"Secret" Chugach Electric Cost Benefit Study

1998 CHUGACH ELECTRIC "SECRET" COST - BENEFIT STUDY OF THE SOUTHERN INTERTIE RELEASED

Citation:
Samuelson, R., Hass, S., and Chugach Electric Association, 2002, Ratepayer Impacts of Proposed Transmission Projects Final Report; Includes Chugach Electric Association's Comments, for Chugach Electric Association Inc.: DFI-Aeronomics, [6 p.], 75 p. [marked "Client Confidential"] NOTE: Released by vote of Chugach Board on 11/20/02. First available on 12/6/02 with six page transmittal memo and recently added "Chugach Review Comments" imprinted throughout the 75 page original document dated 2/16/98. Each page and cover has background watermark, "Includes Chugach Comments".

In 1997, Chugach Electric Association (Chugach) hired DFI-Aeronomics (DFI) to study the rate impacts of the Northern and Southern Interties. The study was intended to assist Chugach in making a decision whether to participate in the Interties. DFI prepared a confidential report, "Ratepayer Impacts of Proposed Transmission Projects Final Report, February 16, 1998" ("Ratepayer Report"). The report concluded that participation in the Northern Intertie did not benefit Chugach ratepayers. The study also showed the Southern Intertie was beneficial to the Chugach ratepayer. Chugach did not have to act on this information, since the projects had not reached their decision date.

The Chugach Board of Directors approved the release of the confidential "Ratepayer Impacts of Proposed Transmission Projects Final Report, February 16, 1998" at the November 20, 2002 Board meeting.

In early 2002 Chugach sought to update the Ratepayer Report's production costing model, Alaska Transmission Line Analysis Spreadsheet (ATLAS). Unlike the goal of the 1998 study to evaluate the most conservative or "worst case" rate impact, the goal of the 2002 update was to accurately assess the "most likely" case for the Chugach ratepayer. When Chugach reviewed the ATLAS model to update it, it was discovered that it did not represent Railbelt generation and transmission precisely enough to capture the magnitude of production savings. The latest economic evaluation was performed using an in-house model.

When reviewing the Ratepayer Report, the reader should keep in mind that there are four major areas why ATLAS does not model the Railbelt system with sufficient accuracy to meet the goal of providing a "mostly likely" impact on the Chugach ratepayer.

The industry-accepted approach to load modeling for a production simulation is to create a standard 8,760 hour load shape, then create typical weekly or monthly load shapes to which an economic dispatch algorithm of generation resources is applied. ATLAS represents 8,760 hours as 100 segments, each representing 87.6 hours. ATLAS averages the 87.6 hours into one data point to represent a single demand value for a given 87.6 hour segment. This approach leads to the same level of unit commitment and dispatch for the entire segment. However, it ignores the reality that in each hour, system generation will have a different dispatch level, and that there are different unit commitment requirements over many hours within each 87.6 hour "block" segment of an actual 8,760 hour load curve. While ATLAS does calculate commitment costs for each unit, the simplified averaging of demand over an 87.6-hour segment limits the model's capability to commit units in a manner that accurately represents actual commitment and dispatch practices. This understates fuel cost,

ATLAS represents scheduled generation maintenance and forced outages by de-rating the capacity of units. Units are also shown always available at the de-rated capacity. For example, a 40MW unit that would be off-line for an annual 30-day maintenance period would be represented as a 36MW unit always running in ATLAS.

In actuality, major maintenance on base-load units requires that less efficient intermediate and peaking units operate to meet load. Typical major overhauls or inspections for the larger, more efficient base-load units in the Railbelt can last 90 days or longer, requiring appreciable dispatch of less efficient units during that period. ATLAS essentially runs the most efficient units all the time, but at de-rated capacities. In reality, units are not always available and when they are, they often run at full capacity. ATLAS's approach understates the costs that the Southern Intertie could save.

In general, ATLASs production costing algorithm is more appropriate for modeling a large system such as in the lower 48 where a single component is a very small part - less than 1 percent - of the whole system. On a small system like the Railbelt, where a single generator is up to 10 percent of the entire system, this approach produces errors in the amount of fuel used.

Even more importantly, this error is magnified when applied to the smaller Kenai sub system with the result that Intertie flows are not represented correctly. If the Kenai Peninsula had 40 generators of similar size, showing some units off-line with others at full capacity may not affect the power transfer much relative to running all units at Be rated capacity. However, because the Kenai only has eight generators (2 at Bradley Lake, 1 at Nikiski, 2 at Cooper Lake and 3 at Bernice Lake) of dissimilar size, running each unit at full capacity or taking any one unit down significantly changes the amount of power transferred over the Kenai-Anchorage tie.

ATLAS's approach of showing all units available all the time at de-rated capacity tends to smooth or average out power transfers across the proposed Southern Intertie. For example, consider an autumn scenario. Reservoirs are full and it's raining, so Bradley and Cooper Lake are generating full output around the clock to avoid spilling water. Nikiski is at full load per contract. The Kenai Peninsula load is not very high this time of year, so the existing intertie is exporting power at full capacity. Routine fall maintenance is underway on one or more Beluga units so the next economic unit to run would be Bernice 3 or 4. But the tie is loaded to capacity, so Beluga 1 and 2 and maybe IGT units must run instead which is much more expensive. They are also small units, so more of them have to be started.

In contrast, ATLAS runs Bradley and Cooper Lake at lower output whenever loads are down, which leaves more tie capacity available. The model also runs Nikiski at de-rated capacity, freeing up still more tie capacity. So ATLAS perceives the intertie is exporting well below capacity and starts a unit at Bernice. In this example, the ATLAS algorithm for dispatching fails to model that the intertie is at full capacity. It runs a Bernice unit, which costs much less to operate than Beluga 1 and 2, or IGT units. Production costs are under estimated and over several weeks, this becomes a significant understatement.

This concept is illustrated simplistically in the following "Actual System Operation and ATLAS Model" diagram.

ATLAS does not remove retired units from consideration in unit commitment and economic dispatch. ATLAS developers apparently assumed that retired units would not be run in the economic dispatch since their efficiencies would be significantly lower than the generic replacements. However, there is no mechanism to prevent existing units from continued dispatch. Many units scheduled for retirement in ATLAS are more efficient than intermediate and peaking. units remaining on the system, so retired units continue to be dispatched in ATLAS well 'after their specified retirement dates. For example, the ATLAS model scheduled ML&P's combined cycle Unit 6/7 to retire in 2003. ATLAS simulation results showed that the units continued to be dispatched with a capacity factor of nearly 100 percent through the end of the study period (2045). Similar problems were found with retirement of Chugach Beluga Units 3, 5, 6, 7, and 8. Removing retired units would force less efficient units to run. ATLAS does not capture the cost to run these less efficient units. This understates the costs that the Southern Intertie could save.

Failure to remove retired units also miscalculates power transfers between areas. Because ATLAS is dispatching units that no longer exist, it understates energy production necessary from generators in other locations. Therefore the model does not accurately represent power transfers that would occur on the Southern Intertie.

As loads increase over the study period and new generation needs to be added, ATLAS will add a generic generating unit to a utility's resources at a level that just meets the generation deficit. ATLAS's generic unit is a typical 120MW combined cycle combustion turbine. If a utility experiences a deficit of 5MW in a given year, ATLAS will 'commission' a 5MW resource that has the high-efficiency characteristics of a fully loaded 120MW combined cycle combustion turbine. The cost to operate a 120MW unit at 5MW is different than operating a 5MW unit at full load.

In reality a new 120MW unit installed in the Railbelt would be one of the most efficient units and it would operate at full capacity in lieu of other units. This will change power transfers depending on where the 120MW unit is located and which units are displaced. This is another reason why the model does not accurately represent power transfers that e would occur on the Southern Intertie.

ATLAS performs a full commitment and dispatch simulation for every tenth year (i.e., 2005, 2015, 2025, etc), then interpolates production cost results between those years. Related to Item 3 above, if a significant number of existing units are retired just after a simulation year, ATLAS will ignore those retirements until the next simulation. The result is inaccurate representation of the dispatch for the intervening years and associated errors in production cost and power flows.

Three studies have examined the value of the Southern Intertie. While all three studies had different objectives and assumptions, they all showed the Southern Intertie was a good investment. However, they cannot be compared directly to one another because they each measured benefits differently.

1. The original 1989 and 1998 updated economic analysis prepared for the Southern Intertie's Final Environmental Impact Statement (FEIS) measured the economic value of the project from a statewide perspective. The study estimated the benefits and costs based on cash flows. It did not include the state grant as a benefit.

2. The 1998 Ratepayer Report compared the Southern Intertie's benefits and each utility's share of the Southern Intertie's cost. The State grant was included as a benefit. The comparison was made using benefit/cost ratios based on the net present value of each utility's change in a 40-year projection of revenue requirements. It also estimated the change in a ratepayer's electric bill.

3. The Chugach 2002 in-house analysis measured the economic, financial and electric bill impact of the project from a Chugach system perspective. The State grant was included as a benefit.

Variable	FEIS	ATLAS	Chugach
1. Discount Rate	4.5% real	9.0% nominal	8.0% nominal
2. Fuel Price Escalation	1% to 2020 real Flat 2020+	4.0% nominal	2.2% nominal
3. Inflation	0%	3.0% nominal	2.8% nominal
4. Present Value	1997	1999	2003
5. In-Service Date	2004	2004	2006
6. Anchorage & Mat Su in 2010 Load	510 MW	541 MW	560 MW*
7. Submarine Cable Replacement	1/2 in 17 years 1/4 in 34 years	Not included	All in 23 years per 1999 depreciation study
8. Grant with interest earnings	$0	NA**	$70 Million
* 2000 ASCC Coordinated Built Supply Report for Anchorage area. ** NA - not available - study does not show these values.

The 1998 Ratepayer Report provides insight to how the Southern Intertie may affect its owners. However, it does not represent Railbelt generation and transmission precisely enough to capture the magnitude of production savings. Chugach decided not to update the ATLAS model and used its in-house production costing model.

Following is the Ratepayer Report that includes review comments that identify out-of-date or incorrect statements.