*ISO Rule – Loss Factor and Loss Factor Methodology*

*ISO Rule – Loss Factor and Loss Factor Methodology*

This decision follows in a series of decisions regarding a complaint made by Milner Power Inc. (“Milner”) on August 17, 2005 regarding Independent System Operator (“ISO”) Rule 9.2: Transmission Loss Factors and Appendix 7: Transmission Loss Factor Methodology and Assumptions (collectively, the “Line Loss Rule”), which was implemented by the Alberta Electric System Operator (“AESO”) on January 1, 2006.

On February 28, 2011, Proceeding 790, which is the subject of this decision, was bifurcated by the AUC to consider the following issues separately:

(a) Phase 1: Whether the AESO’s Line Loss Rule contravened section 19 of the Transmission Regulation; and

(b) Phase 2: What remedy, if any, could be awarded to Milner in the event the AUC held in favour of Milner in Phase 1.

Phase 1 of Proceeding 790 is completed, with the AUC having upheld Milner’s complaint, in Decision 2012-104 and Decision 2014-110, that the Line Loss Rule was unjust, unreasonable or unduly preferential, arbitrarily or unjustly discriminatory or inconsistent with or in contravention of the Electric Utilities Act (the “EUA”) or the Transmission Regulation (the “T-Reg”). The AUC also found that the Line Loss Rule, as it exists today, does not support the fair efficient and openly competitive operation of the market.

# Background

The AUC, in its previous decisions in Proceeding 790, provided a simplified explanation of how line losses are calculated on transmission lines. Losses are typically expressed through the equation L = aP2, whereby a is a constant number, and P is the power flowing over a given power line. Therefore, the amount of losses on a line increases exponentially with the flow of power. As an example, the AUC noted that a line with a power flow of 100-megawatts would incur four times more losses than if that same line had a power flow of 50-megawatts. As a result of the relationship between distance, power flow and line losses, the proximity and size of a given generator to load customers plays a significant role in reducing or increasing line losses.

Under the T-Reg, the overall cost of transmission line losses is borne by generators, and the AESO is responsible for preparing line loss factors amongst generators to distribute the cost of losses. Some generators receive a credit for reducing losses, whereas others incur a charge for increasing losses, depending on their location and contribution to line losses on the transmission system.

# Allocation of Losses Amongst Generators

The issue at hand in this decision was how to properly allocate the loss charges and credits amongst the various generators in the province.

As a first step in determining the loss factor for each generator, the AESO is required to generate raw loss factors. The AUC described several methods that can be used to calculate loss factors.

# Raw Loss Factor Approaches

A Marginal Loss Factor (“MLF”) refers to the last loss caused by the last unit of power generated. Given the square relationship between line losses and power flow, losses increase at an increasing rate as more power is generated. As an example, the AUC noted that under the equation L = aP2, if a generator is delivering 99-megawatts, it will create 9,801a of losses, whereas at 100-megawatts, it will create 10,000a of losses, and at 101-megawatts it creates 10,201a. Therefore, the marginal losses for the last increment of generation are around 200a at 100-megawatts, which the AUC noted was expressed by the formula MLF = dL / dP = 2aP.

However, the AUC noted that the scenario becomes more complex as more generators are added to a system. The AUC noted that if a generator connects next to a local load, and begins generating, total system losses will decline overall, but as the new generator approaches 100-megawatts of generation, the reduction of losses is reduced until it no longer has an impact on the amount of losses on the system with its last increment of generation. The AUC expressed the impact of the second generator’s connection in the figure below:

The AUC noted that while MLF allows the measurement of the impact of the last unit produced (the red line in the figure), another methodology attributes losses by looking at the discrete impact before and after the new generator is connected (the yellow line in the figure). The AUC referred to this method as the Incremental Loss Factor (“ILF”). In the above example with the new generator connecting, the ILF looks to the losses on the system prior to the operation of the new generator, and the losses after the operation of the new generator.

The mathematical expression of ILF by the AUC was described as ILF = [ L(P) – L(0) ] / P, where L(P) is the total losses calculated at the final output of the new generator, and L(0) is the total losses associated with the generator before it generates anything. Using the figure, the new generator creates approximately 2,500a of losses at its final generation output, and the system losses prior to the operation of the generator are 10,000a. Therefore the net impact on losses for that generator is -7,500a.

The AUC explained that the Line Loss Rule, which was the subject of Milner’s complaint, used a modified version of MLF. While the AUC described it as mathematically complex, it could be simplified as essentially an averaged snapshot of line losses from a generating unit during twelve periods throughout the year (high, mid, and low scenarios for each season), which is then divided by two. Therefore the AUC explained that in its simplest expression, the Line Loss Rule used what is called MLF/2. The resulting loss factor would be multiplied by the price and energy produced in each hour, and then summed for all energy produced in a year.

The AUC noted that losses generally take two forms:

(a) A generator’s “own losses” created by losses in the form of heat resulting directly from the generator’s own transmission of electricity; and

(b) “Aggregate system losses”, either positive or negative, which result from that generator’s power flows displacing and changing the power flows and consequential losses of other generators on the system.

In Decision 2014-110, the AUC found that the MLF/2 methodology did not calculate these raw loss factors in a manner that was compliant with the T-Reg or the EUA.

The AESO proposed a variant on the ILF methodology in this application. However, the AUC also noted that the proposed new Line Loss Rule filed by the AESO was not a filing contemplated under Section 25(7) of the EUA, noting that the AUC did not direct the AESO in respect of what a compliant Line Loss Rule should include. Therefore the AUC found that it was exercising its jurisdiction under Section 25(6) of the EUA to specify what changes, if any, are required to make the Line Loss Rule compliant with the EUA. Therefore, the AUC did not consider the AESO’s proposal to be the only method properly before the AUC, and held that proposals made by all parties would be entitled to a full consideration.

Two main approaches to calculating loss factors were put forward in the proceeding:

(a) The ILF methodology, proposed by the AESO and supported by a number of parties; and

(b) The Superposition methodology, proposed by ENMAX Energy Corporation (“ENMAX”) and supported by TransAlta Corporation (“TransAlta”).

# ILF Methodology

The AESO described the ILF methodology as essentially a “but for” approach, calculating the difference between system losses with and without each generating unit by examining the changes to the system losses between the average level of net-to-grid generation from a specific generator, and reducing that generator’s output to zero.

The AESO (and the parties supporting it) submitted that an ILF methodology gives effect to the requirement that the loss factor measures the impact of a generating unit on average system losses as it recognizes the full range of output, in contrast with the previous MLF/2, which only recognized the last increment of generation. The AESO also contended that the proposed methodology accounts for a generator’s location as well, and satisfied all the legislative requirements for a Line Loss Rule.

While most parties supported the AESO’s proposed Line Loss Rule, the AUC noted that a number of parties had concerns related to implementation, notably as it relates to the location at which the ILF is calculated, and various technical issues related to the AESO’s choice of swingbus to rebalance the system once a generator is withdrawn in calculating raw loss factors.

# Superposition Methodology

ENMAX, however, took serious issue that the only problems with ILF were related to implementation, and challenged the validity of the ILF approach as a whole. ENMAX submitted that the ILF methodology failed to accurately reflect how an actual power system works, and assumed a fundamental misallocation of losses. ENMAX instead proposed a Superposition loss factor methodology, based on a theorem of superposition. ENMAX summarized its proposal as reliant on the element voltages and currents from each applied source acting separately, essentially act together to form the algebraic sum of currents and voltages on the system. ENMAX likened its Superposition methodology as attempting to track the net flow of electrons from each generator to various points, to determine its contribution to system losses. TransAlta summarized the Superposition methodology as follows:

(a) Assess whether each generating unit’s full injection is distributed by the system topology to serve loads;

(b) Assign a credit to a generating unit for reducing flow across a transmission element and assign a charge to a generating unit for increasing flow on a transmission element; and

(c) Aggregate the credits and charges to determine a generating unit’s contribution to system losses, and divide this aggregate by the generating unit’s injection to derive a raw loss factor.

ENMAX and TransAlta submitted that the Superposition methodology was superior since it could not assign a different loss factor to co-located units, while still recognizing the full range of a generator’s output.

The City of Medicine Hat (“Medicine Hat”) took issue with ENMAX’s Superposition method, arguing that it did not calculate any incremental impacts of system losses, but rather attempts to deconstruct and assign power flows within a single operating state. Medicine Hat contended that the Superposition method was therefore not compliant, since the AUC held, in Decision 2014-110, that the contribution or impact of a generating unit was based on a change in system losses, therefore necessitating a base case and a change to the base case.

Milner Power Inc. (“Milner”) and ATCO Power Ltd. (“ATCO”) also took issue with the Superposition Method, notably that the allocation of losses to co-located generators was without economic justification and arbitrary, violating the principles of cost causation.

# Ruling Whether the Methodologies Are Consistent with the Legislation

The AUC held that its concern at this stage of the proceeding was to determine whether any (or neither) of the proposed Line Loss Rule methodologies complied with the existing legislation. The AUC held that if it could be determined that a methodology did not comply, even on a single ground, that could not be remedied, it would be removed from further consideration.

The AUC determined that the ILF methodology was reasonably capable of producing a Line Loss Rule and of calculating line loss factors that were not unjust, unreasonable, unduly preferential, arbitrarily and unjustly discriminatory, nor inconsistent with the applicable legislation. The AUC found that this was due to the fact that ILF by definition provides a given generating unit’s impact across the full range of its output. The AUC also found that ILF calculated loss factors that were representative of the impact on the average system losses of each generating facility relative to load since it measured the difference in losses with and without each generating facility.

The AUC further held that the Superposition method failed to comply with the legislation and was incapable of being remedied to comply with those requirements. The AUC provided the following reasons, among others, for making the above finding. The Superposition methodology:

(a) Failed to recognize a generator’s contribution to line losses across the full range of the facility’s output, since the Superposition method did not compare the impact on losses between a base case and a second scenario, but rather only the base case condition; and

(b) Did not allocate losses to the generating unit causing them, as it allocated losses equally to facilities injecting power onto the same transmission element. The AUC determined that this would inevitably result in the socialization of line losses, thereby violating the principles of cost causation.

# Location

Under the AESO’s proposed filing, the “location” of a unit was considered to be its metering point identifier. This approach, according to the AESO, would provide identical loss factors to co-located units that are operated in the same manner, while allowing differently sized or operated units to be assigned different loss factors. The AESO submitted that this would remove the potential for discrimination between closely located generators with different impacts on system losses.

The AUC directed the AESO to make changes to the Line Loss Rule to specify that the location of a “generating facility” as defined in the EUA, be the location of each metering point identifier for a generating unit or group of generating units. This determination, according to the AUC would allow for generators that own or control generating facilities to aggregate or disaggregate their generating facilities as they choose, and at their own expense.

Such aggregations, or disaggregations, in the AUC’s determination, would allow market participants to contract for as many or as few metering point identifiers as they wish. However, any units that are aggregated must also offer into the energy market as a single source asset through one set of price/quantity pairs. The AUC further requested comments from the AESO in its compliance filing regarding how the ability to aggregate or disaggregate would apply to units subject to a power purchase arrangement.

# Swingbus and Base Cases for ILF Analysis

The AUC noted that the principal advantage of the ILF method is that it applies a “but-for” approach, examining the full range of a generating unit’s output. However, a shortcoming of this approach is that the power system must be “rebalanced” through a swingbus to account for the removal or addition of the generating unit. The AUC noted there are two ways of rebalancing a power system:

(a) Scaling down load; or

(b) Scaling up generation to replace the output of the removed generating unit.

The AESO, in scaling up generation has two further options of hypothetically re-dispatching generation:

(a) Rely on the generic stacking order to rebalance the system; or

(b) Rely on the energy market merit order to rebalance the system.

Parties supporting the ILF methodology generally favoured the load scaling method over re-dispatching through the generic stacking order, relying on previous findings by the AUC in Decision 2014-110 that load scaling was “not inconsistent” with the T-Reg. The AESO however, offered its reasoning that reducing loads proportionally across the system removed any subjectivity in selecting the necessary adjustments following the removal of the generating unit.

Milner’s reasons favouring load scaling was appropriate since the T-Reg required that load not pay directly for losses.

With respect to re-dispatch through the generic stacking order, the AUC determined that the witness testimony provided during the oral portion of the hearing made clear that the generic stacking order was never intended to reflect which generating facility would be dispatched next in a system rebalancing scenario. The AESO opposed the use of the generic stacking order on the basis that re-dispatch under the generic stacking order is dependent on the relative location of the generating unit being removed, and the unit being re-dispatched. The AESO also opposed the use of the generic stacking order, since it is a forecast based on a historical analysis of losses on the system, and is therefore subjective.

The AUC held that losses cannot be determined in a vacuum, as they depend on a number of factors on the system in real-time. The AUC held that scaling load down to rebalance the system introduced a conceptual problem in that what is measured it not what actually occurs on the system when a generator is removed. In reality, the AUC found that the AESO would dispatch other sources of generation to take up the lost generation when a generator is removed from service (such as during an unforeseen maintenance event.) Therefore, the AUC held that it would be reasonable to expect the “but-for” analysis to examine the system at a constant load, and thereafter model the total line losses by dispatching other generating facilities to match the load. The AUC also held that scaling down load would be an abnormal operating condition under section 31(2) of the T-Reg, since the AESO very rarely, if ever, curtails load on the system in day-to-day operation. However the AUC stopped short of finding that load re-balancing would violate the T-Reg, since the entire exercise is entirely hypothetical.

The AUC therefore determined that the full load on the system is required for rebalancing, rather than scaling down load. The AUC considered that dispatching up using the merit order more closely reflects what would occur in reality if a generating unit were to suddenly go offline. Therefore, the AUC held that re-balancing the system through merit order dispatch would be the most practical.

The AUC further held that the AESO has the merit order readily available, since it is compiled 8,760 times each year (one for each hour), and should therefore be used as a base case for the calculation of loss factors. Additional reasons were provided by the AUC for directing that the AESO use 8,760 base cases. Among these reasons were:

(a) A larger number of base cases instills greater confidence in forecast loss factors;

(b) Application of the merit order will reduce the necessity of manual interventions by the AESO in developing loss factors;

(c) The merit order is a transparent and publicly available record for the prior year;

(d) Using 8,760 merit orders allows the AESO to use a simple average of raw loss factors rather than a weighted average, which can then be clipped and shifted to within the appropriate collars.

The AUC recognized the potential administrative ramifications of moving from 12 base cases to 8,760 base cases, and therefore requested the following information from the AESO prior to making its compliance filing for the Line Loss Rule:

(a) The operational ramifications for developing 8,760 base cases, including labour, equipment and processing timeframes and costs;

(b) Whether a different number of bases cases would provide the same potential accuracy as 8,760 base cases, and any potential savings associated with such a lower number; and

(c) An estimate of when a new Line Loss Rule could be ready for implementation.

# General Issues

The AUC noted that the ILF method inherently leads to a global over-recovery of losses, requiring a shift factor to be applied to each generator to compensate for the over-recovery. The AESO proposed to make such adjustments at each step to offset the risk of an anomalous raw loss factor materially affecting the final loss factor. ATCO proposed as an alternative that the AESO’s proposal be simplified to applying a volume weighted average thereby obviating the need for multiple rounds of adjustments.

However, the AUC held that its prior determination on the use of the merit order to calculate 8,760 base cases (one for each hour in a year), would result in it not being necessary or desirable to adjust raw loss factors at each base case, and noted that it expected the AESO to address this issue in its compliance filing.

The Line Loss Rule further applies “collars” to the loss factors that fall outside of the limits prescribed by section 31(2)(f) of the T-Reg and charges two times the system average, and credits one times the system average. The AESO proposed to clip and shift the loss factors repeatedly until all loss factors for generating units outside the collars fall within the required limits, as this would be simpler than applying linear compression to all generating units.

The AUC held that the AESO’s proposal was acceptable, finding that it satisfied the requirements of the T-Reg to use a common method to fit loss factors within the collars.

# Order to the AESO

The AUC therefore directed the AESO to file a plan to implement the AUC’s findings in this decision, by February 1, 2016. The AUC directed that this filing by the AESO include a plan to develop a revised Line Loss Rule for approval. The AUC noted that once the plan is reviewed and approved, it would direct the AESO to submit the Line Loss Rule as a compliance filing for review and approval on a date to be determined.

In setting a potential effective date, a number of parties urged the AUC to set as early an effective date as possible, noting that the current unlawful Line Loss Rule has been in effect since January 1, 2006. However, the AUC held that it was constrained in setting an effective date by Section 25(9) of the EUA, which states that the earliest date a rule may become effective is the day on which the AESO files a revised rule.

The AUC also noted that the changes it directed were significant, and would likely require several internal changes for processing and information by the AESO. As a result, the AUC noted that these changes may take time, and that other unanticipated implementation issues may arise.

The AESO expressed some concern about its need to comply with Section 31(2) of the T-Reg, which requires loss factors to apply for a period of at least one year, but not more than five. The AESO’s concern arose from its capability to implement a rule prior to January 1, 2016. The AUC noted that the AESO had several options at its disposal, such as lengthening the time that the 2015 loss factors are in place.

The AUC cautioned that its findings on the effective date for the new Line Loss Rule had no effect on its parallel authority under Sections 119(4) and 121 of the EUA, to adjust line loss charges from January 1, 2006 to the new effective date, and its authority to determine final line loss charges in Module C of this proceeding up to the effective date of the new loss factors.