Loading...
HomeMy WebLinkAbout20071101Report Addendum.pdfRECEI\/ IDAHO~POWER~ An IDACORP Company ZOOl OCT 3 I PH It: 4 7 Barton L. Kline Senior Attorney UTi j i~- ~:J ~H Ji'~ \ ~ S I (; : October 31 2007 Jean D. Jewell , Secretary Idaho Public Utilities Commission 472 West Washington Street P. O. Box 83720 Boise, Idaho 83720-0074 Re:Case No. I PC-07 - In the Matter of Idaho Power Company s Petition to Increase the Published Rate Eligibility Cap for Wind Powered Small powerProduction Facilities; and To Eliminate the 90%/110% Performance Band for Wind Powered Small Power Production Facilities Dear Ms. Jewell: In the Settlement Stipulation filed in this matter, Idaho Power agreed to prepare an addendum to its Wind Integration study. Please find enclosed for filing eight copies of Idaho Power s Report Addendum for Operational Impacts of Integrating Wind Generation into Idaho Power s Existing Resource Portfolio. I would appreciate it if you would return a stamped copy of this transmittal letter in the enclosed self-addressed , stamped envelope. Very truly yours~I~ Barton L. Kline BLK:sh Enclosures O. Box 70 (83707) 1221 W. Idaho St. Boise, 10 83702 ~ort A~dendum ". "..... -... OCT '.3! :;;~- DAHO R.. Operational Impacts of 11J~~~~~ti8gi;;~:~;iC Wind Generation into Idaho Power Existing Resource Portfolio An IDACORP Company IPC- E-O7 - F;;ECE.i\' Re ort Addendum 2007 OCT 31 Pi'l L:: l?8 U T ! L 8 -f~~Jc!~_H!ri.. ~j \ ~ S i 0 Operational Impacts of Integrating Wind Generation into Idaho Power s Existing Resource Portfolio Prepared by: t n e N e"x i' IIWI) POWER~ An IDACORP company EnerNex Corporation 1 70C Market Place Boulevard Knoxville, Tennessee 37922 Tel: (865) 691-5540 ext. 149 FAX: (865) 691-5046 bobz(Q)enernex.com www.enernex.com Idaho Power Company O. Box 70 Boise, Idaho 83707 October, 2007 Table of Contents Section 1 Section 2 Section 3 Section 4 Section 5 Overview........ ........................................................ ....... ................ .............. 1 Study Methodology ..................................................................................... Public Workshops....................... ............ ........ ............................ ................. Revised Analysis and Impacts .................................................................. Revised Regulating Reserve Requirements ......................................................... Load Reg-Up/Reg-Down.................................................................................... Wind Reg-Up/Reg-Down ................................................................................... Total Reg-UpjReg-Down ................................................................................... Revised Model Inputs........................................................................................ Sensitivity Runs ................................................................................................ Market Price Assumptions ................................................................................. Jim Bridger Regulating Reserve........................................ ................................. Updated Study Results.............................................................................. List of Figures Figure 1.Distribution of errors for hour-ahead forecast load ................................................. Figure 2. Actual operating hour load vs. forecast hour-ahead load......................................... 17 Figure 3. Actual operating hour wind vs. forecast hour-ahead wind ....................................... Figure 4. Modeling results as a percentage of market prices using actual market prices and no Bridger coal plant for regulating reserves ............................................. Figure 5. Modeling results using 2006 market prices and the Bridger coal plant for regulating reserves ............................................................................................ Figure 6.Updated study results as a percentage of 2006 market prices without use of the Bridger coal plant for regulating reserves............................................................... 25 Figure 7. Idaho Power s updated cost estimate (in $/MWh) for wind integration ........... 26 List of Ta bles Table 1. Table 2. Table 3. Table 4. Table 5. Table 6. Wind machine capacity factor comparison....................................................... Assignment of extraction points to wind generation scenarios........................... 7 Average levels of regulating reserve by wind penetration level................................ 20 Modeling results as a percentage of market prices using actual market prices and no Bridger coal plant for regulating reserves............................................................... 23 Modeling results using 2006 market prices and the Bridger coal plant for regulating reserves .............................................................................................................. 24 Updated modeling results as a percentage of 2006 market prices without use of the Bridger coal plant for regulating reserves............................................................... 25 Section 1 OVERVIEW The objective ofthis study is to assess the costs that Idaho Power will incur in modifying its operations at the Hells Canyon Complex for "integrating" or incorporating wind energy onto its system. The intennittent and unpredictable nature of wind generation requires a utility to have generating resources available which can increase or decrease generation on short notice in order to keep the interconnected power system balanced. While hydroelectric power plants are well suited for perfonning this function, there are operational impacts and costs associated with operating Idaho Power s hydroelectric plants in a manner that maintains reliability and facilitates integration of energy from wind generation facilities. Under the Public Utility Regulatory Policies Act of 1978 (PURP A), Idaho Power is required to offer independent developers a power purchase contract based on a standard avoided cost rate for a qualifying facility with an output of 10 aMW or less. Due largely to federal tax incentives and favorable PURPA rates, a large number of wind project developers came to Idaho Power in 2005 requesting PURP A contracts. Because uncertainty in integrating this large volume of wind generation on its system Idaho Power sought temporary relief from PURP A requirements until the impact of wind integration could be more fully studied. The Idaho Public Utilities Commission (IPUC) granted this relief by temporarily reducing the PURP A cap of 10 aMW to 100 kW for PURPA wind projects. Variability and uncertainty are the two attributes of wind generation that underlie most of the concerns related to power system operations and reliability. In day-ahead planning, for either conventional unit commitment or offering generation into an energy market forecasts of the demand for the next day will drive the process. In real-time operations the output of generating resources must be continually adjusted to match the ever- changing demand pattern. The inherent variability and uncertainty of wind generation may complicate the ability of matching these generating resources to loads. Adding wind resources may also increase the challenge of meeting demand at the lowest cost while maintaining system reliability. The primary focus ofthis study has been to detennine how the real-time operation of Idaho Power s Hells Canyon Complex would be impacted by the addition of significant amounts of wind generation. Previous wind integration studies (of large amounts of wind generation) have shown that the impacts of wind generation uncertainty and variability on the bulk power system are primarily economic, and manifested in increased system costs. These costs are a consequence of the additional controllable generation capacity that must be allocated to manage the incremental variability of the Balancing Authority area due to wind generation, and the increased uncertainty that must be dealt with in operations planning. Following the completion of the original report in February 2007, which resulted in an integration cost of $1 0.72 per MWh, Idaho Power conducted a public workshop on March 15 2007 to fonnally present the results of the study and to solicit feedback from representatives from the wind industry, environmental groups, customer groups and governmental and regulatory entities. At this workshop, a list of 18 items consisting of questions, concerns and requests was developed for Idaho Power to address. On June 20 2007, a second workshop was held to address questions and concerns raised in the first workshop and to present updated modeling results based on suggestions from the first workshop. The updated modeling resulted in a wind integration cost of$7.92 per MWh. This addendum to Idaho Power s original wind integration study addresses the issues discussed at both workshops and presents modeling results which were updated as a result of the workshops. Section 2 contains a brief summary ofthe methodology used to conduct the study followed by a summary of the issues raised at the public workshops in Section 3. Section 4 presents the work done since the completion of the original report regarding the detennination of reserve requirements as well as the results of sensitivity analyses. Lastly, Section 5 presents updated study results. Section 2 STUDY METHODOLOGY While there is no fonnal or rigorous definition , " integration cost" is the tenn used to describe the economic impact of wind generation variability and uncertainty on the utility company charged with accepting and delivering that energy. The tenn applies to the operational time frame, which comprises the real-time management of conventional generating units and the short-tenn planning for demand over the coming day or days. As evaluated in this study, the tenn does not include costs related to transmission upgrades required to deliver wind generation to serve load or for off system sales. A chronological operations simulation methodology has become the de-facto standard analytical approach for wind integration studies. This framework utilizes synchronized hourly load and wind generation patterns, and mimics the scheduling and real-time operation activities for the company or area of interest. For this study, Idaho Power used the Synexus Global Vista Decision Support SystemTM (Vista DSS) to assess the impacts of wind generation on the real-time operation of its system. Vista DSS is a hydro optimization model that simulates the operating characteristics of Idaho Power s system. The model has detailed generating unit definitions, a simplified bus level transmission architecture and hourly inputs for hydro inflows, loads, electricity prices, reserve requirements and energy contracts. This software is capable of optimizing generation scheduling for the Hells Canyon Complex, while observing hydraulic transmission, and regulatory constraints. The generation scheduling computed by Vista DSS for the Hells Canyon hydro facilities includes generation from other Idaho Power resources as well as off-system market transactions. Seasonal water conditions playa critical role in Idaho Power s ability to utilize its fleet of hydroelectric resources. Because of this, three different water condition years were modeled for this study: 1998 (a good water year), 2000 (a nonnal water year), and 2005 (a poor water year). In addition to varying water conditions, the amount of wind generation on Idaho Power s system, or "penetration level " was modeled in the original study for four different cases: 300 MW, 600 MW, 900 MW, and 1 200 MW. The 1 200 MW penetration level was removed from consideration for the updated analysis presented in this report addendum. The study evaluates the changes in operations and the resulting cost that wind variability and uncertainty introduce into Idaho Power s system at the varying levels of wind penetration for each of the three water years modeled. Two Vista DSS runs were needed to evaluate each wind penetration level for each water condition. The first run (flat wind case) modeled wind generation in flat blocks to simulate a predictable resource. The second run (variable wind case) modeled the same amount of wind generation with its inherent unpredictability and variability. The difference between the values of these model runs is the basis of determining the cost to integrate wind. In the original analysis, the flat wind case wind generation was calculated as the average wind energy for a day and was determined by summing 24-hours of wind generation and dividing by 24. This average energy is then applied to each hour during that day resulting in a 24-hour flat block of energy, which removes the variability of wind for that day. The second run incorporates the actual (hourly variable) wind output and the required additional regulating reserves necessary to maintain a consistent level of system control performance. In the updated analysis, the flat wind case was revised such that the daily wind generation was separated into two flat blocks, one for heavy-load hours and one for light-load hours. The wind integration cost per MWh is calculated as the difference between the dollar value of the total annual generation from the flat wind case run valued at market and that of the total annual generation from the actual wind run also valued at market, divided by the total wind energy produced during the year in MWh. This process was completed for each wind penetration level and water year which resulted in a total of six Vista DSS model simulations per wind penetration level to complete the analysis. In order to understand to how Idaho Power calculated the cost of wind integration, it was important to first review the methodology of the study. Further details regarding study methodology can be found in Section 2 of the original report. Section 3 of this addendum presents the questions raised at the March 15 , 2007 public workshop along with Idaho Power s responses which were presented at the second workshop on June 20 2007. Section 3 PUBLIC WORKSHOPS Idaho Power Company completed its original wind study report in February 2007. Following the submittal of the report, the Idaho Public Utilities Commission asked Idaho Power to conduct a public workshop to present the results and to explain the methodology used to conduct the study. The workshop took place on March 15, 2007 and was attended by representatives from the wind industry, environmental groups customer groups, and governmental and regulatory entities. The first workshop resulted in the following list of items (grouped by topic) for Idaho Power to consider with regard to the study methodology and observations concerning the results of the study. The items that were deemed actionable were analyzed and incorporated in updated Vista DSS modeling. Following this additional work, a second public workshop was held on June 20, 2007 to present the updated results. The complete list of items developed at the first workshop and a brief description of actions taken are shown below: Wind Modeling WindLogics should address concerns regarding west to east diversity of wind modeling (re: Idaho National Lab (INL) wind data). WindLogics reviewed the wind data used in Idaho Power s study and compared it with data provided by INL. The INL and WindLogic data were sampled from significantly different heights. This difference made a direct comparison difficult; however there appears to be good similarity of tracking stonn fronts and synoptic patterns. The review of the wind data does not invalidate it as a reasonable basis for detennining wind generation characteristics over the three year study period. Additional infonnation on simulating Idaho wind resources can be found in the original study on pages 20-21. The capacity factors used in the modeling appear to be low. Would going to a different power curve reduce variability? EnerNex has evaluated the wind data using a turbine curve for the GE 1. MW SL and the results showed an even lower capacity factor than the Vestas V82 used in the study. It appears the Vestas V82 turbine works well with the wind resource found in southern Idaho. Table 1 compares the two machines as they are modeled with the wind simulation profiles. Table 1, Wind machine capacity factor comparison 1998 2000 2005 12. 300 673 25,656 25, 600 1,549 29.514 28, 900 265 28.7%211 28, 1,200 007 28,935 27, 300 702 26,686 26, 600 585 30,550 29.4% 900 286 29.232 28, 200 013 28,941 27. 300 617 23,601 22, 600 1,450 27.1,414 26, 900 140 27.085 26.4% 200 835 27,2,764 26. Idaho Power needs to review the data behind wind variability and scaling issues. Scaling issues were primarily related to the increase from the 900 MW to 200 MW penetration level. Because the 1 200 MW penetration level was shown to be beyond Idaho Power s ability to integrate, the 1 200 MW penetration level was dropped from further consideration in the updated analysis. Therefore, the scaling of wind data is no longer an issue in the updated analysis. Item 12 also contains additional information on turbine scaling issues. Can the analysis be re-run at the 300 MW penetration level to account for the recently approved wind contract being in eastern Oregon rather than southern Idaho? The wind data has been updated to reflect the recently approved Elkhorn Valley wind contract (101 MW) in northeastern Oregon. In the original study, the Cotterel site was included in the 300 MW penetration level as 102 MW taken from 6 extraction points. These points were moved to the Elkhorn Valley site and condensed into 5 extraction points. Idaho Power also took this opportunity to remove scaling as much as possible between the penetration levels. This was accomplished by redistributing the Cotterel generation at the 600 MW penetration level to 5 sites. An additional 24 MW site was added at the 900 MW penetration level. These changes enabled a reduction in scaling by reconfiguring the build out, however 3 MW of scaling remained at one site between the 300 and 600 MW scenarios. Table 2 below (Table 5 in the original study) has been updated to reflect these changes. Table 2, Assignment of extraction points to wind generation scenariosRelative Area 300 600 900to Borah Near/Nome Site w~~:/o MW MW MW 1200 MW West Fossil Gulch 10,10,10,10, West Tuana Gulch 10,10.10.10, West Pilgrim Stage 10,10,10,10, West Thousand Springs 10,10,10,10, West Oregon Trails 10,10,10,10, West Salmon Falls West Notch Butte East Milner Dam East Burley Butte East Golden Valley 10,10,10,10, East Lava Beds 10,10.10.10, East Ammon East Ammon East Parker East Parker East Ammon East Ammon East Ammon East Ammon East Ammon East Ammon East Basalt East Basalt East Basalt East Basalt East Basalt East Basalt East Rockland East Rockland East Rockland East Rockland East Rockland East Rockland East Rockland East Rockland East Cotterel East Cotterel East Cotterel East Cotterel East Cottere! Relative Area 300 600 900to Borah Near/Nome Site w~~:/o MW MW MW 1200 MW East CoHerel West Magic Mt West Magic Mt West Salmon Falls West Salmon Falls West Salmon Falls West Glenns Ferry West Glenns Ferry West Glenns Ferry West Glenns Ferry West Mt Home 1.0 West Mt Home 1.0 West Mt Home 1.0 West Mt Home 1.0 West Mt Home 1.0 East Geiger 1 10. East Geiger 2 10, East Geiger 3 10, East Geiger 4 10, East Schwendiman Farms East Windy Pass 10,10, West Tennessee Mt 10. West Glenns Ferry 10,10, West Glenns Ferry 10,10, West Glenns Ferry 10,10, West Magic Wind West Cassia Gulch West Cassia Farm 10,10,10,10, West Glenns Ferry 10, West Glenns Ferry 10. West Glenns Ferry 10, West Glenns Ferry 10, Oregon Elkhorn Valley Oregon Elkhorn Valley Oregon Elkhorn Valley Oregon Elkhorn Valley Oregon Elkhorn Valley Not included in totals above Montana Horse Shoe Bend 11, Montana Arrow Rock 11.0 19,19.19.19. Flat Combustion Turbines & Coal Can Idaho Power utilize existing natural gas-fired combustion peaking facilities (CTs) to provide reserves and load following capability more economically than using the hydro system? An independent analysis evaluated operations using several historic gas price scenarios and shapes against several historic electricity pricing scenarios. The plant under normal system operating and market conditions is generally run about 400 hours per year. The simulation evaluated running the plant for all 8 760 hours in 2005. The economics of running the existing Bennett Mountain simple cycle combustion turbine to provide 10-minute regulation at the 300 MW penetration level for year 2005 was modeled using Microsoft Excel. The model results using the existing Bennett Mountain project to provide regulating reserves proved to be more costly than using the Hells Canyon Complex. The high heat rate of the plant makes the operation uneconomical during most hours of the year and therefore more costly than using the hydro system to provide reserves. 14.Can Idaho Power include the new Evander Andrews unit when investigating the use of combustion units to integrate wind? The new simple cycle peaking plant will have the same operating characteristic of the existing plant described in #3. The regulation benefit of the plant when it is running is limited as it would be available only for reg- down reserves during heavy load hours during which time there is plenty of reg-down reserve available on the hydro units. 18.Run just a combustion turbine analysis (possibly using Aurora). See #3 & #14. Can Idaho Power modulate its coal-fired projects in order to integrate wind? Idaho Power theoretically could modulate the Jim Bridger and Boardman coal-fired plants to a certain degree in order to integrate wind into its system. However, because of the low variable operating cost ofthese facilities, it only makes economic sense to use these resources for reg-down capability during light load hours when market prices are low and generation from the hydro system is reduced and less able to provide reg-down regulating reserves. The updated analysis includes a sensitivity case with 48 MW of reg-down reserve capability assumed from the Jim Bridger Power Plant. It is emphasized that the use of the Jim Bridger plant for this purpose is a pronounced departure ITom current operating practice, and is expected to be problematic considering Idaho Power s position as a non-operating partner at its jointly owned coal-fired resources. Idaho Power s coal-fired resources are typically fully dispatched and operated in a manner that minimizes thennal fluctuations and cycling. Thennal cycling increases the maintenance cost and decreases the reliability of coal-fired units. In addition, Idaho Power is not the operating partner at these facilities and a change in operations would need to be coordinated and agreed to by the operating partner. Therefore, the Company is reluctant to agree to a long-tenn integration cost which assumes deployment of its coal-fired resources in this manner. The purpose of the wind integration study was to detennine the operational impacts arising ITom integrating wind generation, under the baseline assumption that Idaho Power s current system of generating resources, the wholesale energy market with which it interacts, and the general operating practices currently followed would be used to conduct the study. Idaho Power has acknowledged that as experience is gained in operating its system with greater amounts of wind generation and potential cooperative agreements between control areas are developed, a future analysis of the impact of wind generation may indicate a lower cost of integration. However, Idaho Power feels it would be imprudent to detennine the current cost of integrating wind generation into its system based on the speculation of future operating conditions. Regulating Reserves 10. Does the study double-count regulation requirements? In the original study, Idaho Power assumed that regulating reserve was necessary to cover variability in high-resolution load and wind data along with instantaneous la-minute load and wind data. These two sources of variability were combined through a root-mean-square operation, not a straight arithmetic addition. Idaho Power recognizes that the instantaneous la-minute data may include a portion of the variability present in the high- resolution data, and consequently regulating reserves calculated ITom both time series may reflect double-counting. The use of smoothed (e. averaged) la-minute data would rectify this situation. However, smoothed la-minute data for wind generation were not available for the study. While the double-counting likely has a small overall effect, Idaho Power elected to consider only the la-minute instantaneous data in regulating reserve calculations for its updated analysis, removing ITom consideration variability in the high-resolution data. Idaho Power needs to investigate using an "all reg-down" methodology as proposed by Renewable Northwest Project. An all reg-down methodology of maintaining reserves was considered, but was not analyzed because Idaho Power is not prepared to commit to such a significant departure from current operating practices at this time. The asymmetric methodology used in the modeling (described in Section 4 in the discussion on revised regulating reserve requirements) is a more realistic depiction of how Idaho Power will operate using wind forecasts to maximize the hydro operational revenue. What additional reserves does Idaho Power carry to maintain a CPS2 compliance level of 98%? As discussed in the two public workshops, the assumed level of CPS2 compliance is a factor in the estimation of regulating reserve requirements. For its study of wind integration, Idaho Power has assumed a 98% CPS2 compliance level. Because ofFERC Standard of Conduct regulations discussions with personnel from the Company s transmission group for the purpose of quantifying the extra reserve necessary to maintain 98% compliance versus a lower level of compliance are not permitted. However it is possible to use the methodology for estimating reserve requirements described in Section 4 of this addendum report to calculate the additional reserve imposed in the study as a result of the assumed compliance level. It is understood that relaxed compliance assumptions would reduce the estimated regulating reserve requirement for both wind study cases - the flat wind case where regulating reserve is based on analysis ofload data alone and the actual wind case where the reserve level is calculated from analysis of load data and wind data - although the reduction for the flat wind case is expected to be smaller than that for the actual wind case. Using the study methodology for calculating regulating reserve and an alternative compliance level of 95%, the 98% compliance level requires an additional 17 MW of regulating reserve for the flat wind case in the study. For the actual wind case at the 600 MW wind penetration level, an additional 29 MW of regulating reserve is imposed because of the 98% compliance level. While it was not estimated, it is expected that the disparity in the additional reserve between the flat wind and actual wind cases (12 MW at the 600 MW wind penetration level) would be less for the 300 MW penetration level and greater for the 900 MW scenario. It is emphasized that these estimates have relevance only with respect to the reserve levels as imposed in the study, and should not be considered to represent actual reserve relationships as applied in practice. 17.Can Idaho Power calculate the reg-down component of reserves? What about spilling wind? What about the impacts of using a 20-minute ahead forecast? Yes, the reg-down component of reserves can be calculated. The updated study results are based on separate (asymmetric) reg-down and reg-up reserve levels. Furthennore, the asymmetric reserve levels are defined dynamically as functions of load and wind level. That is, given a forecast load of X MW and a forecast wind ofY MW, functions have been derived to estimate the amount of reg-up/reg-down associated with each of the load and wind forecasts. These amounts are added together through a root-sum-square operation to yield a total reg-up/reg-down. This approach is considered a practical way to consider the problem of wind integration from the perspective of scheduling real-time operations. In the original study, a single static, bi-directional, regulating margin was used for each year. Because of the timing issues involved in the scheduling of real-time operations in an hourly market, the use of a 20-minute ahead forecast is considered impractical. In practice, generation schedulers would be unable to derive benefit from such a forecast with regard to relying on the market to make adjustments. Spilling or controlling the up ramp of wind generation is an option for integrating wind generation. This option is based on an economic decision depending on the cost of the wind generation vs. the cost associated with either purchasing or carrying additional reserves on Idaho Power s system. Controlling the up ramp or curtailing wind generation was not factored into any of the regulation reserve scenarios analyzed. Winds down ramps are not controllable from the wind turbine side of the interconnection and would have no effect on reg-up reserve requirements. Further discussion of methods used to estimate regulating reserves is provided in Section 4 of this addendum. General Questions 16.The flat wind HULL bias should be removed from the model. In the flat wind or base case, actual wind generation was originally input as a flat block for the entire day. To address this concern, in the updated model actual wind generation has been separated into two flat blocks, one for heavy- load hours and one for light-load hours. Moving wind remotely has built in a very high transaction cost. This needs to be investigated. This question demonstrated some confusion in how Vista DSS treated the wind generation under the flat versus variable scenarios. The total wind energy is equal between scenarios on a daily basis. The available water was 13. shaped to maximize the economics of the hydro generation subject to regulating reserve constraints. Any transaction cost differential between the flat and variable wind cases is due to the differing reserve constraints and timing of generation subject to the economic optimization algorithm in Vista DSS. In evaluating the purchase or sale of electricity outside of Idaho Power Control Area, an average transmission "wheeling" expense of $5 per MWh was used in the Vista DSS model. The Vista DSS model accounts for this expense when making economic decisions to optimize operation of the system. This transmission expense applies to any market purchase or sale (not just wind generation) and is always a factor when considering the economics of making purchases or sales in the market. Can a further breakdown of the costs associated with the $10.72 be shown in regards to the amount attributed to the hour-ahead forecast, the wind forecast error, reserves and transmission costs (delta between flat and variable case)? The $10.72 cost figure has been highly processed which makes stratifying the components difficult and highly subjective. The $10.72 is a synthesis of model results from three years, interpolated to a 472 MW penetration level which is then applied to a PURP A contract price. The clearest way to think about the components of the cost is what contributes to the reserve requirements. The reserve requirements changed based on the changed variability between cases. For the workshop this equates to an hour ahead forecast error evaluated on a 10 minute time step which was used to construct a 98 percentile confidence interval fonnula adjusted seasonally to derive the hourly up and down regulation reserves modeled in Vista DSS for each ofthe three years. The fast fluctuation component was ignored in setting hourly reserves. 11.Market prices from year 2000 should not have been used due to market anomalies. Idaho Power needs to investigate other pricing alternatives. To address this concern, 2006 market prices were used in the updated analysis for all study years. Can Idaho Power utilize the regional markets to integrate wind more economically? Yes, but there are limitations. The utilization of hourly regional markets is an integral part of operating Idaho Power s system. The initial study results included the utilization of both the east side market and the west side (Pacific Northwest) hourly markets to the degree transmission capacity was available to either import or export energy. Further review of the model results 15. uncovered an exaggerated arbitrage opportunity that had an adverse impact on the results ofthe original study. The modeling ofthe regional markets was modified by setting the east side prices equal to the west side prices in order to eliminate arbitrage opportunities. The arbitrage effects were the result oflarge price differentials between the Mid-C and the Palo Verde markets. The perfect foreknowledge in Vista DSS allowed the model to take advantage ofthis arbitrage situation in an excessive manner. The modeling in the updated analysis removed the arbitrage opportunity by setting the prices at Mid-C levels for both markets. All other market assumptions remained unchanged in the updated analysis. The within hour regulating requirements, which were the focus of the study, are not available for support in the hourly market structure in which Idaho Power operates. The western electricity market operates on an hourly basis which means power is transacted in whole hour blocks and within-hour products are not available. Investigate "what-ifs" associated with expanding the size of the control area. Utilities across the Northwest are investigating the impacts of integrating wind generation and ways of working together that would lessen the impact of the variable and intennittent nature of wind generation. Members of the Northern Tier Transmission Group (NTTG) along with British Columbia Transmission Company (BCTC) have developed and implemented an ACE Diversity Interchange (AD I) pilot program. The program pools Area Control Error (ACE) to take advantage of control error diversity (momentary imbalances of generation and load). This project and others like it will undoubtedly be developed in the future, however the focus of Idaho Power study is to estimate the current cost of integrating wind generation. Idaho Power acknowledges the results will change over time as additional experience is gained and programs like ADI are implemented. In this section, Idaho Power has attempted to address the questions raised at the first public workshop. These questions were the basis for additional work completed between the first and second workshops. Section 4 provides a more detailed explanation of the work completed in regards to the detennination of reserve requirements. In addition Section 4 presents the results of two sensitivity analyses perfonned as a result of issues raised at the first workshop. Section 4 REVISED ANALYSIS AND IMPACTS REVISED REGULATING RESERVE REQUIREMENTS Since completing the original study, Idaho Power has incorporated substantive changes to its approach for estimating regulating reserve requirements. Because regulating reserve requirements are the basis for determining the cost of integrating wind, the revised estimation techniques warrant further discussion. In general, the revisions to the estimation process are related to the simulation of hour-ahead forecasts for system load and wind generation, and the ability of the Vista DSS model to impose regulating reserve requirements dynamically and asymmetrically. As discussed previously in this addendum report, regulating reserve requirements in the original study were input to Vista DSS at a constant and bi-directionallevel. In this approach, the amount of regulating reserve the model was forced to carry was independent of system load and level of wind production. The reserve level carried was determined simply by calculating the amount ofbi-directional regulating reserve covering 98% of the variability in both load and load net wind. In the revised approach discussed here, the regulating reserve level for a given hour is determined as a direct function of the projected load and wind generation for that hour. This results in decreased regulating reserve requirements, and consequently lower wind integration costs. A detailed description of the previous method used to calculate reserve requirements can be found in Section 6 of the original study report. A discussion of the revised process developed in work leading up to the June 20, 2007 public workshop follows. LOAD REG-Up/REG-DOWN The first step in the investigation was to develop a process for representing system regulating reserve requirements associated with variability and uncertainty in load alone. The objective is to estimate the amount of regulating reserve needed to cover deviations in 10-minute instantaneous measurements ofload from hourly average load as forecast on an hour-ahead basis. This is based on the premise that the hour-ahead forecast load dictates the generation scheduling (and market activity) for the next operating hour, and deviations from that forecast load must be managed by increasing or decreasing the output of other generating units. Deviations from the hour-ahead load forecast occur because (i) the 10-minute instantaneous load data are variable, and (ii) the hour-ahead forecast is in error. That is, even if the load forecast is exactly correct as evaluated on an hourly average basis, deviations occur simply because the instantaneous load varies above and below the observed average during the course of an operating hour. Conversely, even if instantaneous load somehow remains constant during the course of an operating hour, deviations occur unless the load forecast is exactly correct. To mimic load forecast error, a random error was applied to each hourly average load. For example, consider an hour for which the six 10-minute instantaneous load readings average to 2 000 MW. If the randomly selected error for the hour is -00%, then thehour-ahead forecast hourly average load is considered to be 1 980 MW. The randomlyselected errors follow a normal distribution having a mean of 0., standard deviation of 1.3%, and an average absolute error of 1.0%. The distribution of the 8 760 errors usedfor calendar year 2005 is shown in Figure 1 below. 350 100 300 250 200 !:: ::J4b .t 150 ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ .~ ~ ~'r ".J 'I,!Ii. ri, '? e!.'11 ~ " ,, 1::'1 I(,e!. ri,'P co ~.J ~ ri,' 'l,v "q, '" t:)" W '0"' '1r 1(,'" \Y' 'Cf' ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ n. n. n. ~ ~ ~ ~ ~ Bin Figure 1.Distribution of errors for hour-ahead forecast load Taking into account the variability in the 10-minute load measurements and the elTor inthe hour-ahead load forecast, Idaho Power must schedule resources with an expectation of how much higher or lower system load might be during the actual operating hour relative to forecast hourly average system load. Using calendar year 2005 load data, anhourly time series representing the hour-ahead load forecast was devised using the process desclibed above for calculating forecast load. The actual 10-minute load data were then compared to the hourly forecast loads. The data were binned according to the forecast load, resulting in the curves shown in Figure 2. 3000. 1000. LOAD --- REG UP . '----- 2800. 2600. 2400.= 0.000068x' + 0.823961. + 184.725391 = 0.999724 ~ 2200. .:! 2000. ~ 1800. -------- ------- --- ti 1600.1% PROBABILITY ABOVE 1400. 1200. NOTE: FORECAST NEXT-HOUR LOAD BASED ON ACTUAL AVERAGE LOAD PLUS RANDOM ERROR. RANDOM ERROR FROM NORMAL DISTRIBUTION HAVING AVERAGE 0.0%. STANDARD DEVIATION 1.3%, AND AVERAGE ABSOLUTE ERROR 1.0%, 800. 800.1000.1200.1400.1600.1800.2000.2200.2400.2600.2800.3000. forecast next-hour load (MW) Figure 2.Actual operating hour load YS. forecast hour-ahead load Based on a simple empirical analysis of the data, the probability of observing a 10-minute system load measurement exceeding the upper fitted line above is I %. Similarly, the probability of a 10-minute system load measurement less than the lower fitted line is 1%. For example, given an hour-ahead load forecast of 2,400 MW, there is a 1% probability of observing a 10-minute load measurement exceeding 2 554 MW and a I % probability of observing a 10-minute load less than 2 252 MW. In other words, the power system dispatcher can have 98% confidence that the system load (at least as measured at 10- minute intervals) will remain between 2 252 MW and 2 554 MW. Therefore (neglecting any interaction with wind), to assure CPS2 compliance of 98%, Idaho Power should allow for 154 MW of reg-up reserve to cover possible upward movement in load (relative to forecast load) and 148 MW ofreg-down reserve to cover possible downward movement in load. It is important to note that this explanation has yet to consider the LIO band. That is, the above-noted 154 MW of reg-up reserve was derived based on the assumption that generation/load imbalances need to be reconciled to a balanced position. However, CPS2 regulations require only that imbalances are reduced to within a utility LIO level, which for Idaho Power is 38.52 MW. Consideration of the LIO level will be discussed in a later step. WIND REG-Up/REG-DOWN In this case, the objective is to estimate how much regulating reserve is needed to cover deviations in 10-minute instantaneous measurements of wind generation versus hourly average wind generation as forecast on an hour-ahead basis. As with load, deviations from the forecast wind generation must be managed by increasing or decreasing the output of other generating units. Also similar to load, deviations between instantaneous wind generation observed during the course of an operating hour and the associated hour- ahead hourly average wind generation forecast come about for two reasons: ) the instantaneous wind generation during the course of the operating hour varies above and below the hourly average, and ii.) the hour-ahead hourly average wind forecast is in error. With regard to wind generation, an hour-ahead forecasting process can be simulated through the use of an autoregressive time-series model expressing hourly average wind generation for an operating hour as a function of the six 10-minute readings occurring 65 , 85 , 105, & 115 minutes prior to the start of the operating hour (e.g. wind generation forecast for 9:00-10:00 is a function of instantaneous wind at 7:55 7:45 7:35 7:25, 7:15, & 7:05). The hour-ahead wind forecasting utilized in the study was further refined through the derivation of season-specific forecast models (i.e. separate winter spring, summer, and fall forecast models). The autoregressive forecast technique is a marked improvement over the persistence forecast used in the original analysis (February 2007 report), where the hourly average wind generation was forecast to persist from the wind generation occurring at 65 minutes prior to the start ofthe operating hour. The fundamental question for the power system dispatcher is similar to the load alone case - given an hourly average wind forecast, how much higher or lower might system wind generation be during the actual operating hour? Using calendar year 2005 wind data, the following curves were derived. It should be noted that Figure 3 is provided for illustration purposes only. The actual seasonal hour-ahead wind forecast models differed slightly from the example presented below: Given an hour-ahead wind forecast of 00 MW, there is a 1 % probability of observing a 10-minute wind measurement exceeding 428 MWand a 1 % probability of observing a 10-minute wind measurement less than 162 MW In other words, Idaho Power can have 98% confidence that the wind generation (at least as measured at 10-minute intervals) will remain between 162 MWand 428 MW Therefore, to assure CPS2 compliance of 98%, Idaho Power should allow for 138 MW of reg-up reserve to cover possible downward movement in wind generation (relative to forecast wind) and 128 MWofreg-down reserve to cover possible upward movement in wind generation. Consideration of the L 10 band, which is essentially the extent to which loads and resources can be out of balance without constituting a control performance violation, will be reserved for the following discussion on total regulating reserve requirement. 600. 100. 1% PROBABILITY ABOVE 500. 400. :! 300. 'Ji~ 200. ';;; 1:; NOTE: FORECAST NEXT-HOUR WIND BASED ON AUTOREGRESSIVE TIME-SERIES MODEL INCORPORATING OBSERVED 10-MINUTE WIND READINGS AT 65, 75, 85, 95, 105, & 115 MINUTES PRIOR TO THE START OF THE OPERATING HOUR. 100. 100.200.300. forecast next-hour wind (MW) 400.500.600. Figure 3.Actual operating hour wind VS, forecast hour-ahead wind TOTAL REG-UP REG-DOWN In typical real-time operations, load and wind generation share the characteristic of being largely outside the control of the electrical load serving entity. Other factors in the load/resource balance are either dispatchable or highly predictable in the time frame of real-time operations. For example, generation at a run-of-river hydroelectric plant on an hour-ahead basis is very predictable, barring unforeseen outages related to equipment failure. Because of the similarity between load and wind with respect to real-time operations, it is useful to couple their separate regulation components into a single total regulating reserve level. It is understood that because of interaction between load and wind, a straight arithmetic sum of the separate components results in reserve levels that are inappropriately high. Using the examples given in each section, the load forecast of 2,400 MW requires 154 MW of reg-up reserve and the wind forecast of300 MW requires 138 MW of reg-up reserve. It would likely be excessively conservative for a system to carry reserve equal to the sum of these components (292 MW). For modeling purposes Idaho Power combined the components through a root-sum-square operation. Using the same example, the total reg-up reserve calculated in this manner would equal: Total Reg-up = SQRT ((154 MW)2 + (138 MW)J = 207 MW At this point, the LIO band can be applied, resulting in a reg-up reserve requirement of 168 MW (207 MW - 38.52 MW). This process can be followed to generate an hourly regulating reserve time-series for each study year and wind penetration level that is a dynamic and asymmetric function of hour-ahead forecast load and wind. Table 3 provides average regulating reserve levels calculated by the above described process where the load reg-up reserve and load reg-down reserve columns are averages for the load alone (flat wind) cases and the load net wind columns are for the root-sum-square combined load and wind. Table 3.Average levels of regulating reserve by wind penetration level Wind Load Load Reg-Penetration Reg-up down Load Net Wind Load Net WindLevel (MW) (MW) (MW) Reg-up (MW) Reg-down (MW) 300 600 900 51.4 51.4 51.4 49. 49. 49. 66. 87.4 109.4 74. 104. 140. REVISED MODEL INPUTS As part of its continuing study of wind integration following the March 15 , 2007 public workshop, Idaho Power recognized six primary modeling revisions expected to improve the accuracy ofthe study results. The incorporation of these changes produced an overall decline in the estimated cost to integrate wind generation. In this section, the individual revisions and their associated cost impact are described. It is emphasized that interdependence between the modeling revisions make it difficult to isolate the cost impact attributable to an individual modification, consequently the cost impacts presented in this section should be considered approximate. The estimated cost impacts can be considered indicative of the effect of the utilized modeling revisions in a relative sense. A summary of the Vista DSS modeling changes used in the updated wind integration cost detennination are summarized below: In the original study, wholesale electricity markets to the west and east of Idaho Power s system were respectively assigned historical observed prices reported for the Mid-Columbia (Mid-C) and Palo Verde (PV) electricity markets. Price differences between these two markets caused the Vista DSS model to consistently take advantage of arbitrage opportunities across Idaho Power system. While in practice these opportunities do occur on occasion, review of the modeling results indicated that Vista DSS's arbitrage activity was far too frequent and preferential to the flat wind case. Therefore, the arbitrage opportunity was removed by replacing the PV price data for the wholesale market to the east with Mid-C prices. Thus, the two markets available to the model contained equivalent price data, thereby removing the arbitrage opportunity across Idaho Power system. The elimination of this arbitrage opportunity had a significant impact and resulted in a reduction of the wind integration cost of approximately $1.00/MWh. In the original study, regulating reserves were imposed by the Vista DSS model at a constant and bi-directionallevel. Since completion ofthe original study, Synexus Global has incorporated the ability to input asymmetric reserve requirements into the model. This new feature coupled with the ability to specify dynamic reserves on an hourly basis has allowed the assignment of varying levels of reg-up and reg-down regulating reserves on an hourly basis. These reserve levels are considered to more realistically simulate the connection between reserve obligation and load/wind conditions than the constant, bi-directional reserves used in the original study. The impact of this change was also significant and reduced the wind integration cost by approximately $1.00/MWh. As discussed in Section 3 of this addendum report, it was suggested in the workshop process that the reserve estimation methodology of the original study double-counted" the amount of necessary reserves. To remove the potential for double-counting, Idaho Power excluded high-resolution load and wind data from the reserve estimation process, and instead based its estimates on the amount of reserve necessary to cover variability solely in the instantaneous 10-minute data for load and wind. This change had a small impact and reduced the wind integration cost by approximately $0.1 O/MWh. In the original study, the flat wind case was constructed such that wind generation was input at constant levels by day. The 24 hourly wind generation levels were set equal to each other, and equal to the average generation for the day. However because average light load generation in the synthetic wind time series exceeded heavy load, the value of the flat wind case was favorably biased prior to consideration of any effects related to wind integration. To remove this bias, the design of the flat wind case was modified such that wind generation was separated into flat blocks for both heavy load and light load hours. This change resulted in lowering the wind integration cost by approximately $0.25/MWh. The distribution of wind projects used to model the 300 MW penetration level was updated to reflect selection ofthe Elkhorn Valley Wind Project (Horizon) in northeastern Oregon in Idaho Power s recently concluded wind RFP. The 300 MW penetration level was amended to include the Elkhorn Valley project and to move the southern Idaho Cotterel site to higher penetration levels. Overall 100 MW from the Elkhorn project was added to the 300 MW scenario and 100 MW from the Cotterel site was removed. In addition, the sizes of several individual extraction points were adjusted to address scaling issues between the 300, 600 and 900 MW penetration levels. This change provided a greater diversification of the wind resource and resulted in a reduction in the wind integration cost of approximately $0.20/MWh. The wind forecasting methodology used in the model was improved by utilizing a seasonal, autoregressive method rather than a persistence forecast taken at 65 minutes before the hour. This change reduced the wind integration cost by approximately $0.25/MWh. SENSITIVITY RUNS In addition to the modifications mentioned above, Idaho Power perfonned additional Vista DSS simulations for the purpose of exploring the sensitivity of the results to two issues raised at the public workshops: 1) the use of actual Mid-C market prices as recorded for the three study years, and 2) using the Jim Bridger coal-fired generating facility for reg-down reserves. The selection ofthese factors for sensitivity testing is a product of the workshop process where considerable discussion was focused on the market price assumptions used in the Vista DSS modeling and the practicality of providing regulating reserve with resources other than the Hells Canyon Complex. These sensitivity results are considered exploratory, and are not included in Section 5. The updated study results provided in Section 5 are based on a feasible modification of current operating practices for Idaho Power s generating resources. MARKET PRICE ASSUMPTIONS In the updated work since the completion of the original study, Idaho Power selected Mid-C prices recorded for calendar year 2006 for modeling. The use of actual prices in the original study received considerable attention at the workshops, particularly with regard to calendar year 2000 prices. As a consequence, it was decided for the updated work to input prices observed for calendar year 2006 for all three study years. The following table provides the results of the sensitivity test in which the actual market prices observed in the three study years were restored, with all other modeling revisions discussed in this section implemented. The results of this analysis are presented in Table 4 and Figure 4. Table 4,Modeling results as a percentage of market prices using actual market prices and no Bridger coal plant for regulating reserves Study Year 300MW 14,3% 14.1 %8% 11 .1% 10,8.7% 11.9% 900MW A VG Market Price $27, $132. $58. 1998 2000 2005 Average 19. 17.2% 15. 17. PURPA Wind Contract: Wind Integration Cost: Adjusted PURPA Wind Contract: $62.40 $5.43 $56. $62.40 $7.41 $54. $62.40 $10.75 $51. Interpolated Wind Integration Cost at 492 MW:$6.70 25. 20.0% . 15. 10. -- 1 998 -- 2000 -.- 20051 I 0.0%- 300MW 600MW 900MW Figure 4.Modeling results as a percentage of market prices using actual market prices and no Bridger coal plant for regulating reserves JIM BRIDGER REGULATING RESERVE The second sensitivity analysis illustrates the impact of using thennal units to provide reg-down reserves. To incorporate the reg-down capability assumed to be provided by Jim Bridger, the thennal units were not actually modeled providing reserves; rather the reserve requirement was reduced on the Hells Canyon Complex. It is important to note that cycling these units would result in increased maintenance costs which are difficult to quantify and are not included in the results presented in Table 5 and Figure 5. Additional discussion ofthe use of Jim Bridger for this purpose is included in response to item 6 in Section 3 of this addendum report. All other modeling revisions discussed previously in this section were implemented for this analysis. Table 5,Modeling results using 2006 market prices and the Bridger coal plant for regulating reserves Study Year 300MW21.9% 19.4%8% 4.8% 6.0% 10.1 % 900M W A VG 2006 Price $44.44 $44.44 $44.44 1998 2000 2005 Average 22. 12. 11. 15.4% PURPA Wind Contract: Wind Integration Cost: Adjusted PURPA Wind Contract: $62.40 $4. $57.42 $62.40 $6. $56. $62.40 $9. $52. Interpolated Wind Integration Cost at 492 MW:$5. 25. 20.0% - 15. I 5. --+- 1 998 --- 2000 -- 2005 I 10. 300MW 600MW 900M W Figure 5.Modeling results using 2006 market prices and the Bridger coal plant for regulating reserves The detennination of required regulating reserves is a major component of Idaho Power wind integration study and this section has provided details ofthe work completed in the time since the original study was published. Section 5 presents the updated study results in regards to the cost of integrating wind generation on Idaho Power s system. Section 5 UPDATED STUDY RESULTS In the updated study, Idaho Power incorporated the six primary modeling revisions described in the previous section to derive a new estimated cost to integrate wind generation. Table 6 and Figure 6 below show the results of the updated analysis and integration costs by study year as a percentage of 2006 market prices. Table 6 also summarizes the average wind integration cost by penetration level. Table 6.Updated modeling results as a percentage of 2006 market prices without use of the Bridger coal plant for regulating reserves study Year 300MW 15.1% 15.7.4% 11.0% 17.2% 14.7% 900MW A VG 2006 Price PURPA Wind Contract: Wind Integration Cast: Adjusted PURPA Wind Contract: $62.40 $5. $56. $62.40 $9. $53, 18.$44.44 12.4%$44.44 14.$44.44 15. $62.40 $9. $53, . . 1998 --- 2000 ~ 2005 900M W 1998 2000 2005 Average I 25. 20. , 15.0%- 10. 0% 300MW 600MW Figure 6,Updated study results as a percentage of 2006 market prices without use of the Bridger coal plant for regulating reserves Figure 7 shows Idaho Power s estimate of the cost it will incur (in $/MWh) to accommodate wind generation for a range of penetration levels for both the initial and updated studies. Study results suggest that as wind penetration levels increase, the resulting reserve requirements at higher wind penetration levels will ultimately overwhelm the current system s reserve capacity. Idaho Power believes that given current technology and market structure, the upper limit on the amount of wind generation that can be integrated on its current system lies between 600 and 900 MW. the time the original study was completed, Idaho Power had signed contracts or commitments to develop 384 MW of wind generation. It should be noted that 600 MW of wind generation corresponds to a penetration level of approximately 19% (according to the convention of expressing penetration level as percentage of peak system load), which is an ambitious level of development by current standards. To arrive at a single cost estimate to account for system impacts due to wind integration Idaho Power proposed using the estimated cost at the midpoint of wind development between the current committed level of384 MW and 600 MW utilizing current PURPA rates to convert trom percent of market to a dollar amount. In the original study, the cost at this midpoint level (492 MW) was estimated via a 3rd order line fitted to the original study results. This cost ($10. 72/MWh) is illustrated in Figure 7 below. In the updated study, the cost at this midpoint level (492 MW) is estimated via linear interpolation between the updated study results at 300MW and 600MW. This process results in a wind integration cost of $7.92/MWh and is also illustrated in Figure 7 below. $14,. Initial Study Average $12, . Updated Study Averoge $10.72/MWh ". 'O',""w..',$10, $8,00 "" --"... n--'-o-....-,,..'o.,--...--.--.,--.. $7,92/MWh 00.--' ""'0""'0"....0-- - .----.-- :-.... $6, $4, $2, :;; - ..so ..... :;;:;; , $0, 300 WIND PENETRATION LEVEL 600 Figure 7. Idaho Power s updated cost estimate (in $/MWh) for wind integration Idaho Power supports society s desire to have future energy supplies that come from clean, renewable energy sources. Renewable, emission-free electricity production has been a part of our company s history for over 90 years and wind power will be an important part of continuing that legacy. One thing is for certain - the cost of wind integration will change over time. Regional wind integration efforts, improvements in wind forecasting, regulatory changes and actual "hands-on" experience will all have an impact on the cost of integrating wind energy. In recognition ofthis fact, Idaho Power will continue to evaluate wind integration costs as model assumptions change and new and improved study methods are developed.