PURPA) of 1978 began a change in the industry, allowing Qualifying Facilities (QFs) to require its local utility to buy the output of their generation.  PURPA also created the concept of net metering, where a single meter would measure the net flow between the customer and the utility, instead of different meters being used for customer consumption versus customer generation.  Net metering requirements changed the customer load patterns experienced by the utility, invalidating traditional rate design.

NET METERING

Electrical engineers often wince when they hear associates use KW and KWH interchangeably.  One is power, one is energy.  The two are related in that energy is the integral of power, as is shown in Figure 1.  Alternatively, power is the average amount of energy over a given period of time.  Figure 1 is a "stylized" daily usage curve for a residential customer.  The height of the curve is power, measured in KW.

The area under the curve (shaded in color) is energy, measured in KWH.  The usage curve is "stylized"  by making it into a straight line, just so some of the calculations, including calculus, is easier later on, for anyone who wishes to replicate the calculations.  The maximum demand is 10 KW.  The minimum demand is 4 KW.  The total energy consumption is 168 KWH.  For a residential customer, this is a large load.

Figure 2 replicates Figure 1 but overlays a daily distributed generation production curve, such as solar, wind, or a home cogeneration plant.

Again the production curve is stylized as a straight line to make the calculus easier later on.  In contrast to the load curve, the production curve slopes in the other direction going from a low of 2 KW to a maximum of 8 KW, for a total production 120 KW.  The opposing slopes were chosen just to make the differences starker.

The concept of net metering is using a single meter to measure the power going into or out of the customer's premises.  Generally, the measurement is the integral of that power over an interval of time.  This integral of power is energy.  Historically, the high cost of metering meant that the interval of time was a month in most parts of the U.S., though some utilities had a standard reading interval of two months, or even a year for some utilities around the world. The electronic revolution has reduced the cost of interval metering and also reduced the standard interval from a month.  In some places the standard interval is one hour, other places 15 minutes, and even as short as 6 seconds, as is used for Australia's Frequency Control Ancillary Services (FCAS).

Short billing intervals are important if loads change rapidly while prices change rapidly.  If neither price nor load changes, then long billing intervals can be used.  For instance, prior to restructuring of the U.S. electric industry, prices changed infrequently.  Rate cases occurred less than once a year.  Fuel clause adjustments (FCAs) were occasionally unchanged month to month.  So metering intervals shorter than a month were not needed.  The desire for increased economic efficiencies suggested time varying prices, though such efficiency gains have been disputed.

The alternative to net metering is gross metering, which involves two meters.  One meter determines the customer load and the other meter determines the customer production.  Besides the added cost to the utility of owning and processing two meters, there is the added cost to the customer to have the production separately wired back to the meter point.  Net metering reduces these expenses but at the confusion of the utility not knowing the load it was meeting and the production that was being supplied by the customer.  From simple algebra in Figure 2, the utility measures 64 KWH as going to the consumer and measures 16 KWH as coming to it, or a net of 48 KWH during this 24 hour period.  Unmeasured is the 104 KWH that the consumer self generated and used itself.

EFFECT OF SMALLER PRICING INTERVALS

The issue associated with net metering begins to appear in Figure 3, which overlays a pricing curve on top of the power curves of Figure 2.

The pricing curves slopes in the same direction as the load curve, going from a maximum value of \$90/MWH to a low of \$10/MWH, for a simple average of \$50/MWH over the period.  Given the slope of the pricing curve, a price responsive generator would have a slope in the opposite direction.  On a simple average basis, the customer could pay the utility \$3.20, or \$50/MWH for the 64 KWH delivered by the utility, and receive \$0.80, or \$50/MWH for the 16 KWH delivered by the customer to the utility.  This is summarized in the first column of Table 1 which uses a simpler calculation.  This simpler calculation looks at the 48 KWH of net deliveries by the utility and has the customer simply pay a net of \$2.40.  But this looks at the entire 24 hour period as a single metering interval, and does not look at the small time intervals.

Breaking the day into three metering intervals of 8 hours each results in the analysis shown to the right four columns of Table 1.

Again, the first column reproduces the analysis for treating the entire day as a single time interval.  The second column presents data for the first 8 hours of the day.  The simple average price during those 8 hours is \$76.66/MWH.  During those 8 hours the customer takes 48 KWH and pays \$3.68.  During the middle 8 hours, the simple average price is \$50/MWH, the customer consumption is 16 KWH, and the payment is \$0.80.  During the last 8 hours, the customer delivers 16 KWH at a price of \$23.33/MWH and receives \$0.37.  The net payment by the customer is \$4.11 or \$1.71 more that the \$2.40 calculated using the simple average over the 24 hour period.  The \$1.71 represents an increase of 71% as a result of using finer billing intervals, at least in this example where the prices and loads are both varying, though admittedly slowly.  Again, the example is for generation that varies contra-cyclically to the price of energy.

The "hourly" line near the bottom of Table 1 presents a summary of similar calculations for each hour within each of the 8 hour periods.  The net payment increases again for this use of even finer billing intervals, though only by another \$0.21.  The "continuous" line at the bottom of Table 1 presents a summary of the calculations using calculus, which essentially uses vanishingly small measurement intervals, similar to the 6 second intervals associated with FCAS.  For this set of data, there is little additional benefit to sub-hourly pricing, essentially not more than a rounding difference.

SETTING REAL TIME PRICES

Utilities long ago realized that economic principles could be used to lower the fuel and other operating costs of generating plants by equalizing the marginal costs of all operating generating plants.  Marginal costs roughly monotonically increase over the operating range of generators.  The resulting optimization protocol is to increase the output of generators with low marginal costs and to decrease the output of generators with high marginal costs.  This protocol moves the marginal costs of the various units closer together.  The protocol also slowly adjusts the operating levels of all generators as load changes throughout the day, week, and year.  This equalization process for marginal costs reflects geographic issues such as line losses (on a marginal basis) and transmission constraints.

The equalized marginal cost concept is a continuous process applicable to those units that are operating.  Turning a unit on or off is a non-continuous unit commitment process.  Each unit commitment action changes the marginal cost equilibrium point, requiring an adjustment to the operating levels of the various units.  The unit commitment process greatly affects the profitability of individual generating units and contributes to the infra-marginal cost of the entire generating system.  For generating systems that include generators with non-uniform ownership interests, the unit commitment process and the equalized marginal cost calculations are greatly scrutinized.  This scrutiny has been addressed by the establishment of Independent System Operators (ISOs).  Most of the continental U.S. is in the footprint of an ISO.

Just as the dispatch of generators is optimized by operating them at an equalized marginal cost level, economic theory suggests that the optimal price to charge purchasers for this electricity is a price equal to the equalized marginal cost.  The ISO dispatch prices for generation provides appropriate marginal cost prices for this purpose.  These dispatch prices are typically developed every 15 minutes, though often averaged over an hour.  The example presented previously in Table 1 suggests that little economic efficiency is lost by using an hourly average price instead of minute by minute prices.  But Table 1 does not have rapidly changing prices or rapidly changing loads.  Rapidly changing prices, such as when generators are switched on or off or when transmission lines are opened or closed can cause great variations in the value of electricity.

Engineers have designed operating systems to respond to sudden changes in the configuration of the generation and transmission system, providing signals to increase or decrease generating levels as appropriate in the various parts of the network.  In extreme cases, under-frequency relays will also interrupt load.  The driver for these rapid response operating systems is frequency and Area Control Area (ACE).  ACE is a combination of frequency error with inadvertent interchange, the unscheduled flow of electricity into or out of a utility.  To some extent, ACE can be viewed as a free storage device, absorbing excess electricity or supplying electricity during a deficit.  Though there is no price for ACE, NERC has mandatory reliability standards that can results in fines when a utility abuses ACE.

Fred Schweppe, the MIT professor who during the 1980s developed some of the early theories for ISO operation and pricing, pointed out that the marginal cost theories were only applicable when the system was in balance.  This balance requirement suggests that the presence of ACE (or of frequency error for an isolated utility) reduces the robustness of ISO pricing mechanisms.  One approach to addressing this imbalance issue is to treat the imbalance as an economic resource.  This concept was presented in "Tie Riding Freeloaders--The True Impediment to Transmission Access,"[1] “Microgrids And Financial Affairs - Creating A Value-Based Real-Time Price For Electricity,”[2] and “The WOLF in Pricing: How the Concept of Plug, Play, and Pay Would Work for Microgrids.”[3]

For instance, a storage device makes money by absorbing electricity when prices are low and should be producing electricity when prices are high.  Thus, when there is excess electricity, as is indicated by a positive frequency error and/or an export of inadvertent interchange, which are physical manifestations of using the grid to store the excess, the system price should be low.  Conversely, when there is a shortage of electricity, as is indicated by a negative frequency error and/or an import of inadvertent interchange, which are physical manifestations of using the grid to take electricity out of storage, the system price should be high.  It is unclear whether NERC mandatory reliability standards would need to be eliminated to accommodate such an automated, dynamic pricing mechanism.  The above mentioned papers suggest using such a pricing mechanism for unscheduled flows of electricity between and among ISOs and/or utilities, for ISO internal pricing within the normal 15 minute dispatch intervals, and for independent utilities that are dealing with independent power generators within their footprint, such as the net metering loads described in Figure 2.

UPLIFT

A big issue with net metering is Uplift, a term common among the independent system operators (ISOs) for the cost not recovered by the hourly calculation of revenue from loads.  For a vertically integrated utility, Uplift can be considered to be the difference between the total installed and operating cost of the generation function versus the revenue associated with the generation function (or perhaps the generation and transmission function).  There can also be Uplift associated with the other functions.  Most utilities in the U.S., are distribution utilities, either because they were originally constituted as distribution utilities (such as many rural electric cooperatives and some municipalities) or because of the wave of restructuring that occurred near the turn of the millennium.  For distribution utilities, Uplift is the difference between purchased power costs and the associated "generation related" revenue.

In the example so far, the revenue from gross load would have been \$9.36 or \$55.72/MWH, probably much less than the utility's generation costs.  The revenue from net load would only be \$4.32, as shown in Table 1.  Hourly or sub-hourly pricing of net metered deliveries goes only a little way toward collecting Uplift costs.  Since Uplift costs are net of collected revenue, net metering can increase Uplift costs, though in some cases metering can decrease uplift costs.  But, sizeable uplift costs will remain.

Under general economic theory, a producer will build new productive capacity when the expected marginal price for sales will exceed the average cost of new productive capacity.  For electric utilities operating in  competitive market such as an ISO, the marginal sales price is generally set at the marginal cost of production.  Studies of the Australian ISO have claimed that the marginal cost of production never reaches the cost of new production capacity, requiring non-market motivations for producers to add capacity.  Some have speculated that part of the under-collection of operating costs partially relates to a practice of restraint on how high real time prices are allowed to go.

Non-market motivations may include requiring utilities to purchase sufficient capacity to achieve a defined reserve margin.  Some reserve margins are defined as installed capacity versus project load.  Other reserve margins have been defined as operating reserves, such as a real time requirement for spinning reserves.  In the U.S., some of these reserve and operating requirements have been promulgated as Mandatory Reliability Standards under the Electricity Policy Act of 2005.  The Australian studies have been confirmed by internal staff studies at the Federal Energy Regulatory Commission and at the U.S. Department of Energy (DOE) Energy Information Agency.  The studies show that ISO hourly markets do not collect enough money to pay for new capacity.  Further, a major portion of the loads within the ISOs are actually traded outside of the ISO hourly markets through bilateral contracts.

DEMAND CHARGES TO COVER UPLIFT COSTS

Electric utilities typically functionalize their costs among distribution, transmission, and distribution.  Each function has a peculiar mix of revenue drivers, such as short run marginal costs for the generation function and to an extent for the transmission function.

Distribution Uplift

Utilities incur distribution costs for both metering the customer's use of electricity and for the wires that historically moved electricity from the transmission grid to the customer.  The growth of distributed generation now has some of the electricity moving away from the customer.  Generally this electricity can be considered as going to other customers.  It is rare that electricity moves off a distribution feeder back into the transmission grid, though the growth of rooftop solar in Hawaii has been reputed to have resulted in some distribution feeders delivering electricity to the transmission grid.

The cost of customer metering and customer connection equipment is highly dependent on the size of the customer.  Larger customers require more expensive equipment, though the cost of the equipment does not go up linearly with the size of the customer's demand.  Indeed, these costs are very class dependent, sometimes being treated as being uniform within the residential class, within the commercial class, and within the industrial class.  Determining the monthly customer charge necessary to recover these costs is outside the scope of this paper.  The distribution uplift can be considered to be the total distribution revenue requirement after netting the revenue associated with monthly customer charge.

Utilities typically classify costs among customer, energy, and demand.  The distribution grid incurs relatively minor costs that vary with the amount of energy that goes through the distribution wires.  Indeed, about the only "energy-related" costs that a utility incurs on the distribution grid are associated with the electrical losses, and these are typically minor and are treated as generation related costs.  The vast majority of the uplift costs after the netting of revenue from the monthly customer charge are determined the size of the wires that the utility has installed for the distribution grid.[4]  The size of the wires are determined by the demands that the utility expects its customers to impose on the system, whether for bringing electricity from the transmission system or for delivering electricity from the distributed generation to other customers.

In the context of Figure 1, the customer imposes a demand on the utility of 10 KW, at the extreme left side of Figure 1.  Under a gross metering context, the utility would use 10 KW as the billing demand for the customer.  Most utilities use the maximum demand during a month as the primary billing demand.  Some utilities impose a ratchet on the billing demand, which carries the billing demand forward to subsequent months.  A demand of 10 KW imposed on the utility in January would then be used for the billing demand each month afterwards until superseded by a higher actual demand or the expiration of the ratchet.  A demand rate of \$5.00/KW-Mo would result in the customer being billed a monthly demand charge of \$50.  A one year ratchet would result in annual demand charges of \$600.

In the context of Figure 2, the customer imposes a net demand of only 8 KW on the utility, the difference between the gross demand of 10 KW and the gross simultaneous generation of 2 KW.   The reduced demands associated with net metering will necessarily increase the unit price imposed on customers, perhaps \$5.60/KW-Mo instead of the aforementioned \$5.00/KW-Mo.

The implications of using a demand charge to collect the utility's distribution uplift costs are discussed in "Curing the Death Spiral"[5] and "Demand a Better Utility Charge During Era of Renewables: Getting Renewable Incentives Correct With Residential Demand Charges."[6]  ATCO Gas, a distribution only utility in Alberta, Canada, uses a customer charge and a demand charge to bill it large customers with no commodity charge, the gas equivalent of an energy charge[7].  ATCO Gas eliminated its commodity charge for large use customers as of September 1, 2007.  Thus, the use of a customer charge and a demand charge without a commodity charge is a relatively new phenomenon, at least for ATCO Gas.

Generation Uplift

Some ISOs use a coincident demand variation of the demand charge.  Under a coincident demand charge customers are billed a demand charge based on their consumption during periods of stress on the system, generally when the system achieves it maximum demand.  A period of stress could also be when generation is most out of balance relative to excess load, such as has been described in the California Duck Curve or the Hawaii Nessie Curve.

Figure 4 builds on Figure 2 by adding two coincident demand indicators, one at the 1 hour mark, the other at the 18 hour mark.

Table 2 develops coincident demand charges under the assumption that these two demand indicators are the only ones used by the utility for charging the uplift fee.

At the 1 hourmark, the customer has a load of 9.75 KW and generation of 2.25 KW.  The customer's net metered amount is 7.5 KW.  At the 18 hour mark, the customer has a load of 5.5 KW and generation of 6.5 KW.  The customer's net metered amount is 1.0 KW, leaving the customer.  Assuming generation uplift costs of \$20/KW-Mo, the customer would be billed \$150/mo for the coincident demand it imposed on the utility at the 1 hour mark and would receive a credit of \$20/mo for the relaxation of the coincident demand at the 18 hour mark.  Some ISOs use more than two coincident demands for the recovery of uplift costs.

The bottom section of Table 2 provides an advanced coincident demand mechanism, with uneven rates for the two coincident demands.  The difference between the two rates is a factor of two (2), with the coincident demand at the 1 hour mark being billed at half of the rate at which the coincident demand is billed at the 18 hour mark.  The bottom section of Table 2 is just meant to show that the two coincident demands can be billed at different rates, not to establish the mechanism by which the different rates are determined.

Transmission Uplift

The real time price for the use of the transmission system is determined by geographically differentiating the real time price for generation.  The geographic differentiation reflects marginal electrical losses and transmission constraints.  The sudden and often unpredictable occurrence of transmission constraints makes the real time price for the use of the distribution system highly variable, even more so than the real time price for generation.  Some ISOs may even collect more revenue for the use of their transmission systems than the associated revenue requirement.  A coincident demand uplift fee such as that presented in Table 2 may be appropriate for collecting the transmission uplift fee.  To the extent that the transmission costs are less than the real time price, the transmission uplift fee could be a credit to the generation uplift fee.

CONCLUSION

PURPA resulted in greater variation in the perceived value of electricity by forcing utilities to buy wholesale electricity from Qualifying Facilities.  As the market grew, ISOs were created to manage the transactions and provided better quantification of the real time variability of the value of electricity.

PURPA also allowed customers to be both consumers and producers through a new process known as net metering.  The power flow across the interface between the utility and its customers became more variable and less certain.

The combined variability of value and the variability of power flow made finer measurements of the power across the interface more important.  This interval metering concept was facilitated by the advances in the electronics industry, which reduced the installed cost of interval meters as well as the cost of processing the data produced by interval meters.

Marginal cost pricing does not provide sufficient revenue to cover the costs of the utility, a point made most clearly by ISOs that impose Uplift charges to recover the difference between costs and revenue.  The concept of Uplift charges are equally applicable to utilities other than ISOs and can be collected using demand charges.  Coincident demand charges seem to be most applicable for the recovery of generation and transmission Uplift costs.  Customer demand charges have traditionally been found to be most applicable for recovering distribution costs not recovered through monthly customer charges.

Notes:

[1] Public Utilities Fortnightly, 1989 December 21

[2] Cogeneration and On-Site Power Production, September, 2007. http://www.cospp.com/articles/article_display.cfm?ARTICLE_ID=307889&p=122

[3] IEEE Power & Energy Magazine, January/February 2009

[4] It should be noted that sometimes a portion of the cost of wires is treated as a customer charge.

[6] USAEE Dalogue, United States Association for Energy Economics, 2015 January. http://dialog.usaee.org/index.php/volume-23-number-1-2015/271-lively

[7] High Use Delivery Service Rates--ATCO Gas North

Fixed Delivery Charge (FDC): \$5.584/day
Variable Delivery Charge (VDC): \$0.000/GJ
Total Demand Charge: \$0.383/GJ/day of 24 Hour Billing Demand

http://www.atcogas.com/Rates/Current_Rates/