XII. Demand Forecast

A. Summary of Demand Forecast

The demand forecast addresses PG&E's forecast of on-system demand, off-system deliveries, and total throughput for the 2004 period. This annual demand forecast is then used for cost allocation and ratemaking purposes. For 2004, PG&E forecasts total throughput of 2273 MDth/d, or on-system demand of 2054 MDth/d⁶⁰ and 219 MDth/d of off-system delivery. This is shown in Table 13-1 of Exhibit 1.

There are four main market segments that make-up on-system end-use demand. These four segments are core, noncore industrial, electric generation, and wholesale. The core is composed of mainly residential and commercial customers. The noncore industrial consists of large customers who are engaged in industrial activities and who qualify for service under the G-NT rate schedule, and noncore natural gas vehicle customers. The electric generation segment consists of generators and cogeneration facilities that use natural gas to make electricity. The wholesale segment consists of municipal or private entities that purchase transportation-only service for gas for resale through non-PG&E distribution systems.

PG&E's 2004 forecast of demand for these four market segments are as follows in MDth/d: core - 800; noncore industrial - 450; wholesale - 11; electric generation and cogeneration - 745. PG&E notes that on-system demand has declined about 3.6% per year over the 2001-2004 period.

PG&E's electric generation demand forecast assumes that the decline in electric generation demand from 2001 to 2002 is likely to continue through 2004 because new combined cycle plants⁶¹ will be added much faster than the growth in electricity demand. Since these newer plants are more efficient, gas usage should decrease.

Various parties have raised the issue of whether the electric generation demand forecast should be revised due to the lower number of gas-fired combined cycle power plants that are expected to be built. Due to the lower number of new plants, some parties contend that this will result in an increase of gas consumption at the less-efficient gas-fired generating plants, which should raise the electric generation demand forecast. PG&E asserts that if the electric generation demand forecast is updated, other forecasts should be updated as well.

The off-system delivery forecast is for gas that is transported through PG&E's backbone transmission system and delivered to SoCalGas' transmission system for final delivery to customers in the Southern California market. These off-system customers are generally looking to buy gas that is produced in Canada or in Northern California. Some of these off-system customers have firm G-XF contracts for Redwood path capacity that were signed when Line 401 was built. PG&E forecasts 2004 off-system demand of 219 MDth/d.

PG&E proposes to adjust the backbone throughput to account for the fact that some backbone contracts have a rate higher than the annual firm rate⁶² and some contracts have a lower rate. The annual firm rate is increased by 20% to determine the seasonal firm rate and the as-available rate. In order to account for the higher revenues from the seasonal and as-available service and lower revenues from discounted contracts, the backbone load factor adjustment is added to the forecast of throughput prior to calculating the load factor. The backbone load factor adjustment is the amount of throughput paying the higher rate multiplied by the percentage over the annual rate, less the throughput paying a lower rate multiplied by the percentage over the annual rate, less the throughput paying a lower rate multiplied by the percentage discount to the annual rate. PG&E estimates that the net increase to backbone throughput is 45.9 MDth/d.

B. Position of the Parties

1. CCC/Calpine

CCC/Calpine assert that PG&E has materially understated its proposed 2004 electric generation (EG) throughput forecast. The Commission should instead adopt CCC/Calpine witness Beach's proposed EG throughput forecast of 665 MDth per day.

PG&E forecasted gas demand for electric generation by using the MarketBuilder model of the Western Electricity Coordinating Council (WECC) electricity market in 2004, assuming average hydro conditions and all known new resources expected to come on line before or during the forecast period. CCC/Calpine note that all of the parties who took a position on the throughput forecasts, except for PG&E, agree that PG&E's electric generation throughput forecast is too low.

Beach mentioned several factors why PG&E underforecasted EG gas demand. The first factor is because EG gas demand increases more in a dry year than it decreases in a year that is comparably wet. CCC/Calpine assert that because EG throughput in wet years does not decline symmetrically with increases in EG throughput during dry years, use of average throughput systematically biases throughput forecasts downward.

The second factor is that new resource additions, i.e. renewables, cogeneration, more efficient combined cycle plants, and demand-side management, tend to reduce gas demand for electric generation. Although PG&E used a relatively up to date list of new resources and on-line dates, there has been significant slippage in the on-line dates for many of these projects. These delays will remove 4,531 MW of new resources that PG&E's model assumed would be available in 2004. This represents 71% of the 6,355 MW of new generation in the WECC that PG&E expected to come on line in 2004, as listed in Table 13-4 of Exhibit 1. NCGC had asked PG&E to rerun its model under the assumption that 100% of PG&E's assumed new resources are delayed by one year. This sensitivity run resulted in a 14% increase in EG demand in 2004.

CCC/Calpine assert that since the record in this case shows that known delays in new plant completions already exceed 70% of the delay that PG&E itself modeled in the sensitivity run for NCGC, new plant delays can be expected to raise PG&E's EG loads in 2004 by 10% (71% of 14%) above the levels shown in PG&E's forecast.

Beach's testimony shows that PG&E's reliance on average hydro conditions has resulted in 10% to 12% under-forecasts of actual EG demand. The combination of the known plant delays, PG&E's own sensitivity run on the impact of a one-year delay in new plants, and the under-forecasts due to reliance on average hydro conditions, all support Beach's EG throughput forecast of an additional 15% in EG demand. For these reasons, CCC/Calpine recommend that the Commission adopt a 2004 EG demand forecast of 665 MDth per day.

PG&E contends that if the forecasting model is to be updated, that all elements of the model be updated rather than just selected inputs. CCC/Calpine assert that PG&E has already had the opportunity to present a forecast, and that parties responded. The Commission should not allow PG&E to redo all of its modeling assumptions. Instead, the Commission should decide this issue based on the record that has been developed concerning PG&E's forecasts, other parties' analyses of those forecasts, and the utility's responses to those critiques.

CCC/Calpine also assert that PG&E understated its forecast of off-system backbone level throughput. PG&E contends that off-system backbone-level throughput will decrease below the historical level of 298 MDth per day experienced during the original Gas Accord period. PG&E proposes the following downward adjustments to historical off-system throughput: (1) reduce G-XF contract volumes by 35 MDth/day; and (2) reduce Baja off-system flows by 44 MDth/d.

CCC/Calpine contend that both of PG&E's adjustments are incorrect. PG&E's adjustment for G-XF volumes is attributable to the expiration of certain existing G-XF contracts. Although CCC/Calpine do not dispute that off-system G-XF volumes may have expired, this does not necessarily translate into an overall reduction in off-system throughput. CCC/Calpine contend that it does not follow that because certain G-XF contracts are terminating, that they will not utilize short-term off system service, which is what PG&E assumes.

CCC/Calpine assert that the demand for off-system service does not depend on whether some customers are signing or terminating long-term contracts for such service. What matters is whether there is an economic incentive to take off-system service, on either a short or long term basis. If the price for and supply of off-system service are at least as high as in the past, then the demand for the service should also be as robust as it has been historically. Thus, it is reasonable to forecast that the demand for Redwood off-system service in 2004 will at least reach historical levels.

With respect to demand for Baja off-system service, PG&E asserts that this will shrink to zero. However, the demand for Baja off-system service is driven by the spread in prices at Topock between gas going into the PG&E system, and gas flowing into the SoCalGas system. PG&E witness Wilson recognizes that the price spread at Topock is the result of capacity constraints moving gas onto the SoCalGas system, which result in a much higher Topock into SoCalGas price. CCC/Calpine assert that these constraints will not change in 2004 compared to 2002. Thus, the price spreads at Topock should persist, resulting in similar levels of Baja off-system throughput to those experienced in the past.

CCC/Calpine recommend that PG&E's forecast of off-system demand be revised upwards to levels consistent with historical experience. That is, an off-system throughput forecast of 298 MDth/d should be adopted.

2. CMTA

CMTA contends that PG&E's proposed backbone rates are based on a gas throughput forecast which is too low. CMTA contends that PG&E understated the throughput by underestimating forecasted gas demand by electric generators, and by using off-system throughput that falls below historical levels.

PG&E assumes that Redwood off-system throughput will decrease. However, this is contrary to PG&E's own observation that demand for Redwood off-system service is likely to remain at historic levels, due to the expected continuation of the historic $0.21 per Dth price spread between Malin and Topock. CMTA contends that it is this price spread, and not the G-XF contract volumes as PG&E assumes, that drives demand in the transportation market. PG&E acknowledges that price spreads drive market demand in its Market Builder model that is used to forecast EG throughput.

3. Duke

Duke recommends that the key elements of PG&E's forecast of EG gas demand be updated to reflect the cancellation and deferral of planned new generation units. Duke points out that PG&E's forecast was developed when many new efficient plants were expected to come on-line in 2003 and 2004. PG&E's EG throughput forecast relied on a list of expected power plant additions that the California Energy Commission (CEC) issued in July 2002, and a similar list prepared by WECC in January 2002. It was expected that these more efficient plants would displace less efficient, older gas-fired plants and, as a result, EG gas throughput would decline.

PG&E forecasted that EG throughput would decrease by 18% in 2004. Since the time PG&E's forecast was made, many plant sponsors have cancelled or deferred their proposed new plants. Duke believes the Commission should have a forecast that reflects the most recent available list of expected power plant additions because of the effect these cancellations will have on the EG demand forecast.

4. Mirant

Mirant notes that it would be prudent for the Commission to consider and take official notice of the most current forecasts of power plant construction and new power plant additions, as those forecasts may affect PG&E's EG gas demand forecasts for 2003 and 2004.

5. NCGC

NCGC witness Pretto testified that the EG throughput forecasted by the MarketBuilder model that PG&E used is sensitive to assumptions about whether or not projected new power plants will come online. PG&E's forecast of EG usage in 2004 is 580 Mdth/d. NCGC asserts that due to plant deferrals and cancellations, 2004 EG throughput could easily be higher because those deferred or cancelled plants are more efficient. If less efficient plants inside PG&E's service territory are utilized at higher load factors, the result could be increased throughput to EG customers.

PG&E provided a sensitivity analysis to the NCGC witness about the impact of plant cancellations or deferrals on PG&E's EG throughput forecast. According to NCGC, the analysis showed that if the plants listed in Table 13-4 of Exhibit 1 were delayed by one year, 2004 EG throughput would be 663 Mdth/d, 14% higher than forecasted by PG&E. If the plants were delayed by two years, the 2004 EG throughput would be 791 Mdth/d or 36% higher than forecasted by PG&E. NCGC asserts that these results illustrate that PG&E's projected EG throughput is highly sensitive to the operative dates of the plants listed in Table 13-4 of Exhibit 1. Due to the sensitivity, NCGC recommends that PG&E's EG throughput forecast be updated as near as possible to the issuance of a final decision in this proceeding to reflect the status of new power plants at that time.

NCGC contends that it would be easy to accomplish an update of PG&E's proposed EG throughput forecast. The CEC uses a version of the MarketBuilder model called the North American Regional Gas Model (NARG). Using NARG, the CEC is developing a new forecast of EG throughput. The CEC's throughput forecast was scheduled to be released on May 23, 2003. Due to the similarities between the two models, the CEC's forecast should be used to update PG&E's EG throughput forecast.

Instead of updating the EG throughput to consider just actual plant additions, PG&E urges a comprehensive update, using the most recent information from the same data sources that were used in the prepared testimony. NCGC asserts that a comprehensive update is unnecessary. If a comprehensive update of PG&E's throughput forecast is done, NCGC recommends that it focus on EG throughput, off-system backbone throughput, and as-available backbone throughput. The as-available backbone throughput should take into account the most recent available information on the mix of backbone services, i.e., firm service and as-available service, that shippers are taking on the PG&E system today.

6. TURN

TURN disputes PG&E's forecast of 745 Mdth/d of load for electric generation. TURN contends that PG&E's forecast underestimates electric generation throughput by assuming that 56,000 MW of new combined cycle generation will come online in 2004 throughout the western United States.

TURN witness Marcus disagrees with PG&E's throughput estimate for EG demand. PG&E's projected decline in EG throughput from 2001 to 2004 presumes the construction and availability of numerous out-of-state combined cycle power plants, many of which have been cancelled. TURN has identified over 4500 MW of generation that has been cancelled or delayed. Marcus also noted that 2004 is likely to be the lowest year for EG gas demand in California in the next few years. Marcus suggests basing the EG demand estimate on several future years, rather than PG&E's depressed and likely inaccurate projection for 2004.

TURN recommends that the Commission require PG&E to rerun its EG forecast with more recent information regarding project cancellations and delays, and report the results for 2004-2007.

7. PG&E

PG&E developed forecasts of on-system demand, off-system deliveries, and total throughput for the Gas Accord II 2004 period. These forecasts are used for cost allocation and ratemaking purposes. PG&E asserts that its forecasts are based on careful analysis and are consistent with historical experience and anticipated future conditions.

PG&E points out that since no party took issue with PG&E's proposed core and noncore industrial forecast, the Commission should adopt PG&E's proposed core and noncore industrial forecast for 2004.

PG&E's forecast of gas demand by all cogeneration plants except Crockett was uncontested. PG&E contends that because the output and gas demand of cogeneration plants, except Crockett, are not strongly affected by conditions in the electricity market, its forecast was based on an extrapolation of recorded data. Since no one disputes PG&E's cogeneration forecast, PG&E recommends that its cogeneration forecast be adopted.

PG&E also presented a forecast of gas demand for gas fired EG plants whose production levels are affected by changes in the electricity market. PG&E asserts that its forecast is based on a detailed modeling of the power markets, and should be adopted. PG&E's EG forecast was based on the most current data and assumptions, and no changes to its forecast are needed. However, if the Commission is inclined to have PG&E perform an update to its demand forecasts, PG&E recommends that all input assumptions, and all components of the demand forecast (core, noncore, EG, off-system) be reevaluated. PG&E notes that a selective or piecemeal update would result in a biased forecast that is based on internally inconsistent assumptions.

PG&E points out that the intervenors propose using a higher estimate of EG gas demand because that would lead to lower rates. The CCC/Calpine proposes to raise PG&E's EG forecast by 15 percent because of an alleged average understatement of PG&E's EG demand forecast from prior years. PG&E points out that CCC/Calpine's proposed 15% increase is based on the assertion that hydro conditions in 2004 will be biased toward dry conditions. However, hydro conditions for 2003 are forecasted to be close to average.

PG&E also contends that the accuracy of PG&E's prior EG forecasts are not relevant because it is using a new forecasting method, the MarketBuilder model. PG&E asserts that the evidence demonstrates that this model has not underestimated average EG gas demand.

CAPP contends that PG&E's forecasted load factor of 68.4% for 2004 is too low compared to third party forecasts, such as the CEC, which projected a system utilization for PG&E at 75%. PG&E says this argument is without merit because the CEC forecast of December 2002, which CAPP relied on, was to be corrected and updated in Spring 2003. PG&E says this update was posted on May 28, 2003 on the CEC's website. Figure 14 of that CEC document shows that the CEC's forecast of gas demand for electricity generation (including cogeneration) in 2004 is about 750 MMCF/day or 760 MDth/d. The CEC's new forecast is lower than PG&E's which was 845 MDth/d (580 MDth/d for EG as PG&E defines it and 265 MDth/d for cogeneration). Although the CEC forecast is lower than PG&E's, PG&E is not recommending that it be adopted because it is a preliminary staff forecast not adopted by the CEC, and may be based on a different model.

CCC/Calpine argue that new plant delays alone can be expected to raise PG&E's EG loads in 2004 by 10% above the levels shown in PG&E's EG forecast, and that the Commission should not allow PG&E to re-do all of its modeling assumptions. PG&E asserts that forecasts of electricity demand affect both the schedules of power plant development and EG gas demand forecasts, and the record shows that EG forecasts are sensitive to electricity demand. PG&E says it is only fair that if power plant schedules are to be updated, electricity demand forecasts should also be updated.

Mirant says that Beach pointed out the extent to which PG&E's past EG forecasts using similar models have underestimated actual EG demand. But PG&E says nothing in the record or in fact supports Mirant's claim that PG&E's current EG forecast and its past forecasts were developed using similar models. There is nothing in the record about what models, if any, were used previously by PG&E or how they might be similar to Marketbuilder.

TURN claims that PG&E's forecast underestimates electric generation throughput by assuming that 56,000 MW of new combined cycle generation will come online in 2004 throughout the western United States. PG&E says TURN is incorrect because PG&E's forecast never assumed that 56,000 MW would come on-line, and that it did not use the WECC forecast.

PG&E's forecast of its off-system throughput is the sum of the following: the short-term Baja off-system contracts; the short-term Redwood off-system contracts, and the long-term Redwood off-system contracts. CCC/Calpine contend that PG&E's off-system forecast is too low, and recommends a higher forecast of 298 MDth/d based on the average off-system flow since the beginning of the Gas Accord period. PG&E contends that its off-system forecast is reasonable and should be adopted. PG&E asserts there is no evidence in the record that these terminating G-XF customers will continue to use off-system service.

PG&E points out that TURN proposes that PG&E provide EG throughput forecasts for 2005 to 2007. PG&E does not believe such a forecast is needed to set rates for 2004, and TURN's proposal should be rejected.

C. Discussion

The only demand forecasts that parties take issue with are the EG forecast, and the off-system forecast. These two forecasts affect the throughput amount which is used to calculate the system load factor, which in turn is used to develop rates. A higher demand forecast, all else being equal, will result in a higher system load factor. A lower demand forecast, all else being equal, will result in a lower system load factor. The higher the load factor, the lower the rates will be, because there will be more throughput to allocate the costs. The lower the load factor, the higher the rates will be, because there will be less throughput to allocate the costs.

We turn first to the EG demand forecast. The record has many references to the reduction in the number of new gas-fired EG plants. Some of the parties favor an update of the EG forecast in light of the downturn in new plants, while PG&E favors a comprehensive update of all the relevant variables. We do not believe that an update is needed in light of the record, and updating the forecast at this point would be impractical given the time constraint.

The sensitivity runs that PG&E ran for NCGC show that if the plants listed in Exhibit 1 were delayed by one year, 2004 EG throughput would be 14% higher, or using PG&E's EG forecast, to 663 MDth/d. If the plants were delayed by two years, 2004 EG throughput would be 36% higher, or 791 MDth/d. CCC/Calpine's testimony suggests a 15% increase, which is slightly higher than the one-year sensitivity run.

The record also referred to the CEC's forecast that is part of the integrated energy policy report that was to go to the Governor on November 1, 2003. (8 RT 789.) An earlier version of the forecast was cited by PG&E in its brief, and the latest version of that staff report is dated August 2003 and is entitled "Natural Gas Market Assessment." In Figure 14 at page 46 of that report, the forecast of average daily demand for gas in PG&E's service territory by electricity generation (including cogeneration) for 2004 appears to range from 800 to 900 MDth/d.⁶³ This provides a useful comparison to PG&E's combined forecast EG and cogeneration forecast of 845 MDth/d.⁶⁴ Using the one year sensitivity analysis, a 14% increase of PG&E's EG forecast would result in an increase of 81.2 MDth/d or a total combined EG and cogeneration forecast of 926.2 MDth/d. Based on this information, we will not change PG&E's electric generation and cogeneration forecast.

The next forecast to address is PG&E's off-system delivery forecast of 219 MDth/d. PG&E's forecast of this amount is challenged by CCC/Calpine. CCC/Calpine recommend that the recorded off-system throughput of 298 MDth per day during the Gas Accord period be used instead.

We adopt CCC/Calpine's forecast of off-system throughput. Our reasoning is that PG&E is requesting as part of this proceeding that it be permitted to allow eligible off-system end users to connect directly to PG&E's backbone transmission service. One of the reasons that PG&E gives in support of that proposal is that "Customers outside PG&E's service territory have expressed interest in taking service from PG&E's backbone system...." (Ex. 1, p. 18-6.) As discussed later in this decision, we approve that request. Once this service is in place, off-system throughput should remain at recorded levels or increase, and in the words of PG&E would "enhance the use of PG&E's backbone transmission system." (Ex. 1, p. 18-8.) Accordingly, PG&E shall use the 298 MDth/d in its 2004 demand and throughput forecast for off-system delivery, and in the calculation of its load factor.

No one has raised objections to the other forecasts of demand that are shown in Table 13-1 of Exhibit 1, or to the backbone load factor adjustment..

We adopt the forecasts of demand and throughput and the backbone load factor adjustment that are shown in Table 13-1 of Exhibit 1, as modified by the increase to off-system delivery.