Exhibit Petitioners KJ/LK-4

Southern Loop / Brattleboro Area Study and Synchronous Condenser Installation Report November 5, 2006
L. R. Kirby

Acknowledgements
Special thanks to Bruce Bentley for his contributions to the development of the Southern Loop load growth model, and for his DSM screening analysis which formed the basis for much of the DSM analysis in this study. Special thanks to John Jockell for writing the key automation programs used in the loadflow analysis of this study. Special thanks to Kim Jones for her assistance with the cost and financial analysis of this study.

Table of Contents
I Executive summary...page 4 II Introductionpage 4 III Historical perspective.................................................................................page 6 IV Study objectives and guiding principles..page 7 V Study assumptions, methodology, and criteria.. page 8 VI The 5 root problems that afflict southern Vermontpage 10 VII Possible solutions to the 5 root problems .page 13 VIII Integration of solution components to form a strategic plan..page 19
IX The common thread (a synchronous condenser installation at Stratton)..page 23

X Technical analysis of a synchronous condenser installation (alone)..page 24 XI Capital / O&M / loss costs of a synch condenser installation (alone)page 32 XII Recommendationspage 32 Glossary..page 33 Appendices
A B C D E F G H I J An Explanation of QV and PV Analyses.....page 39 Stability Discussion of Synchronous Condensers at Stratton...page 44 Basic Probability Theory.....page 46 Why Chester-Londonderry No Longer Makes Sense..page 53 A New Probabilistic Reliability Guideline Based on the Equal Slope Criterionpage 58 When and Where Load Growth Modeling in Southern Vermont...........page 69 Overcompensation of Network Systems, and its Relationship to the Southern Looppage 73 The Effect of Low Load Growth but High Load Factor Growth on the Southern Loop...page 76 Derivation of Outage Probabilities for the Southern Looppage 79 Strategic Solution Option Descriptionspage 83

I Executive summary The Southern Loop is a 46 kilovolt (46 kV) subtransmission line owned and operated by CVPS, extending from Bennington to Brattleboro. Over time, this line has been subjected to growing electrical demand, particularly near its middle where it is weakest. Its reliability problems appear as low (or high) voltage, not as equipment overloading, and occur more severely in the winter. In the early 1980s, CVPS applied for a new Londonderry to Chester 46 kV tie line to reinforce the Southern Loop. This application was denied by the PSB and the demand growth since that time has been managed primarily with capacitor additions to prop up the weak voltage and load management efforts to limit peak demand. There are other related area reliability problems including a vulnerability of Brattleboro to line or transformer contingencies, a growing threat that the Southern Loops demand will exceed its capability even with all facilities in service, a critical reliance on Vermont Yankees aged 345/115 kV T4 transformer to support area voltage, and emerging regional reliability issues on the overlying 345 kV transmission system operated by VELCO and NEPOOL. Accordingly, CVPS and VELCO developed a spectrum of 10 potential solution options ranging from those emphasizing transmission solutions to those emphasizing distributed resources such as DG and DSM, and with some that combine these approaches. These options are all capable of solving at least some of the areas problems described above, with varying costs and degrees of effectiveness, as well as varying aesthetic, developmental, and environmental implications. These 10 potential solution options were then studied by a group of affected citizens chosen from the public at large, based on a broad demographic mix of interests and perspectives (business leaders, environmental advocates, energy use experts, health care and emergency rescue providers, and local government officials). This unprecedented public outreach process was administered under the direction of a facilitation consultant known as the STAR Group. The final result of this collaboration is that the original 10 solution options have been narrowed down to only 3 options. A proposed synchronous condenser installation near Stratton, which is the subject of this report, is common to all three remaining solution options. Moreover, the public citizen group has explicitly recommended that the synchronous condenser installation be done as soon as is practicable. By itself, the synchronous condenser will not solve all of the problems cited above, but it will dramatically reduce the vulnerability of the Southern Loop to line or transformer contingencies, and will substantially increase the systems ability to serve the areas growing demand with all facilities in service. The reminder of this report provides a more in-depth account of the areas history, the technical analysis performed, the public review process, and the rationale for a synchronous condenser installation near Stratton. II Introduction The Southern Loop is a 46 kV subtransmission line owned and operated by CVPS, extending from Benningtons Woodford 115/46 kV substation to Brattleboros Vernon 115/46 kV substation (see Figure 1 with 46 kV lines in thin red, 69 kV lines in purple, 115 kV lines in black, and 345 kV lines in thick red). The line is 66 miles in length and supplies the town of Manchester, the Stratton and Bromley ski resorts, various manufacturing plants, and several smaller towns.

VELCO 345 kV Line

Southern Loop

Figure 1

The Southern Vermont systems highest electrical demand occurs in winter and has been as high as 120 Mw, comprising 11% of Vermonts highest demand. This 120 Mw includes approximately 25 Mw each in the Bennington and Brattleboro areas and 70 Mw distributed along the loop itself. Much of this demand is concentrated around its weakest point, in the vicinity of Stratton and Bromley, near the middle of the loop. As would be expected in an area having seasonally-dependent economic activity, the summer demand of the 46 kV loop is significantly lower, is less concentrated near its middle, and is of less concern than the winter demand. It is normally operated in a network configuration, and tends to be voltage-constrained, but not thermally constrained. Its pronounced weakness (i.e. its low short circuit strength) may cause excessively low or high voltage during line contingencies that disrupt the main path. This weakness also inhibits the effectiveness of its protection and relay systems. There is virtually no local generation connected to Southern Vermonts 46 kV and 115 kV transmission systems. The Vermont Yankee nuclear plant is more closely associated with the high-voltage transmission system serving all of New England, than with the lower-voltage local systems. There is a scarcity of local generation because there are no significant water sources for hydro plants, as in other parts of Vermont. Moreover, there is no gas distribution system and there are stringent environmental 1 laws which have historically made fuel-burning generators uneconomic in this region . Also, there is little opportunity for CVPS manure-to-methane Cow Power generators because of the lack of large farms in the area. There have been some initiatives involving wind power, but these have all been withdrawn.

Fuel-burning generators in Vermont are often subject to permit conditions that restrict the number of hours they are allowed to run (due to emissions). From the perspective of a project developer, such restrictions impact the economic viability of the project and may make it valuable only if peaking generation is required, or in other special circumstances. 5

There is also a 69 kV line that supplies the Brattleboro area, running north from the Vernon Road Substation. The 46 kV and 69 kV systems that supply Southern Vermont are themselves fed by 115 kV lines terminating at Woodford Road Substation near Bennington and at Vernon Road Substation near Brattleboro. Woodford Road Substation is fed by two 115 kV lines, one from Hoosick Substation in New York and one from Adams Substation in Massachusetts. The redundancy of the Woodford 115 kV supply makes loss of either line fairly inconsequential. Vernon Road Substation is fed by a single 115 kV line from Chestnut Hill Substation in New Hampshire, denoted the N-186 line, which may be seen just to the right of the Vermont Yankee Substation in Figure 1. The lack of redundancy of the 115 kV supply to Vernon Road poses a critical weakness that will be explained in more detail later in this report. At this point, the reader need only be concerned with where these systems are, not how they interact; their interaction will become apparent as this report progresses. III Historical perspective The issues concerning the Southern Loop, are best understood in the context of its recent history. Throughout the 1960s and 70s the expansion of this regions ski resorts led to rapidly-growing electric loads. This growth occurred in facilities directly owned by the ski resorts (i.e. lifts, snowmaking, and condominiums) and in supporting facilities (i.e. ski shops, hotels, restaurants, and the nearby homes of people working in the resort industry). Much of the growth occurred near the middle of the 46 kV line, in and around the Stratton resort. By the early 1980s CVPS had become concerned that the 46 kV Southern Loop was reaching its loadserving limit and that remediation was required to avoid voltage instability during contingencies at higher loads. In 1983, CVPS filed a 248 application to build a 46 kV line from Chester to Londonderry, in order to reinforce the middle of the loop. After a long and contentious hearing process, Vermonts Public Service Board rejected the application, citing several weaknesses in CVPS argument. Since that time, additional load growth on the Southern Loop has been offset by adding dispatchable shunt capacitors along the line, and by demand side management (DSM) including interruptible contracts for the ski resorts, fuel switching, load control, and conservation programs. Efficiency Vermont (EVT) has taken over some of these DSM efforts since the year 2000. Additionally a second 115/46 kV transformer was added in 1995 at the Woodford Road Substation to share the demand with the original transformer. However, these two strategies (capacitors and DSM) do have their limits, and it appears that the point has been reached where current efforts can no longer be counted on to manage the reliability of this older infrastructure. Aggressive DSM programs on the Southern Loop for the past 20 years have been successful, but have now used up many of the available opportunities. Figure 2 is a graphical summary of added DSM savings from 1991-2004. Despite continued effort, the new savings are in a gradual decline. New demand does of course provide new DSM opportunities, and while these emerge at a rate that is too slow to offset the growing threat to system reliability, they could defer capacity upgrades to serve future load growth. Accordingly, CVPS is considering how DSM can be cost-effectively developed to help solve the Southern Loops problems, even though there is not enough DSM available to solve the problems by this method alone.

DSM Energy Savings SL

10,000 9,000 8,000 7,000 6,000 5,000 4,000 3,000 2,000 1,000 -

The use of capacitors to mitigate the effects of new demand growth has also run out of gas but for reasons that are more technical. Simply stated, adding ever more capacitors to meet ever higher demand will indeed prop up a diminished voltage to the desired level, but with increasing difficulty for the operating personnel who must manage their dispatch while the systems demand changes from hour to hour. The oversight of so many capacitors is simply not practical. Furthermore, this heavy reliance on capacitors (sometimes referred to as overcompensation) makes the system more vulnerable to a blackout if something significant goes wrong, such as the loss of a nearby transmission line. This problem is explained in greater detail in Appendix G. Another important historic trend on the Southern Loop is that of an increasing demand factor. Although demand factor sounds complicated, it is simply the average demand of a customer or system, divided by the highest (i.e. peak) demand of that same customer or system. On the Southern Loop, two of the DSM efforts (load control and special contracts) have had the effect of shifting demand from peak-demand hours to low-demand hours. This has kept the peak-demand hours from having their demand grow very much. But it has encouraged the low-demand hours to decline in number and the high-demand hours to grow in number. This trend increases demand factor. Appendix H explains this concept in more detail. The Southern Loop cannot adequately serve more than about half of its peak demand if it suffers the loss of an important facility, such as a nearby transmission line or transformer. Therefore, the more hours that become high-demand hours, the more time the Southern Loop is exposed to this danger, even though its peak load is growing only slowly or not growing at all. Appendix H explains this concept graphically if the reader wishes to explore it further. These various trends have caused or contributed to a handful of distinct and difficult problems that we now face in Southern Vermont. The next section explains what these problems are, and more about how they have arisen. IV Study objectives and guiding principles When asked to describe their main objectives in a planning study, most utility planners will mention acceptable reliability at the lowest possible cost or words to that effect. CVPS agrees with this viewpoint but hastens to add that the quantification of acceptable reliability involves subtle distinctions that must be clearly understood. First, it is generally acknowledged that the larger the system (in terms of voltage, capacity, or load served), the higher its reliability should be. 765 kV transmission facilities serving large regions or metro areas should logically be held to a higher reliability standard than should 34.5 kV 7

Annual MWH

EVT MWH CV MWH

Figure 2

1990

1991

1992

1993

1994

1995

1996

1997

1998

1999

2000

2001

2002

2003

2004

2005

subtransmission facilities serving small localities or towns, and this is in fact the case throughout the US power grid. Therefore it is clear that reliability requirements are, and should be, variable. Unfortunately, there is considerably less agreement as to precisely what level of reliability is appropriate in each specific situation. CVPS believes that, because reliability and costs are interdependent, reliability should not be considered separately from cost unless there is a compelling reason to do so2. CVPS further believes that responsible stewardship of scarce capital resources demands a thoughtful approach in deciding cost and reliability tradeoffs. Accordingly, CVPS planners have adopted a guiding principle known as most for less. The most for less principle recognizes that the expenditure of capital resources generally results in a diminishing return, and that the appropriate level of expenditure is therefore problematic. Ultimately, the most for less approach seeks to capture most of the reliability benefit of a rigorous planning approach for substantially less cost. Appendix E describes the cost versus reliability tradeoff, as well as the most for less philosophy in greater detail, and goes on to explain the consequent development of a new CVPS reliability standard known as the Equal Slope Criterion that dovetails with this philosophy. In addition to the objective of striking an appropriate cost/benefit tradeoff, there are several more objectives for the present study, including the need to: Satisfy the requirements of the Vermont 20 Year Plan Satisfy the requirements of the Section 248 regulatory statute Develop a solution that is aesthetically unobtrusive Develop a solution that is environmentally responsible Develop a solution that responds to customer needs in a timely fashion

V Study assumptions, methodology, and criteria Approximately four years ago, CVPS conducted loadflow studies in the vicinity of the Southern Loop to assess a proposed tap change on the Vermont Yankee 345/115 kV (T4) transformer. In the course of this analysis, it became apparent that the Southern Loop system had poor short circuit strength, excessive reliance on shunt capacitors, and exceptional vulnerability to voltage instability or collapse. The concerns raised by this analysis eventually lead to the establishment of a larger and more formal study to address the deficiencies, which is that described herein. At the time it was initiated, CVPS decided that the unique problems and weaknesses of the Southern Loop required the use of special analytical techniques and related criteria that went beyond conventional indicators such as voltage adequacy and thermal loading. Problems much like those suffered by the Southern Loop (heavy loading in relation to voltage class, overcompensation with static var sources and voltage instability) have emerged in the past twenty years on numerous mature systems throughout the US power grid. New analytical methods and remedial strategies have been developed to cope with these challenges. QV analysis and PV analysis are two of the newer analytical techniques that are intended for such problems. Both techniques are used in this study and are explained in Appendix A. For example, members of regional power pools who operate bulk-power transmission systems may be compelled to meet certain minimum reliability criteria on these systems, regardless of cost. Also, federal law now mandates additional minimum reliability standards for bulk transmission systems. Therefore, the owners of systems governed by these standards may have to settle for the least expensive option that meets the standards, as opposed to treating both the costs and the standards as interdependent variables. 8
2

The use of QV and PV analysis requires the establishment of acceptance criteria for reactive power and real power margins. The precise value that constitutes "adequate" reactive power margin is a judgment call. A typical standard is to require sufficient margin to withstand any single contingency coincident with loss of the largest reactive power source within the localized area of concern, or "reactive power basin", overlapping the loss of an important line, transformer, or generator. For Stratton, the largest nearby reactive power source (that is actually dispatched) is any one of the 3 local 5.4 Mvar capacitors3. Given the voltage dependency of the these sources (shunt caps), the loss of a nominally-rated 5.4 Mvar bank may result in a further reduction in local reactive supply due to a drop in the output of the other two banks. Therefore it seems reasonable to slightly increase the required margin at Stratton to 6 Mvar (5.4 Mvar x 1.1). The largest reactive power sources electrically close to West Dummerston are the two local 2.7 Mvar shunt caps, therefore, the required reactive power margin at East Jamaica (or West Dummerston) is judged to be 3 Mvar (2.7 Mvar x 1.1 @ W Dummerston). The largest reactive power source close to S Shaftsbury is the 5.4 mvar cap at Manchester, therefore, the required reactive power margin at South Shaftsbury is 6 Mvar (5.4 Mvar x 1.1). Note that the larger 10.8 Mvar caps at Woodford Road and Vernon Road are ignored because they are already "lost" during the initial critical contingencies (including loss of the 115 kV N-186 line from New Hampshire and loss of the 46 kV Southern Loop at or near its eastern termination). In the course of the study however, the 6 Mvar margins often resulted in requirements for substantially increased expenditures that a 3 Mvar margin would avoid, particularly in the case of DG and other non traditional solution options. Accordingly, this standard was relaxed to 3 Mvar in some cases, which is still more rigorous than a simple voltage criterion (the conventional approach). As with reactive power margin, "adequate" real power margin is a judgment call - a typical standard is to require adequate voltage (greater than or equal to 0.90 P.U.) at up to 105% or 110% of peak loading (i.e. a 5% or 10% margin). The standard chosen for this analysis is a 5% margin, which is judged to be prudent but not excessive. In addition to the real and reactive power margin criteria, CV also recognizes related, but more conventional planning requirements such as pre and post contingency voltage limits on subtransmission and transmission equipment of 95% to 105% and 90% to 110% respectively, a deltaV limit for switching operations or contingencies of 5%, thermally-based loading limits for transformers, lines and other current-carrying devices, and angular stability requirements post-disturbance (transient instability or sustained oscillations are not permitted). CVs protection and relay engineers also adhere to a variety of standards for maximum fault-clearing times, protection redundancy, fault current limits, fault duties of critical equipment, and protection coordination with internal and external systems. Loadflow base cases are carefully developed and calibrated against recorded loads, real and reactive power flows, voltages, generation dispatches, and capacitor dispatches. These base cases are then modified to simulate proposed system remedies and modifications, contingencies, alternate generation and capacitor dispatches, intermediate and light loading, peak load growth, and other sensitivities. Because this is a forward-looking study, there has been an attempt not only to identify existing problems and issues, but to anticipate those that will emerge over the next decade or so. Central to this effort is the forecasting and simulation of load growth. Owing to the unique character of the loads in this area (prominent seasonal variation, growing load factor due to DSM, centralized concentration, etc) it was decided that conventional load growth forecasting and modeling was inappropriate. Column D of Table 6 (presented later in this report) denotes the size and location of all Southern Vermont capacitors. 9
3

Therefore a more sophisticated modeling method (when and where) was developed specifically for this study, and is explained in Appendix F. All loadflow cases assume constant real power and constant quadrature power components in their load models, as opposed to constant current or constant admittance components. The reader may be unfamiliar with the terms T0+ and steady state which refer to the two time frames typically assumed for contingency simulations. T0+ is the time frame very soon after a contingency (typically ten to thirty seconds). It comes after fault clearing and attenuation of system transients, but before operator action such as generator and capacitor re-dispatch and system switching. Steady state refers to a later time frame (typically ten to thirty minutes after the contingency) in which all operator actions and automatic tap changes have ceased, and during which the systems parameters are unchanging or changing only slowly. VI The 5 root problems that afflict southern Vermont Each problem will be discussed separately in the subsections below. The reader may wish to refer back to Figure 1 while reading the descriptions of the five root problems. All other things being equal, equipment failure during periods of high demand tends to result in the worst problems, and therefore represents the greatest threat to system reliability. The Southern Loop has its highest demand in winter and is therefore most vulnerable to reliability problems during that season. These problems usually appear as low voltages. However, there may be situations during high summer demand in which equipment heating problems become important. These problems usually involve transmission wires or transformers carrying too much demand during hot weather. Both types of problems (low voltage and equipment overheating) will be addressed where applicable. For now, we will defer the presentation and discussion of loadflow or other analytical evidence. Root Problem #1 - The present-day Southern Loop system is vulnerable to an unplanned loss of a transmission line or a transformer much of the time. The 46 kV system that supplies Southern Vermont is itself fed by 115 kV lines terminating at the Woodford Road Substation near Bennington and at the Vernon Road Substation near Brattleboro. As mentioned earlier, Woodford Road is fed by two 115 kV lines, one from Hoosick Substation in New York and one from Adams Substation in Massachusetts. The redundancy of the Woodford 115 kV supply makes loss of either line fairly inconsequential. Vernon Road Substation however, is fed by a lone 115 kV line from Chestnut Substation in New Hampshire, denoted the N-186 line. Vernon Roads dependence on this single line means that its loss may cause unacceptably low voltage in southern Vermont, including the Southern Loop, even at as little as 45% of peak demand. The demand on the Southern loop is at least this high 66% of the time, meaning that the Southern Loop is exposed to a possible blackout from this one line failure almost two-thirds of the time. Although primarily a winter problem, this exposure does include some higher-demand summer hours as well. During periods of high demand on the Southern Loop 46 kV line, much of that demand is concentrated near its geographic center (at Stratton, Bromley and Manchester) where it is weakest. The worst failures on the 46 kV loop itself are line failures at one of its two ends, which leave most of its demand being fed by a long single 46 kV line. Long single lines tend to have more voltage problems than short single lines because the voltage gradually drops along the length of the line, like the water pressure in a long hose. However, because of the Southern Loops centric load dispersal, line failures occurring anywhere along its length may cause unacceptably low voltage, because they generally lead to a relatively long single span of line supporting a relatively heavy demand. Failure of the Vernon Road 115/46 kV transformer feeding the eastern end of the Southern Loop would be electrically equivalent to losing the 46 kV line itself near the eastern end. Transformer failures tend to be less frequent than line failures but they generally take longer to repair. Moreover,

10

this particular transformer is CV-owned, not VELCO-owned as with most 115/46 kV transformers in Vermont, and is therefore not backed up by VELCOs spare and transportable units. This outage was identified in the 2004 Transformer Failure Study Update as the leading threat to the CVPS transformer infrastructure. At the time of that study, it was recommended that remedial action be deferred, pending the outcome of the in-progress Southern Loop Study, as it was hoped that a common solution might be found. There are several ways to quantify the exposure to poor reliability that threatens the Southern Loop. Perhaps the most intuitive way is to calculate the span of time sufficient to cause a 50% probability of failure as well as that for a 90% probability of failure. 50/90 probabilities are often used as handy and meaningful benchmarks. For the Southern Loop, loadflow simulations and statistical analysis have lead to a prediction that a contingency serious enough to cause some loss of load is at least 50% likely within 3 years and at least 90% likely within 10 years if no remedial actions are taken and if the load remains constant. Appendix I explains the analysis that lead to this conclusion. Root Problem #2 - The present-day Brattleboro area system is vulnerable to an unplanned loss of a transmission line or a transformer 100% of the time. The Brattleboro areas 69 kV system is fed by a single (i.e. radial) 69 kV line from Vernon Road Substation. Vernon Road substation is itself fed by a single 115 kV line (the N-186 line that runs to New Hampshire, mentioned earlier). At present, there is no 115 kV circuit breaker to separate this substation from a 115 kV line failure. Therefore, the loss of the N-186 line invariably results in loss of the transformer that supplies most of Brattleboro. Simply adding the requisite circuit breaker will not remedy the problem because the only other source of power available to the Brattleboro 69 kV system, is the weak Southern Loop 46 kV line fed from Bennington. This long chain of lines would be completely inadequate to supply Brattleboro following a failure of the N-186 line, regardless of demand level. Failure of the (radial) 69 kV line feeding Brattleboro will also black out most of that city. Failure of the Vernon Road 115/69 kV transformer is electrically equivalent to losing the 69 kV line, but would likely be of longer duration. Although there are operating remedies available to promptly restore Brattleboro following an outage, this much load (approximately 20 Mw at peak) being continually vulnerable to a first contingency is a serious problem. As it grows, and as the reliability of the electric grid becomes more indispensable for commerce and public safety, the need for a remedy will become more acute. Furthermore, it is conceptually difficult to reconcile CVPSs endorsement of the equal slope subtransmission reliability criterion (see Appendix E) with a substantially-loaded 69 kV system that has no ability whatsoever to withstand certain first contingencies. These comprise the main threats to Brattleboros reliability due to unplanned equipment failures. Root Problem #3 - The Southern Vermont system (and the South Western New Hampshire system) are vulnerable to a long-term outage of the T4 345/115 kV transformer at the Vermont Yankee nuclear plant. The T4 is a large transformer located at the Vermont Yankee nuclear plant. It supplies the local 115 kV transmission system, including the crucial N-186 line. This transformer (owned and operated by Vermont Yankees Entergy Corporation) is 33 years old and has no on-site backup, nor is there a comparable unit within a reasonable distance that has been earmarked as a backup. Although less damaging to voltage than certain other potential equipment failures, the unavailability of this unit could last for as much as a year if it failed catastrophically, thereby escalating the risk of an overlapping outage of two key system components (i.e. the T4 and some other component). Additionally, Nuclear Regulatory Commission rules would force the shutdown of the plant if the T4 remained unavailable for more than 7 days. Although the shutdown would have little effect on system reliability, it would necessitate the purchase of alternate power by Vermonts utilities, probably at much higher cost. Ultimately this increased cost would be born by Vermonts electric customers. 11

Root Problem #4 - The Southern Vermont system will soon be unable to supply peak demand even with all facilities in service, due to demand growth. All of the problems mentioned thus far are related to possible equipment failures, but there is also a concern that this areas demand may grow to the point where the Southern Loop cannot operate properly at peak demand even with all facilities in service. Computer simulations indicate that, depending on the location and dispersal of future demand growth, the 46 kV loop will begin to experience unacceptably-low voltage after only 3 to 5 Mw of additional growth in peak demand. Given the modest rate of average growth over the past several years (only 0.5% per year on the 46 kV loop itself), one might assume that this crisis is still years away. However, as the graph in Figure 3a indicates, this systems peak demand growth has not been smooth and predictable, but instead has been quite irregular. This irregularity is due to weather variances, economic fluctuations, and other unpredictable factors. Note that the peak demand on the 46 kV loop jumped by over 7 Mw between the winter of 2003/2004 and the winter of 2004/2005. Another sudden jump to a peak demand level beyond the capability of this system, is a distinct possibility. Also, keep in mind the earlier discussion in Section III (Historical Perspective) that the Southern Loops demand factor has increased significantly over time, despite its slowing rise in peak demand. This slow growth in the peak value should not lead to complacency, for beneath it lies an expanding period of high demand (although not peak demand) during which blackouts may occur if a critical line or transformer fails.

Southern Loop Peak Winter Demand History

(On the 46 kV loop itself between Bennington and Brattleboro)

D e m a n d (M w )

72.0 70.0 68.0 66.0 64.0 62.0

1995/96

1996/97

1997/98

1998/99

1999/00

2000/01

2001/02

2002/03

2003/04

2004/05

Year

Figure 3a

Peak Winter Demand Trendline (0.5% per year)

Southern Loop Average Demand History

D e m a n d (M w )

(On the 46 kV loop itself between Bennington and Brattleboro)

32.0 31.0 30.0 29.0 28.0 27.0 26.0 25.0

1995/96

1996/97

1997/98

1998/99

1999/00

2000/01

2001/02

2002/03

2003/04

2004/05

Year

Figure 3b

Average Annual Demand Trendline (1.3% per year)

12

Figure 3b illustrates this hidden demand growth. Although the peak demand in the previous graph is only growing at 0.5% per year, the average demand seen in this graph is growing at 1.3% per year. Moreover, this growth is steadier, as it is less affected by weather abnormalities. It is causing increasingly-long periods of high demand that represent a mounting threat to reliability. Root Problem #5 - Future problems that are related to the Southern Vermont system, will soon emerge on the wider regional transmission system. Several issues relating to the regional transmission system and the demand growth in southeastern Vermont, southwestern New Hampshire, and central Massachusetts are peripherally involved with the Southern Loop and Brattleboro areas. These include the effects of possible transmission line failures in the Monadnock area of southern New Hampshire, a possible failure of VELCOs Coolidge to Vermont Yankee 345 kV line (post NRP), and a significant increase in exposure to regional reliability problems following possible loss of the Vermont Yankee T4 transformer. There are serious reliability problems in this area that would be exacerbated by a long-term failure of this transformer. A regional transmission project in New Hampshire and Massachusetts, called the Monadnock project is being considered by ISO-NE and the local transmission owners to address these problems, with possible construction in 2007/ 2008. Although this project is not designed to remedy Vermonts reliability problems, it would improve several of them. VII Possible solutions to the 5 root problems To facilitate the discussion of solutions, it is necessary to introduce a simplified schematic (i.e. a graphical representation) of the Southern Vermont system. Figure 4 depicts the Southern Loop and Brattleboro areas, using color-coded lines to denote transmission and subtransmission lines, and short dashes intersecting these lines to denote connected substations. Transformers are denoted as zigzag shapes connecting lines of different colors (i.e. different voltages). Other important facilities in the area include the Vermont Yankee nuclear plant and its associated substation, and VELCOs 345 kV transmission line, referred to as the 340 line. Notice that the information presented in Figure 4 is very similar to that in Figure 1. Although Figure 4 is not a scale drawing, it is conceptually correct as to its orientation of important features. Figure 4 and variations of it will be used to describe potential solutions to the five root problems afflicting Southern Vermont. Note the critical N-186 115 kV line from New Hampshire in the lower right corner.

13

The five root problems will be addressed momentarily, in terms of potential solutions. First however, there must be a discussion of general solution approaches. Although many solutions are available to rectify the problems in Southern Vermonts overburdened transmission system, they all tend to fall into one of four categories: 1. Transmission upgrades that involve adding new transmission lines or replacing existing transmission lines at substantially higher voltage (e.g. adding 115 kV lines to an area predominantly served by 46 kV lines). The new or upgraded lines take over much of the burden of serving customer demand, thereby relieving the burden on the existing lines and/or transformers. 2. Generation additions that involve either large centralized units or smaller dispersed units referred to as DG (distributed generation). In either case, the generation sources are most effective if located near the customers that use their output. This relieves the burden on the transmission system by reducing the amount of net load it must carry, and avoids the unintentional severance of these sources from their intended customers, should a line fail. 3. Voltage support devices These tend to fall into one of four subcategories:
Series capacitors are passive devices installed directly in the conductor path to lower net transmission impedance and thereby improve voltage. They have no moving parts. Shunt capacitors are passive devices in substations that inject reactive power into the transmission system and thereby improve voltage. They have no moving parts. Synchronous condensers are active devices in substations that inject reactive power into the transmission system and thereby improve voltage. They are more effective than shunt capacitors but also more expensive. They have rotating components like a motor. Thyristor-based voltage controllers are active devices in substations that inject reactive power into the transmission system and thereby improve voltage. They are more effective than shunt capacitors but also more expensive. They have no moving parts, although their ancillary support systems may include fans or pumps.

4. Load management / load reduction These strategies are also referred to as DSM (demand side management) and tend to fall into one of four subcategories:
Conservation measures such as retrofitting incandescent lighting with high-efficiency florescent lighting, or adding insulation to buildings. Special rates and contracts that encourage customers to conserve power and/or to move their demand to off-peak hours, in order to avoid overburdening the transmission system that supplies them. Load control systems that disable non-essential customer appliances during peak load hours, such as hot water heaters. Fuel switching measures that replace electricity-based heating, cooking, and drying equipment with combustion-based equipment, such as the replacement of electric ovens with gas ovens.

Of course, some solution alternatives may utilize elements from several of these categories, and are therefore combinational or hybrid alternatives. This approach combines the advantages of multiple strategies to fit the unique dimensions of a problem, often making the hybrid solution better than those that rely on a single strategy. The various potential solution alternatives that will be discussed below have been shown through engineering analysis to be effective at solving one or more of the five specific root problems. But these are really just pieces of a larger puzzle that has yet to be put together. Some of these pieces will be used and some will not in the ultimate plan that solves the full range of problems in a costeffective manner. Having explored the five root problems and some general solution categories, we now turn our attention to Table 1, which is organized in terms of these five problems. Notice that the potential transmission solutions are depicted in the mini schematic diagrams just below the table, which are similar to Figure 4. Note that the N-186 line is left out of each one due to lack of space. Except for the synchronous condenser installation, the solution options involving DG and DSM are presented only textually in the table.

14

We will re-consider the five root problems one by one, and discuss the potential solutions to each, as summarized in Table 1. The way in which each solution works to remedy each root problem is rather technical, but their basic workings will be explained. The discussion of costs, funding sources, aesthetics, and environmental impacts will be deferred for now. Instead, we will focus simply on what measures can potentially solve the root problems. Once we have a list of workable candidates we can begin to consider their relative merits and how they may be combined in a strategic solution that cost-effectively solves all of the five root problems, with due consideration to environmental factors, aesthetics, and economic risk. Root Problem #1 - The present-day Southern Loop system is vulnerable to an unplanned loss of a transmission line or a transformer much of the time. Beginning with potential transmission solutions, the best transmission solution to this root problem appears to be the installation of the following three components: A new synchronous condenser installation at a new Stratton 46 kV substation. A new Vermont Yankee to West Dummerston 115 kV transmission line on the existing VELCO 345 kV line right-of-way between Vermont Yankee and Coolidge substations (possibly constructed for future operation at 345 kV). A new substation at West Dummerston with a 115/46 kV transformer, connecting the new 115 kV line from Vermont Yankee with the 46 kV Southern Loop.

The synchronous condenser installation helps to hold up the voltage near the middle of the 46 kV loop. This locality has heavy demand in winter and is distant from the nearest strong source of voltage support at Bennington, whenever part of the 46 kV line just west of Dummerston fails. It is also distant from the nearest strong sources of voltage support at Brattleboro and Dummerston, whenever part of the 46 kV line just east of Bennington fails. The proposed new 115/46 kV substation at Dummerston fed by the new 115 kV line from Vermont Yankee provides a strong source of voltage support for the eastern portion of the 46 kV loop that is presently vulnerable to a failure of the 115 kV N-186 line. This completes the list of likely threats that contribute to this root problem, and the ways in which transmission solutions can help. Moving now to potential DG solutions, the best DG-only solution (as opposed to a hybrid solution involving some DG) to this root problem appears to be 80 Mw of DG, dispersed as follows: 100% offset of the Stratton demand (21 Mw), 100% offset of the Bromley demand (6 Mw), 100% offset of the Brattleboro demand with islanding capability (16 Mw) and partial offset of the remaining demand on the 46 kV loop (37 Mw out of a remaining total of 56 Mw), and no offset of the Bennington demand. This concentrates the DG near to the demand that is most detrimental to voltage during an outage of the same transmission or subtransmission line segments noted above. Last are the potential DSM solutions. DSM solution options tend to be more limited than transmission or DG because the opportunities to reduce existing demand are finite, whereas DG and transmission additions are basically limited only by their cost. On the other hand, whatever modest DSM potential may be available is usually worth considering because its cost is typically low or negative4 and its effect on electric system reliability can only be beneficial. A preliminary screening of the Southern Vermont area shows a potential for approximately 26 Mw of cost-effective DSM dispersed as follows: 7.2 Mw in Bennington, 5.4 Mw on the 46 kV loop west of Stratton, 0.4 Mw at Bromley, 1.6 Mw at Stratton, 4.2 Mw on the 46 kV loop east of Stratton, and 7 Mw in Brattleboro (note however that this 7 Mw is only useful in one particular solution option). This estimate was subsequently verified by an independent consultant, Optimal Energy. Its analysis determined the total available economic DSM to be about 29 Mw, which is judged to be close enough to the initial estimate to preclude a reassessment of the DSM-related analysis. The cost of DSM may be negative because it avoids many other costs such as building new distribution, transmission, and generation systems, and the cost of fuel for generators. 15
4

Table 1 - Root Problems and Potential Solutions

root problem appropriate reliability guideline potential transmission solution(s) PTF funding aesthetic considerations environmental considerations comments potential DG solution(s) PTF funding aesthetic considerations environmental considerations comments potential DSM solution(s) PTF funding aesthetic considerations environmental considerations comments none low visual impact most DG technologies generate noise and emissions best cost effectiveness at ski resorts (27 Mw) 26 Mw distributed around Southern VT none virtually no visual impact positive environmental impact (reduces emmissions) works best in combination with other solutions Brattleboro DSM could supplement Brat DG N/A N/A N/A N/A
requires complete offset with islanding capability (16Mw)

#1 Existing S Loop contingency reliability exposure equal slope criterion sync cond and VY-Dumm 115 kV line (below) none low visual impact synch. cond. noise, new 115 kV line, but no new ROW

#2 Existing Brat contingency reliability exposure equal slope criterion VY-Dumm 115 kV line (below) none low visual impact new 115 kV line but no new ROW

#3 Existing VY T4 long-term outage exposure deterministic (N-1 / N-2) spare T4 or VY-Dumm-Cool 115 kV line (below) either option would receive significant PTF funding low visual impact for either, but lower for T4 backup new 115 kV line but no new ROW

#4 Future load growth in southern Vermont equal slope criterion sync cond and VY-Dumm-Strat 115 kV line (below) none new 115 kV line on existing 46 kV line ROW new 115 kV line on existing 46 kV line ROW

only transmission can solve this N/A N/A N/A N/A only transmission can solve this N/A N/A N/A N/A

complete offset of 10 yr growth would be 29 Mw none low visual impact most DG technologies generate noise and emissions

none low visual impact most DG technologies generate noise and emissions

available amount is unclear none virtually no visual impact positive environmental impact (reduces emmissions) can slow the growth of new load, and/or supplement DG

potential transmission solution

potential transmission solution

potential transmission solutions

potential transmission solution

or

#5 future regional reliability problems deterministic (N-1 / N-2)

VY-Dumm-Benn or VY-Dumm-Cool 115 kV line (below)

either option would receive significant PTF funding new 115 kV line on existing 46 kV line ROW new 115 kV line on existing 46 kV line ROW

VELCO input needed none low visual impact most DG technologies generate noise and emissions

VELCO input needed none virtually no visual impact positive environmental impact (reduces emmissions)

potential transmission solutions

or (better for regional reliability )

Based on these estimates, DSM will not be sufficient by itself to completely remedy any of the 5 root problems in Southern Vermont, but may be regarded as a useful part of potential combination strategies that also rely on transmission and/or DG. Root Problem #2 - The present-day Brattleboro area system is vulnerable to an unplanned loss of a transmission line or a transformer 100% of the time. The best potential transmission solution to this root problem, of those identified, appears to be the installation of the following three components: A new Vermont Yankee to West Dummerston 115 kV transmission line on the existing VELCO 345 kV line right-of-way between Vermont Yankee and Coolidge substations (possibly constructed for future operation at 345 kV). A new substation at West Dummerston with a 115/46 kV transformer, connecting the new 115 kV line from Vermont Yankee with the 46 kV Southern Loop. A new 115 kV circuit breaker at the existing Vernon Road Substation to separate its 115/69 kV transformer (and hence the 69 kV line feeding Brattleboro) from the 115 kV N-186 line when it fails. Note that this circuit breaker cannot be seen in the schematic.

Interestingly, two of these three transmission components are included in the three transmission components cited earlier as the best transmission solution to the first root problem. An examination of the transmission schematics in Table 1 will confirm this commonality. Why are these same two components effective for this distinctly different problem? As noted earlier, the new 115/46 kV substation at Dummerston fed by the new 115 kV line from Vermont Yankee, provides a strong source of voltage support for the eastern portion of the 46 kV loop that is presently vulnerable to a failure of the 115 kV N-186 line to New Hampshire. This outage will not only cause low voltage on the Southern Loop, but will unavoidably black out most of Brattleboro as well because there is no breaker to separate Brattleboro from the faulted N-186 line. The addition of such a breaker in this solution allows the strong voltage source created at Dummerston to not only carry the 46 kV loop, but also the Brattleboro 69 kV line after the N-186 fails. Of course, the failure of the single (i.e. radial) 69 kV line feeding Brattleboro will still result in its immediate blackout, but it can be quickly restored in most cases by switching in supplies from the north and south. Turning now to potential DG solutions; recall that most of Brattleboro is presently fed by a single (i.e. radial) 69 kV transmission line. Therefore, the failures that affect Brattleboro are generally ones that completely cut it off from the rest of the electrical system. Any DG solution that may be proposed must recognize and address this problem. The only way for DG to work under these circumstances is for it to be capable of operating in an island configuration. This requires enough DG to carry the entire demand of the island (presently 16 Mw, at peak), as well as special controls to match the islands demand to the output of its generators. These two requirements make the DG solution expensive in relation to the amount of demand that is jeopardized by this root problem. Because of the islanding issue, DSM is not technically capable of solving this root problem by itself, but could offset some of the required DG cited in the paragraph above. Root Problem #3 - The Southern Vermont system (and the South Western New Hampshire system) are vulnerable to a long-term outage of the T4 345/115 kV transformer at the Vermont Yankee nuclear plant. There are two potential transmission solutions to this root problem. The first is to simply connect a second transformer at the Vermont Yankee substation, similar in capability and design to the one presently operating, as a backup or spare. The second potential transmission solution is to build the following two components:

17

A new Vermont Yankee to West Dummerston 115 kV transmission line on the existing VELCO 345 kV line right-of-way between Vermont Yankee and Coolidge substations (possibly constructed for future operation at 345 kV). A new West Dummerston to Coolidge 115 kV line on the existing VELCO 345 kV line right-ofway between Vermont Yankee and Coolidge substations (possibly constructed for future operation at 345 kV).

The effect of the first transmission alternative (the spare T4 transformer) is self-explanatory. The effect of the second transmission alternative is to provide a strong 115 kV source to the Vermont Yankee 115 kV Substation from Coolidge 115 kV Substation, in order to back up the source of 115 kV supply provided by the T4 transformer. Coolidge is very strong because it is supplied by a highvoltage transmission line (VELCOs 345 kV 340 line). Therefore the line coming from it to Vermont Yankees 115 kV substation would also be strong. With this strong new source of 115 kV supply, the failure of the T4 transformer would be of little consequence. For various reasons, neither DG nor DSM appears capable of providing a practical solution to this root problem. First, the N-186 lines voltage weakness following a failure of the T4 transformer, is a result not only of the demand in Vermont, but also of that in New Hampshire. And, because DG/DSM installations are most effective when located at or near the demand they seek to offset, much of the DG/DSM would need to be put in New Hampshire (outside the jurisdiction of CVPS, VELCO, and the State of Vermont). Furthermore, it is likely that DG/DSM would be an unacceptable backup to the T4 in the event of its failure. This means that the Vermont Yankee nuclear plant would likely have to shut down for a longterm outage of the T4, leading to long-term power supply disruptions and high power replacement costs for Vermont customers. Root Problem #4 - The Southern Vermont system will soon be unable to supply peak demand even with all facilities in service, due to demand growth. Based on the analysis done thus far, the best transmission solution to this root problem appears to be the installation of the following five components: A new synchronous condenser installation at a new Stratton substation. A new Vermont Yankee to West Dummerston 115 kV transmission line on the existing VELCO 345 kV line right-of-way between Vermont Yankee and Coolidge substations (possibly constructed for future operation at 345 kV). A new 115/46 kV substation at West Dummerston, connecting the new 115 kV line from Vermont Yankee with the 46 kV Southern Loop. A new West Dummerston to Stratton 115 kV transmission line on the existing Southern Loop 46 kV line right-of-way. A new 115/46 kV transformer at a new Stratton substation.

The synchronous condenser installation and the new 115 kV transmission line from Vermont Yankee to West Dummerston to Stratton, would help support future demand growth on the Southern Loop, particularly that near the middle of the 46 kV loop, where robust growth is expected over the next several years. These two sources of voltage support would also back each other up if one suffered a failure, thereby permitting the expected new demand to be served even during a contingency. As explained earlier, the strong support of the 46 kV system at Dummerston provided by the new 115 kV reinforcements would help to support the voltage at the Vernon Road Substation and therefore that of the Brattleboro 69 kV system as well (because it is fed from the Vernon Road Substation). This improvement would support future Brattleboro demand growth. Based on the analysis done thus far, the best DG solution to this root problem is simply to build new generation in the same quantity as new demand, and dispersed in the same locations, to the extent possible. The amount of new demand in Southern Vermont over the next 10 years is forecast to be 18

approximately 29 Mw. This matching of new generation to new demand means the transmission system supplying the area would experience no change in its net burden. DSM could offset some of the necessary DG. Root Problem #5 - Future problems that are related to the Southern Vermont system, will soon emerge on the wider regional transmission system. Based on the analysis done thus far, the best transmission solutions to this root problem are new network (not radial) transmission lines that run parallel to the existing lines, like parallel roads. And, like new parallel roads, these new parallel lines share the traffic (i.e. the demand) and thereby reduce the burden on the original pathways. Two potential transmission solutions have been identified, both of which provide parallel pathways for existing regional transmission lines that may become overburdened, especially after one of them has failed. The individual components of each of the two solutions are listed below: A new Vermont Yankee to West Dummerston 115 kV transmission line on the existing VELCO 345 kV line right-of-way between Vermont Yankee and Coolidge substations (possibly constructed for future operation at 345 kV). A new West Dummerston to Stratton 115 kV transmission line on the existing Southern Loop 46 kV line right-of-way. A new Stratton to Woodford Road (Bennington) 115 kV transmission line on the existing Southern Loop 46 kV line right-of-way. - or A new Vermont Yankee to West Dummerston 115 kV transmission line on the existing VELCO 345 kV line right-of-way between Vermont Yankee and Coolidge substations (possibly constructed for future operation at 345 kV). A new West Dummerston to Coolidge 115 kV line on the existing VELCO 345 kV line right-ofway between Vermont Yankee and Coolidge substations (possibly constructed for future operation at 345 kV).

The increased connectivity created by these new transmission lines would help electricity to move more freely across central New England, even if some transmission pathways were to fail. Potential DG and DSM solutions, sufficient to defer the need for regional transmission improvements for 10 more years, would require generation additions and/or demand reductions on the order of hundreds of mega-watts. The precise amounts and dispersal of these resources will require in-depth engineering studies, yet to be performed5. VIII Integration of solution components to form a strategic plan Having explored the five root problems of Southern Vermont and a broad spectrum of potential transmission, DG, and DSM solutions, we may now consider how these puzzle pieces may fit together to create an overarching solution.

In an on-going Vermont regulatory investigation (Docket 7081) parties have proposed to establish the responsibility for these studies within the state. An association of affected distribution companies, VELCO, and public representatives operating under the supervision of a new planning authority known as the VSPC (Vermont System Planning Committee) is expected to perform this task in the near future, if the proposal is approved.

19

So far, this report has emphasized the discrete technical problems of Southern Vermont and the associated solution elements that might be used to solve them. Now we advance from these fundamentals to the more nuanced process of synthesizing a long-term strategy for Southern Vermonts electric system. The best way to describe the many combinations of solution components that are to be discussed, is to develop a streamlined nomenclature, that is, a special shorthand. Figure 5 provides a reference schematic for this nomenclature.

Notice that Figure 5 is very similar to Figure 4, upon which it is based. The dotted facilities are potential new 115 kV transmission lines, 115/46 kV substations and transformers, and a synchronous condenser installation at Stratton. Cost-effective DSM achievable within 10 years is quantified according to its location, in boxes. It adds up to 26 Mw, which is the amount estimated to still be available in Southern Vermont. The potential new 115 kV transmission lines are assigned names according to the blue segments: A is a new Vermont Yankee to West Dummerston 115 kV transmission line on the existing VELCO 345 kV line right-of-way between Vermont Yankee and Coolidge. B is a new West Dummerston to Coolidge 115 kV transmission line on the same right-of-way. C is new West Dummerston to Stratton 115 kV line on the existing West Dummerston to Stratton 46 kV right-of-way. D is a new Stratton to Woodford Road (Bennington) 115 kV transmission line on the existing Stratton to Woodford Road 46 kV right-of-way. The blue S denotes the synchronous condenser installation.

Most of the reasonable and practical strategic solutions for the Southern Loop and Brattleboro areas can be described using this schematic. However, DG installations are difficult to portray graphically and must be described textually instead, because they tend to be dispersed in many locations and in different sizes.

20

Note that this nomenclature will assume any new transmission lines to include whatever new substation and transformers are located at their end points. For example, the A,C combination of new transmission lines, and the A,C,D combination of new transmission lines would both be assumed to include the new West Dummerston and Stratton 115/46 kV substations and transformers. CVPS and VELCO have developed 10 strategic solution options using the solution components described earlier, most of which may be visualized by using the reference schematic. These 10 possibilities may not be exhaustive, but capture the most effective of the strategic solution options that CVPS and VELCO have identified for Southern Vermont. Table 2 lists these 10 options, with some general categorization and a few brief comments. Appendix J provides detailed descriptions for all 10 solution options.

Table 2 - Strategic Solution Options

Option 0 1 2 3 4 5 6 7 option components Existing System A,B,S A,B,C,S A,B,C,D A,B,C,D,S A,C,D,S A,T4,S A,B,S,DG (25 Mw) A,B,S,DSM (19 Mw) / DG (21 Mw) 8 A,T4,S,DG (25 Mw) A,T4,S,DSM (19 Mw) / DG (21 Mw) 9 S,DG (42 Mw) S,DSM (26 Mw) / DG (38 Mw) 10 DG (80 Mw) DG (105 Mw) pure generation hybrid hybrid option type N/A pure transmission pure transmission pure transmission pure transmission pure transmission pure transmission hybrid The "T4" component is a backup transformer for the existing Vermont Yankee T4 initial configuration final configuration (by 10th year) initial configuration final configuration (by 10th year) initial configuration final configuration (by 10th year) initial configuration final configuration (by 10th year) comments

As explained earlier, the table uses our special nomenclature for describing the components of each strategic option. For example, Option #7 includes the following transmission components (compare with the Figure 5 reference schematic): S - A new synchronous condenser installation at a new Stratton substation. A - A new Vermont Yankee to West Dummerston 115 kV transmission line on the existing VELCO 345 kV line right-of-way between Vermont Yankee and Coolidge substations. B - A new West Dummerston to Coolidge 115 kV line on the existing VELCO 345 kV line right-of-way between Vermont Yankee and Coolidge substations. Implied by A and B - A new substation at West Dummerston with a 115/46 kV transformer, connecting the new 115 kV line from Vermont Yankee to the 46 kV Southern Loop.

The reader will notice that Option #7 has two color-coded listings in the table instead of just one. The first listing is for the current year, and the second listing is for the tenth year. No immediate DSM load reduction is assumed to be possible, so the first years DSM value is zero. However, over a period of 10 years, it is predicted that 19 Mw of useful DSM will have been achieved6. This will permit reliance on DG to be reduced from 25 Mw in the first year to 21 Mw in the tenth year, despite the growth in demand over the same period. So in summation, this option includes several transmission improvements, some immediate DG installations, and steady progress in DSM load reduction over ten years time that will permit gradual reductions in required DG use.

An additional 7 Mw of DSM would be available in Brattleboro but is not needed with a strong West Dummerston source. 21

The reader is encouraged to review each of the 10 strategic solution options listed in Appendix J and to become familiar with their general configuration and approach7. We can help this process by discussing each of them according to their measures of performance and other attributes. We will start by offering some additional detail as to the location and size of the assumed DSM and DG values (we have provided the DSM values and locations earlier, but are restating them here for the sake of completeness).

Table 3 - DSM and DG locations and amounts (Mw)

Option 0 1 2 3 4 5 6 7 option components Existing System A,B,S A,B,C,S A,B,C,D A,B,C,D,S A,C,D,S A,T4,S A,B,S,DG (25 Mw) A,B,S,DSM (19 Mw) / DG (21 Mw) 8 A,T4,S,DG (25 Mw) A,T4,S,DSM (19 Mw) / DG (21 Mw) 9 S,DG (42 Mw) S,DSM (26 Mw) / DG (38 Mw) 10 DG (80 Mw) DG (105 Mw) comments none none none none none none none 20 Mw DG @ Stratton, 5 Mw DG @ Bromley 19 Mw DSM dispersed per screening tool*, 21 Mw DG @ Stratton 20 Mw DG @ Stratton, 5 Mw DG @ Bromley 19 Mw DSM dispersed per screening tool*, 21 Mw DG @ Stratton 21 Mw DG @ Stratton, 6 Mw DG @ Bromley, 15 Mw DG @ Brattleboro 26 Mw DSM dispersed per screening tool*, 31 Mw @ Stratton, 7 Mw @ Bromley 21 Mw @ Stratton, 6 Mw @ Bromley, 16 Mw @ Brattleboro, 17 Mw @ west loop, 20 Mw @ east loop 31 Mw @ Stratton, 9 Mw @ Bromley, 18 Mw @ Brattleboro, 21 Mw @ west loop, 26 Mw @ east loop

* 7.2 Mw in Bennington, 5.4 Mw on the 46 kV loop west of Stratton, 0.4 Mw at Bromley, 1.6 Mw at Stratton, 4.2 Mw on the 46 kV loop east of Stratton, and ZERO Mw in Brattleboro (or 7 Mw if Option # 9 is assumed).

Table 4 is a summary of how effectively each of the 10 strategic solution options solve the 5 root problems that afflict Southern Vermont. Note that only the final configuration (in year 10) of any multiphased solution option appears in Table 4, that is, only the final configurations are noted in Table 4 for Options #7, #8, #9, and #10. Note that all 10 strategic solution options would also solve an important element of Root Problem #1, that of the possible long-term failure of CVPS Vernon 115/46 kV transformer, mentioned earlier. Table 5 is a summary of the costs of each of the 10 strategic solution options. By directly comparing the effectiveness and costs in Tables 4 and 5, the reader may gain a sense of the advantages and disadvantages of each strategic approach. Be aware that these estimates are preliminary and are therefore subject to significant changes as market conditions for labor, materials, and expertise fluctuate. Bear in mind that the selection of certain options does not necessarily preclude others from being built later on. For example, choosing Option #1 does not preclude the possibility of building Option #2 at a later time because the first projects components (A,B,S) are all common to that of the later project (A,B,C,S)8. On the other hand, a decision to build Option #10 would not lend itself to any other option listed, because the initial investment in 80 Mw of DG is not common to any other option. Note however, that transmission improvements rarely preclude later investments in DG or DSM.
7

Interestingly, the Chester-Londonderry 46 kV line option, mentioned in Section III Historical perspective is not included. The explanation for this omission is given in Appendix D. 8 Of course, the cost to progress from Option #1 to Option #2 would only be the incremental cost to build line C, not the entire cost to build Option #2.

22

Table 4 - Strategic Solution Option Effectiveness Summary

Southern VT root problem Option 0 1 2 3 4 5 6 7 8 9 10 Existing System A,B,S A,B,C,S A,B,C,D A,B,C,D,S A,C,D,S A,T4,S A,B,S,DSM (19 Mw) / DG (21 Mw) A,T4,S,DSM (19 Mw) / DG (21 Mw) S,DSM (26 Mw) / DG (38 Mw) DG (105 Mw) #1 - Existing Southern Loop contingency reliability exposure #2 - Existing Brattleboro contingency reliability exposure #3 Existing T4 long-term outage exposure #4 Future load growth in southern Vermont #5 future regional reliability problems combined effectiveness score (on a scale of 10)

0 8 10 8 10 9 6 10 9 6 7
.

Legend

= very effective

= somewhat effective

= ineffective

Note: The "combined effectiveness score" for each solution option is determined by assigning 2 points for a full circle, 1 point for a half circle, and zero points for an empty circle, in each problem category.

Table 5 - Preliminary Strategic Solution Option Cost Summary

Option 0 1 2 3 4 5 6 7 8 9 10 Existing System A,B,S A,B,C,S A,B,C,D A,B,C,D,S A,C,D,S A,T4,S A,B,S,DSM (19 Mw) / DG (21 Mw) A,T4,S,DSM (19 Mw) / DG (21 Mw) S,DSM (26 Mw) / DG (38 Mw) DG (105 Mw) Capital cost ($M) PTF ($M) Non-PTF ($M) New England 20 yr PWRR ($M) VT 20 yr PWRR ($M)

$ZERO $79.6 M $106.6 M $136.1 M $146.6 M $108.0 M $71.0 M $161.6 M $153.0 M $130.5 M $105.0 M

$ZERO $65.2 M $65.2 M $115.2 M $115.2 M $76.6 M $30.0 M $65.2 M $30.0 M $ZERO $ZERO

$ZERO $14.4 M $41.4 M $20.9 M $31.4 M $31.4 M $41.0 M $96.4 M $123.0 M $130.5 M $105.0 M

$ZERO $124.7 M $146.5 M $186.2 M $200.7 M $147.0 M $111.7 M $50.0 M $36.8 M ($51.4 M) $504.5 M

$ZERO $73.0 M $94.8 M $94.9 M $109.4 M $86.3 M $87.9 M ($1.8 M) $13.0 M ($51.4 M) $504.5 M

A final decision has not yet been made by CVPS as to which of these strategic solution options should be selected for a 248 application. That decision will depend, in part, on the recommendations received from the public through the USC / CWG process. In making this decision, the tradeoff between solution costs and solution effectiveness must be carefully considered because both will impact CVPS customers. We wish to strike a balance, and to avoid an unreliable system on the one hand, or an overly expensive system on the other. Of course, there are further considerations such as environmental impact, noise (of possible generators), and aesthetics (particularly of transmission lines). At the time that this report was issued, the CWG had recommended that CVPS and VELCO narrow their consideration to only Solution Options #2, #4, and #7, but had decided not to recommend a single option. IX The common thread (a synchronous condenser installation at Stratton) The problems on the Southern Vermont system have by now become a serious and immediate threat to reliability. Windows of opportunity to perform maintenance on critical lines, transformers, and other

23

equipment have diminished to the point where CVPS can only remove certain equipment from service during temperate weather coinciding with weekends. On September 27th 2005 an equipment failure at the Woodford Road substation in Bennington, during such a maintenance operation, lead to a blackout of several hours for tens of thousands of Southern Vermont electric customers. Had this maintenance been necessary during a much higher demand period, a rolling blackout would have been required even without the coinciding equipment failure. Conversely, had the equipment failure occurred during a much higher demand period, even without the coinciding maintenance operation, an unplanned blackout would have occurred. Accordingly, CVPS must act to restore Southern Vermonts reliability in the short run. This issue was carefully considered at the two-day USC meeting on January 30/31 of 2006. The USC recommended that the synchronous condenser installation be added as soon as possible because it appeared to have the best benefit-to-cost ratio of any single solution component and was compatible with most of the solution alternatives defined thus far. CVPS agrees with this rationale therefore intends to file a 248 application for the synchronous condenser installation, using 2 x 15 Mvar (capacitive) machines. In support of this objective, consider that eight of the ten strategic solution options in this report include the synchronous condenser installation as one of their key components. A synchronous condenser is a relatively low-cost device that has been shown in extensive engineering studies to permit dramatic reductions in the durations of exposure on the Southern Loop to critical equipment failures. The only two strategic solution options that exclude the synchronous condenser installation are Strategic Solution Option # 3 (A,B,C,D) and Strategic Solution Option #10 (pure DG). Let us briefly consider these two options. Note that the synchronous condenser installation does not really preclude the use of Strategic Solution Option # 3 (A,B,C,D) because Strategic Solution Option # 4 (A,B,C,D,S) is effectively the same option with the synchronous condenser installation included, and its reliability performance is significantly better in the summary tables, with only a modest cost premium. The conclusion is that the opportunity cost of pre-installing the synchronous condenser facility in this case would be small or zero. Turning now to a comparison with Strategic Solution Option #10 (80 mw of DG to start and eventually 105 Mw of DG), we note that once again, a synchronous condenser installation would not really preclude the use of Strategic Solution Option #10 because Strategic Solution Option #9 (S, 42 Mw of DG to start and eventually 26 Mw of DSM, and DG reduced to 38 Mw), is effectively the same option but with the synchronous condenser installation included and DSM, which reduces the necessary DG amount. Its performance is significantly below that of Strategic Solution Option #10, but its cost is much lower. So again, the conclusion is that the opportunity cost of pre-installing the synchronous condenser facility in this case would be small or zero. However, the synchronous condenser installation would provide little or no relief for the Brattleboro exposure, but this problem could be managed in the short run in a variety of ways. One possible remedy is the installation of a relatively inexpensive capacitor at the Brudies Road Substation, to support voltage following system restoration via the 69 kV line to Bellows Falls. Recent analysis and operating experience with a temporarily-installed portable capacitor have verified the effectiveness of this approach. There are also proposals by National Grid to make system changes that may result in upgrades to, or reconfigurations of the 69 kV line to Bellows Falls. These changes may mitigate, or at least partially mitigate the voltage weakness in this area. X Technical analysis of a synchronous condenser installation (alone)

24

So far most of this report has described the system inadequacies in Southern Vermont and has explained the benefits and costs of possible strategic solution options. However, the ultimate purpose of this document is to justify a synchronous condenser installation at Stratton as the first step of a broader strategic solution still to be determined. Therefore additional technical detail about the synchronous condenser installation, operating without further reinforcements, is now in order. Table 6 provides the results of various load flow computer simulations of the Southern Vermont electric system with and without the synchronous condenser installation. The table may be referenced by row and column using numbers and letters respectively. The columns are organized as follows: Column A provides the name of the simulation case, which is probably useful only to the author. Column B provides the assumed load level as a percentage of peak winter load (100%). Note that cases at each simulated load level were calibrated according to actual substation loads in southern Vermont, recorded within the last several years. Column C provides the annual percent duration during which loading is equal to or lower than that simulated. Column D, which is really 13 adjacent columns, denotes the assumed shunt capacitor dispatch in each simulation. The stars indicate a dispatched capacitor. The Vernon and Brudies capacitor columns have a gray background because they are more closely associated with the Brattleboro area than with the Southern Loop. Column E provides the amount and location of assumed interruptible load that has actually been shed in the simulation (usually zero). Column F provides the amount and location of assumed non-interruptible load that has actually been shed in the simulation (always zero). Column G denotes the assumed contingency (if any). Column H provides the assumed time frame of the simulation. Some cases assume the time frame immediately following a contingency, just after breaker operations (T0+) and attenuation of system transients. These cases help determine whether the unfaulted system will successfully ride through the disturbance without criteria violations or system instability. Other cases assume a longer time frame following a contingency, including the time required for corrective actions by system operators (steady state). These actions, which are simulated, may lessen the systems burden (for example, dispatching additional capacitors) or they may increase it (for example, restoring loads that were previously tied to the fault, by re-sectionalizing). These simulations help determine how much disrupted load (if any) may be successfully restored post-contingency, prior to repairing the faulted element itself. Column I denotes whether the simulation assumes the added support of the synchronous condenser installation or not. Thus far, the columns described have denoted independent parameters, that is, inputs to the loadflow simulations, chosen by the engineer performing the analysis. The remaining columns denote parameters that are dependent, that is, outputs from the loadflow simulations. Columns J, K, and L provide the per-unit voltages at three representative locations across the Southern Loop, moving west to east from S Shaftsbury 46 kV to Stratton 46 kV to West Dummerston 46 kV. Columns M and N provide the lowest and highest per-unit voltages and their locations on the 46 kV loop. Columns O, P, and Q provide the reactive power margins at specific locations across the Southern Loop, again moving west to east from S Shaftsbury 46 kV to Stratton 46 kV to West Dummerston 46 kV. Appendix A explains these important measures of system strength. Column R provides the real power margin of the Southern Loop as a whole. Appendix A also explains this important measure of system strength. Column S provides special commentary where appropriate. Additionally, Table 6 contains explicit footnotes and general end-notes that provide further clarification.

25

The key question now before us is How effectively will the synchronous condenser installation, by itself, remedy the five root problems that afflict Southern Vermont? An examination of Table 6, coupled with some deductive reasoning can provide the answers. The first root problem is restated below: Root Problem #1 - The present-day Southern Loop system is vulnerable to an unplanned loss of a transmission line or a transformer much of the time. Table 6 entries 1 through 8 provide the results of a set of non-contingency (all facilities in-service) simulations with gradually declining southern Vermont loading, and tests of performance with versus without the synchronous condenser installation. Although Root Problem #1 is specifically about contingency issues, not non-contingency issues, these non-contingency entries provide insight as to system vulnerabilities that may be exacerbated under contingency. Entry 1 indicates that the existing system is showing signs of stress at 100% of peak winter load even with all facilities in service. Shunt capacitors have been optimally dispatched. Despite this, the 46 kV system voltage at Stratton is only 94% of nominal, which is slightly out of compliance with the criterion for a non-contingency condition (95% minimum). The reactive power margin at Stratton is 2.9 Mvar which is moderately out of compliance with the criterion of 6.0 Mvar. The real power margin of 5.7 Mw is in compliance of the 5.0 Mw criterion, but this standard is based on a minimum of 90% voltage. Some authorities would insist on a 95% minimum voltage for a non-contingency, which would render this too out of compliance. In any event, the system is clearly near its load-serving limit. Entry 2 is the same simulation, but with the proposed synchronous condenser installation added. It shows a dramatic improvement over the marginal results of the previous simulation. The Stratton voltage is now at its ideal nominal voltage, with all reactive and real power margins substantially beyond minimum criteria (the Stratton reactive power margin is now at 32.9 Mvar and the real power margin is at 46.3 Mw) indicating stable and robust system operation. Entries 3 and 4 provide a similar comparison without and with the synchronous condenser installation, again assuming no contingency, but now at 90% of peak load instead of 100%. The existing system, even without the synchronous condenser installation, is now comfortably in compliance with all criteria. However, the corresponding system with the synchronous condenser installation still has better margins. Entries 4 through 8 assume successively lower load levels for the existing system. Despite the declining use of shunt capacitors in this series of cases, the margins of the existing system grow steadily more robust, as expected. Entries 9 through 13 provide our first glimpse of contingency performance. The assumed contingency is the loss of the West Dummerston-North Brattleboro 46 kV line (steady state). Both simulations without and with the synchronous condenser installation are non-convergent9 at 100% of peak load (entries 9 and 10). This indicates a likely voltage collapse under both of these conditions. The same two simulations run at 90% of peak load still indicate a voltage collapse without the synchronous condenser installation (entry 12), but show acceptable performance with the synchronous condenser installation (entry 13). Entry 11 is an ancillary simulation, testing the effect of supervisory loadshedding at Stratton, at 100% of peak load with the synchronous condenser installation.

Non-convergence is the failure of the load flow simulation to achieve a mathematically acceptable answer. This may indicate an ill-conceived simulation model or it may be a portent of actual voltage collapse. All simulations in this study exhibiting non-convergence were carefully reviewed and tested further to determine its true origin. Non-convergent cases presented in this report have been verified to be genuinely indicative of severe voltage depression or collapse, and are not the result of modeling error.
9

27

Entries 14 through 19 assume the loss of the 115 kV N-186 line coming from New Hampshire (T0+). This is a severe contingency that has proven to be definitive throughout this study. It is among the two or three worst contingencies in this area and its remediation has driven much of the recent planning effort for Southern Vermont. All simulations exceeding 70% of peak load (entries 14 through 17) were non-convergent and would likely result in voltage collapse. The simulation at 70 % of peak load with the synchronous condenser installation (entry 19) was marginally acceptable, having a reactive power margin at West Dummerston of 3.1 Mvar versus a minimum criterion of 3.0 Mvar. The same simulation without the synchronous condenser installation (entry 18) was non-convergent and indicates a likely voltage collapse. Entries 20 through 27 assume the loss of the Woodford-Manchester 46 kV line (T0+). All simulations exceeding 70% of peak load (entries 14 through 17) were non-convergent and would likely result in voltage collapse. The simulation at 70% of peak load with the synchronous condenser installation (entry 27) was marginally acceptable, but the same simulation without the synchronous condenser installation (entry 26) was non-convergent. Entries 28 through 34 assume the loss of the Woodford-South Shaftsbury 46 kV line (steady state). Interestingly, this contingency is closely related to the prior contingency (loss of the WoodfordManchester 46 kV line) in that it shows the effect of the simulated response of the operators. The results are strikingly similar to those observed before supervisory actions were taken, because the added burden of the load restoration between Manchester and South Shaftsbury has been counterbalanced by the dispatch of more capacitors. The system, with support from the synchronous condenser installation, can withstand the contingency up to the same 70% of peak load. Clearly, the three worst contingencies tested thus far are the loss of the 115 kV N-186 line coming from New Hampshire (T0+), the loss of the Woodford-Manchester 46 kV line (T0+), and the loss of the Woodford-South Shaftsbury 46 kV line (steady state). Moreover, all three of these, when managed with the help of the synchronous condenser installation, can be withstood up to 70% of peak winter load. But the results presented thus far provide little insight as to the load limit of the existing system for these same three contingencies. We only know is that the existing system is quite incapable of withstanding them at or above 70% of peak load. Entries 35 through 37 are intended to provide a better sense of the existing limit. All three contingencies are simulated at 50% of peak winter load. This load level still does not correspond to the precise limit, but is close enough that we may infer its value from the associated margins and voltages. Entry 35 assumes loss of the 115 kV N-186 line (T0+) at 50% of peak winter load. All voltages are within criteria. The reactive power margins at Stratton and West Dummerston, although positive, are below par. The real power margin is also positive but below par. Based on extensive simulation experience with this system, it is estimated that the load limit at which these criteria would be at least marginally acceptable is 40% to 45% of peak winter load. Entry 36 assumes loss of the Woodford-Manchester 46 kV line (T0+) at 50% of peak winter load. All voltages are within criteria. The reactive power margin at Stratton, although positive, is below par. It is estimated that the load limit at which this criterion would be marginally acceptable is 45% of peak winter load. Entry 37 assumes loss of the Woodford-South Shaftsbury 46 kV line (steady state) at 50% of peak winter load. All voltages are within criteria. The reactive power margins at South Shaftsbury and Stratton, although positive, are below par. It is estimated that the load limit at which these criteria would be at least marginally acceptable is 45% of peak winter load. Entry 38 is a special test of the systems potential for overcompensation, a condition caused by overreliance on shunt capacitors to prop up voltage as demand increases. Appendix G explains this phenomenon in more detail. 28

In summation, it appears that the existing systems load limit is approximately 45% of peak winter load for the three worst contingencies. Our earlier analysis indicates that the addition of the synchronous condenser installation increases this limit to approximately 70% of peak winter load for the same three contingencies. The derivation of these load limits is useful, yet they still fall short of providing an intuitive sense of our exposure to voltage collapse, and the degree to which the synchronous condenser installation provides mitigation. Better insight may be gained by correlating the load limit information with loadduration information, as in figures 6a and 6b. Both figures depict the same load duration curve. A load duration curve is simply a cumulative record of system load levels over a period of time (typically one year). The longer the time spent at a given load level, the flatter the slope of the curve will be in the vicinity of that time value. The load-duration curve depicted in figures 6a and 6b is that of the Southern Vermont system, including the Bennington and Brattleboro areas and the Southern Loop that connects them. Duration is noted as a percentage on the horizontal axis and load level is noted as a percentage on the vertical axis. The load-serving capability (i.e. the load limit) of the existing system is depicted in Figure 6a, and that of the system with the synchronous condenser installation is depicted in Figure 6b.

120.0

Southern Loop / Brattleboro Existing Load Duration Curve With Exposure Duration for Existing System
Critical Load Level (%) = 45.0

120.0

Southern Loop / Brattleboro Existing Load Duration Curve With Exposure Duration for Option S
Critical Load Level (%) = 70.0

100.0

100.0

Load (% of peak)

80.0

80.0

81 Mw load-serving capability (70% of peak, or 81 Mw)

60.0

60.0

52 Mw 40.0 load-serving capability (45% of peak, or 52 Mw)

Load (% of peak)

40.0

20.0

20.0

Figure 6a
0.0 100.0 0.0

Figure 6b
Duration (annual %)
100.0

Duration (annual %)

Peak Load of Base Curve (MW) = Exposure Duration of Base Curve (%) = Load Factor of Base Curve (%) =

116.0 65.9 52.3

Peak Load of Base Curve (MW) = Exposure Duration of Base Curve (%) = Load Factor of Base Curve (%) =

116.0 11.7 52.3

Recall that the existing systems load limit was found to be approximately 45% of peak, as indicated in Figure 6a. This load limit defines a vertical dividing line that separates the period of contingency vulnerability on the left, with the period of invulnerability on the right. The period of contingency vulnerability is known as the exposure duration. The exposure duration of the existing system is presently 65.9%, as indicated in Figure 6a. Notice that in Figure 6b the exposure duration has been 29

reduced to only 11.7% of the time. Today, if one of the three critical contingencies occurs, there is only a 34.1% probability10 that the system will avoid a criteria violation that could lead to an outage. With the synchronous condenser installation, this probability rises to 88.3%11. The Southern Loops greatly improved odds of contingency survival are the main benefit of the synchronous condenser installation. The next logical consideration is whether this improvement would bring the Southern Loop into compliance with CVPS applicable reliability criteria. In short, the answer is almost. Appendix E (A New Probabilistic Reliability Guideline Based on the Equal Slope Criterion) provides an explanation for CVPS reliability guideline for systems like the Southern Loop, that is, non-bulk transmission. This guideline requires an exposure duration of no more than 8.8% for the Southern Loop. The exposure duration achieved with the synchronous condenser installation would be 11.7 %. This is nearly in compliance. However, future load growth would further widen the compliance gap. This is not of any particular concern because it is assumed that additional remedial steps will be taken within a relatively short time as part of a broader strategic solution, based on additional work by CVPS, VELCO, DPS, and various public participation committees. Earlier, it was noted that the key question before us is How effectively will the synchronous condenser installation, by itself, remedy the five root problems that afflict Southern Vermont? Our review of Table 6, our consideration of the resulting changes to the Southern Loops exposure duration, and our review of CVPS applicable reliability standards has lead to the answer. The synchronous condenser installation is very effective at solving the first root problem, although it falls just short of full compliance with reliability criteria. The next root problem that must be considered is restated below: Root Problem #2 - The present-day Brattleboro area system is vulnerable to an unplanned loss of a transmission line or a transformer 100% of the time. The synchronous condenser installation at Stratton would be so electrically distant from Brattleboro as to have no appreciable effect on this root problem. The Brattleboro problem will have to be addressed by the later steps of whatever strategic solution is eventually implemented. The next root problem that must be considered is restated below: Root Problem #3 - The Southern Vermont system (and the South Western New Hampshire system) are vulnerable to a long-term outage of the T4 345/115 kV transformer at the Vermont Yankee nuclear plant. The synchronous condenser installation at Stratton would be so electrically distant from the T4 interconnection to the 115 kV system as to have no appreciable effect on this root problem. The T4 problem will have to be addressed by the later steps of whatever strategic solution is eventually implemented. The next root problem that must be considered is restated below: Root Problem #4 - The Southern Vermont system will soon be unable to supply peak demand even with all facilities in service, due to demand growth. Table 6 entry 2 can provide some insight here. Entry 2 simulates the Southern Vermont system at 100% of peak winter load with the proposed synchronous condenser installation operating. Voltages are excellent and reactive power margins are all ample (more than five times the minimum requirement). Most revealing of all, the real power margin is 46.3 Mw or 41% of peak load. This
10 11

100% - 65.9% = 34.1% 100% - 11.7% = 88.3%

30

means that the system, reinforced with the synchronous condenser installation, has the capability to supply almost half again the amount of load it now supplies with all lines in, with at least 90% voltage. This does not necessarily mean that load could grow by that amount with no operating difficulties. First, the real power margin assumes uniform load growth, but we know from our Southern Vermont load growth modeling efforts that a uniform growth assumption is unrealistically optimistic (see Appendix F for more detail on load growth modeling assumptions). Load growth concentrated near the ski resorts, as expected, would be more burdensome on a per-megawatt basis. Second, a 90% voltage would not be acceptable with all facilities in operation. And third, this data provides no indication of contingency performance with load growth. However, the real power margin does provide some indication of the systems capability to cope with load growth with all facilities in service, which is the concern with Root Problem #4. Therefore it seems likely that the system, assisted by the synchronous condenser installation, could cope with several tens of percent load growth in Southern Vermont. Given the average rate of growth, this would translate into several years time, enough to implement the later phases of a strategic remediation plan. The last root problem that must be considered is restated below: Root Problem #5 - Future problems that are related to the Southern Vermont system, will soon emerge on the wider regional transmission system. The synchronous condenser installation at Stratton would be so electrically distant from the problems cited on the wider regional system as to have no appreciable effect on this root problem. This may be partly addressed by the later steps of whatever strategic solution is eventually implemented, and partly by the regional plans of ISO New England, VELCO, and PSNH. Other Reliability Considerations As discussed in section V of this report (Study assumptions, methodology, and criteria), CVPS requires that its systems have adequate angular stability, also referred to as machine stability. No undue angular stability problems exist today in the Southern Loop or Brattleboro areas, nor would any be expected as a result of installing synchronous condensers near Stratton. Appendix B provides a more detailed assessment of this issue. Another reliability consideration is the synchronous condensers need for inductive capability as opposed to the all-important capacitive capability. Inductive capability is required in cases where the system voltage may rise too high rather than fall too low. High voltage tends to be less problematic on the Southern Loop than low voltage, but was considered nonetheless. Entries 39, 40, and 41 in Table 6 help to clarify the issue of high voltage suppression through inductive synchronous condenser operation at Stratton. Together, these two entries represent the worst conceivable situation for high voltage, one in which a relatively high load at a weak portion of the system becomes separated from the capacitors intended to support it, due to a contingency. This leaves the capacitors free to push the (now even weaker) radial system that remains, to a very high voltage because there is little countervailing load to restrain its rise. This scenario is difficult to imagine at higher load levels because loading tends to be heavy everywhere, precluding localized post-contingency voltage spikes, as all capacitors have sufficient load and losses nearby to curb their excesses. At light load however, there is a significant disproportion in loading between the ski resorts (very light) and larger towns (not so light). In this circumstance, the use of ski resort capacitors to prop up an adjacent towns voltage can set the stage for a severe voltage rise, as a contingency can now separate the capacitors from their intended load.

31

Accordingly, entries 39, 40, and 41 simulate light load (50% of peak). The relatively heavy town loads described above are at Manchester, Arlington, and East Arlington. These are supported by capacitors dispatched at Bromley (1.8 Mvar), Stratton (5.4 Mvar), and Rawsonville (5.4 Mvar), although the loads at these locations are themselves very light. Steady state voltage criteria were temporarily suspended in order to stress the system as severely as possible. The pre-contingency 46 kV voltages recorded in entry 39 are therefore unusually high (1.06 per-unit at South Shaftsbury, 1.10 per-unit at Stratton, and 1.07 per-unit at West Dummerston). The synchronous condensers were assumed to be off in order to permit this extreme deviation. In entry 40 the assumed contingency is the loss of the 46 kV line segment between East Arlington and Wallace tap (post-restoration), which isolates the load concentration to the west from the cluster of capacitors to the east. The synchronous condensers at Stratton remain off for now, to reveal the effect of the contingency alone. Its effect is severe - the voltage at Stratton rises to a blistering 1.17 per-unit, a full seven percent above the emergency criterion. The synchronous condenser installation is brought on-line in the following entry (41) to gauge its ability to suppress the overvoltage. Less than 9 Mvar of inductive absorption was required to bring all voltages into compliance with post-contingency criteria (and in fact also in compliance with precontingency criteria). Although 9 Mvar of inductive power capability therefore appears to be adequate, it is judged that requiring more capability is prudent, given the brevity of the overvoltage analysis and the possibility of unforeseen difficulties as area load grows and changes. Therefore 15 Mvar is judged to be a reasonable value, and is fairly typical when specifying synchronous condenser capability (inductive capability of one-half the capacitive value). XI Capital, O&M, and loss costs of a synchronous condenser installation (alone) The total capital cost of the synchronous condensers, including the associated line and substation work is estimated to be approximately $10 million. Testimony by Ryan Johnson provides a more detailed breakdown of this total. O&M costs are dependent to a large degree on the level of contracted services provided by the vendor(s) of the equipment, which have yet to be determined. Combined frictional losses, windage losses, and average electrical demand of ancillary devices are estimated to be on the order of 350 kW (continuous), with the synchronous operating most or all of the time. XII Recommendations Because a 30 Mvar (capacitive) synchronous condenser installation at Stratton is common to all of the most cost-effective strategic solution options, and, because it would substantially improve near-term reliability and all-lines-in adequacy, its installation is recommended along with all necessary ancillary equipment. Considerations regarding modularity and redundancy have lead to the conclusion that the optimal size and number of machines is two, each with the capability to supply the 46 kV system with 15 Mvar (capacitive) and 7.5 Mvar (inductive), which are the recommended capabilities. The specific designs of the machines step up transformers (GSUs) and of the machines own terminal capabilities must be coordinated to meet this 46 kV requirement. Further study and discussion is required to determine the next steps of a broader strategic plan for southern Vermont as prescribed by the results of Docket 7081.

32

Glossary

33

Note: Items in blue text are defined elsewhere in this glossary.

248 application process - The 248 application process is a regulatory process based on the Vermont law known as Section 248 (30 V.S.A. 248). It is administered by the Vermont PSB to consider proposed projects by utilities serving Vermont, and to determine whether they are in the public interest. Successful applicants are granted a CPG (Certificate of Public Good) which is essentially a license to build and operate the proposed project. The 248 process permits interested parties that qualify by statute and by procedure, to have a say in the hearing process and to influence the decision by the PSB to issue a CPG, a conditional CPG or a notice of denial. ANR (Agency of Natural Resources) - ANR is Vermont states agency in charge of protecting air, water, and soil quality, as well as wildlife. ASC (Area Specific Collaborative) - The ASC is a provisional association of interested parties organized under PSB Docket No. 6805 to jointly develop an integrated, least-cost solution to the Southern Loops problems. It includes representatives from Vermonts DPS, the Stratton and Bromley ski resorts, and CVPS. Breaker - See the definition of circuit breaker below. Capacitor - A capacitor is a device located in a substation that supports voltage within a local area. It has no moving parts and is relatively inexpensive. Capacitors are often added to transmission and distribution systems to keep the voltage acceptably high as electric demand grows over time. Overreliance on this strategy may lead to a system that has high enough voltage, but with poor stability, meaning that its voltage is too easily changed by the daily cycle of demand ramping up and down. This forces utility personnel to constantly re-adjust which capacitors are on and which are off, and makes the system vulnerable to a blackout. Capacity or capability - Capacity is the maximum demand that an electrical component or system can carry without overheating. Circuit breaker - A circuit breaker is a large switch that turns utility equipment on or off. It may be operated manually in order to safely perform preventative maintenance on equipment, or it may operate automatically to turn off equipment that is malfunctioning (see also the definition of fault below). Conductor - The conductor is the part of a transmission line that actually carries the electricity, in other words, the wire itself. The wire or conductor is just one part of a transmission line; other parts include the poles and the insulators from which the conductor is hung. A conductor must have enough capacity to carry the highest demand that it will experience, or it could overheat and fail. Contingency - A contingency is an unplanned outage of a critical system component such as a transmission line, transformer, or generator. The time prior to the contingency is referred to as precontingency and the time after it has begun is referred to as post-contingency. Controlled blackout - A controlled blackout is simply the same thing as a rolling blackout. CPG (Certificate of Public Good) - A CPG is a document that may be granted by the Vermont PSB at the conclusion of a Section 248 application process. It is essentially a license to build a proposed project that has been applied for (such as a new transmission line or substation) and signifies the PSBs conclusion that the project is in the best interests of the public. CVPS (Central Vermont Public Service) - CVPS is an investor-owned and state-regulated electric company responsible for the purchase or generation of electricity and its delivery to customers within its Vermont franchise area. 34

Demand - Demand is the amount of electricity being used at any given moment by a single customer, or by a group of customers. The total demand on a given system is the sum of all of the individual demands on that system occurring at the same moment. The peak demand is the highest demand occurring within a given span of time, usually a season or a year. The total peak demand that a transmission or distribution system must carry sets the minimum requirement for its capacity (see also the definition for energy). Demand factor - Demand factor is the average demand of a customer or system, divided by the highest (i.e. peak) demand of that same customer or system. The peak demand determines how much infrastructure is needed; the demand factor determines how fully that infrastructure is being utilized over time. DG (Distributed Generation) - Distributed generators are relatively small, dispersed electric generators that are intended to serve local electrical demand. They have come into greater use in recent years to satisfy gradual growth in demand, without the need to periodically build expensive, obtrusive, transmission lines. They may be owned and operated by the local utility or by individual customers. Typically, they are driven by gasoline or diesel engines, or by renewable power sources such as solar, wind, and running water. Depending on their size, they may be connected to a transmission system or to a distribution system. Dispatch - Dispatch is the act of turning on or turning off system resources that are needed in varying amounts over time, such as generators and capacitors. Such resources are said to be dispatchable. Distribution - Distribution lines and distribution substations operate at lower voltage than the transmission systems that feed them. They carry relatively small amounts of electricity to local customers. Distribution lines use shorter poles, have shorter wire spans between poles, and are usually found alongside streets and roads, or buried beneath them. Typical distribution voltages include 12.5 kV and 4 kV. DPS (Department of Public Service) - The DPS is Vermont states public advocate in legal proceedings and other forums that involve utility regulation, statues, and related issues. DPS staff often specialize in specific areas such as engineering, economics, or law. DSM (Demand Side Management) - Demand side management, like DG (distributed generation), is intended to satisfy local growth in electrical demand without the need to build new transmission lines. However, it differs from DG in that it strives to reduce the demand itself rather than to increase the supply. DSM generally falls into one of four categories: 1. Conservation measures such as replacing standard light bulbs with high-efficiency light bulbs, or adding extra insulation to buildings. 2. Special utility rates and contracts that encourage customers to conserve energy and/or to move their electrical use to those hours when overall electrical demand tends to be low, in order to avoid overburdening the transmission system that supplies that demand. An example of this is an interruptible rate that provides the customer with a discount in exchange for the utilitys right to interrupt the associated demand when system reliability or economic considerations necessitate it. 3. Load control systems, also known as demand response systems, that disable non-essential customer appliances (e.g. hot water heaters) during high-demand hours. 4. Fuel switching measures that replace electricity-powered heating, cooking, and drying equipment with fuel-powered equipment, such as the replacement of electric ovens with gas ovens. DUP (Distributed Utility Planning) - Planning method that seeks to find the lowest cost of providing reliable energy delivery through traditional means such as transmission, as well as newer approaches such as DG and DSM. Often, these strategies are used in combination. 35

Efficiency Vermont - Efficiency Vermont is Vermonts energy efficiency utility and administers programs under contract with the PSB that conserve energy by utilizing it more efficiently (see also the definition of DSM above). Energy - Energy is the ability to do work. Energy comes in many forms (electrical, chemical, thermal, mechanical, etc). It is measurable in common units, regardless of which form it is in. The rate at which energy is made, used, transformed, or transferred is called power or demand. Energy is expensive to produce and therefore should not be wasted. Fault - A fault is the failure of a line, transformer, or other electrical component. Once such a component has failed (due to overheating, short-circuiting, physical breakage, or other trauma) it is automatically taken out of operation by a circuit breaker that quickly turns the component off. Once it has been tripped off (in the parlance of engineers) it no longer poses a threat to human safety, but its loss may present a difficult burden to the remaining system (see also the definition of redundant below). Generation or Generator - A generator is a device that converts mechanical power from an engine, a water wheel, a windmill, or other source, into electrical power. Generators have internal parts that spin as they make electricity, similar to an electric motor. Hydro - Hydro is electric generation driven by running water such as streams or rivers. Island or Islanding - An island in utility parlance, means an area of the electrical system that is electrically cut off from the main system by switches or by disabled equipment, but that is able to serve its own demand by means of generation located within its own boundaries. This requires that the generation have special controls to match its output to the islands ever-changing demand, and also requires that there be enough generation to satisfy the islands total demand. Islanding is the process of disconnecting from the main system and re-establishing service using these specialized generators. ISO New England Inc. - ISO New England Inc. is the Independent System Operator for all of New England, and is responsible for the coordinated planning, PTF funding, and operation of the transmission system, as well as reliability oversight of generators and other electrical facilities. ISONE is also responsible for the administration of New Englands power supply markets (in which utilities make bids or exercise contracts for other companies generation to meet their own customers demand). kV (kilovolt) - A kilovolt is a thousand volts. Volts and kilovolts are measures of voltage. As an example, the Southern Loop subtransmission line that runs from Bennington to Brattleboro operates at 46 kV or 46,000 volts. Load - Load is simply the same thing as demand. Load Duration Curve - A load duration curve is a mathematically-based graph depicting the magnitude of load (i.e. demand) over a long period of time, usually one year. The graph does not show the constant up and down movements of daily or weekly demand cycles. Instead, it is a continuous function that transitions smoothly from peak demand to minimum demand over the given duration. This is because the individual data points, adjacent to one another along the curve, are sequential in their demand value but are not sequential in their time value. The flatter the slope of the curve in a given vicinity, the greater the duration of the associated demand value. The demand near the middle of the load duration curve is commonly referred to as shoulder load because of its resemblance to a human shoulder. Appendices E and H provide example load duration curves. Load factor - Load factor is simply the same thing as demand factor.

36

Losses - Losses are wasted electrical energy. All components and systems that carry electricity waste a small amount of its energy. This wasted energy is given off as heat to the surrounding air. Losses cost money, but can be minimized by sound engineering practices. Mw (Megawatt) - A megawatt is a million watts. Watts and megawatts are measures of demand. To put this in perspective, the peak demand for the state of Vermont is approximately 1,100 Mw or 1,100,000,000 watts. N-0 or N-1 or N-2 - The term N minus zero (or one or two) refers to the failure of important equipment. Although these terms sound complex, they are actually quite simple. N is the total number of components that the system relies on to operate properly. Only rarely does anyone try to calculate its actual value; it is simply a generic term to describe all the components of a given system. The number subtracted from N is the number of components that may fail in a given scenario, although more information is needed to denote just what component or components are assumed to have failed. Therefore, N-0 means that no components have failed and the system is in a normal condition. N-1 means that only one component has failed. N-2 means that two components have failed at the same time, which is generally worse than having only one fail (see also the definition of contingency above). Network - A network line is one that is capable of carrying power in either direction, similar to a twoway street. Most transmission lines are network lines, while most distribution lines are not (see also the definition of radial below). Peaking Generation - Peaking generation is generation that is designed to run only a limited number of hours per year, during periods of high demand. Power - Power is simply the same thing as demand. PSB (Public Service Board) - The PSB is Vermont states quasi-judicial authority in legal proceedings that involve utility regulation, statues, and related issues. It consists of 3 principal commissioners (appointed by Vermonts governor) and supporting staff. PTF (Pool Transmission Facility) - The precise definition of a pool transmission facility is beyond the scope of this document but, generally speaking, it is any transmission facility operating at 69 kV or higher that is networked (not radial). PTF falls under the authority of ISO New England. The construction of new PTF facilities is funded by the ISO on a pro-rata basis among its member utilities. Vermonts responsibility for such costs is about 5% of the total. Radial - A radial line is one that is capable of carrying power in only one direction, similar to a one-way street. Most distribution lines are radial lines, while most transmission lines are not (see also the definition of network above). Redundant - Facilities that have backups or alternate ways of operating are said to be redundant, that is, their function can be sufficiently provided even after they suffer a breakdown or failure. The more crucial a component or system, the greater the need for it to be redundant. Renewable power source - A renewable power source is any power source that does not rely on a finite resource to keep it running, such as coal, oil, or natural gas, which will eventually run out. Renewable power sources include solar collection systems, wind mills, and hydro generators, because sunlight, wind, and running water will never run out. Generators than burn replaceable fuels also qualify as renewable power sources. Examples include bio-diesel generators that run on crop-derived fuels, and wood-burning generators. Rolling blackout A rolling blackout is the deliberate cutting off of electric service to a limited number of customers during conditions of dire system problems. These blackouts are targeted at different groups of customers, first one and then the next, and continuing this way in a cyclical pattern in order

37

to spread the pain as evenly and fairly as possible until the problem is fixed. A rolling blackout may also be referred to as a controlled blackout. ROW (Right-of-Way) - A right of way is the long but narrow strip of property on which a transmission line is built. It may be owned by the associated utility or it may owned privately, with the utility exercising its state-mandated right to use this private property for the public good. Substation - A substation is a fenced-in area where several transmission and/or distribution lines come together and are connected by various other equipment for purposes of switching, metering, or manipulating voltage. Often they contain transformers. Subtransmission - Subtransmission systems are very similar to transmission systems (see also the definition of transmission below) and differ only in that they operate at somewhat lower voltage and carry smaller amounts of power. Typical subtransmission voltages include 46 kV and 34.5 kV. Synchronous Condenser - A synchronous condenser is a device located in a substation, that supports voltage on electric transmission or distribution systems, much like a capacitor. But unlike a capacitor, a synchronous condenser has moving parts, that is, it spins like a motor, and its outward appearance is very similar to that of a large motor. Synchronous condensers tend to be more effective than capacitors but also more expensive. They run for very long periods of time and produce a humming sound, but do not burn fuel and therefore do not release emissions into the atmosphere. Thermal - This term is related to the terms capacity or capability. Thermal refers to heat or temperature, which are of concern when electrical equipment is carrying a high demand. Electrical components that exceed their capacity or capability are said to be thermally overloaded, meaning that they are carrying too much demand, are growing too hot, and could fail as a result. Transformer - Transformers are the on-ramps and off-ramps of the transmission highway. Specifically, a transformer is a device located in a substation that connects high-voltage equipment to low-voltage equipment and allows power to flow from one to the other. Different voltages are used because higher voltages are better for moving electricity over a distance, but lower voltages are better for using electricity in machinery and appliances. Transformers (and the substations in which they reside) are commonly described by the two (or more) voltages that they connect, such as 115/46 kV, signifying a connection between 115 kV and 46 kV equipment. Transmission - Transmission lines and transmission substations operate at high voltage and carry large amounts of electricity from centralized generation plants to low voltage distribution lines and substations that supply small towns and localities. A few transmission lines or even one may be capable of supplying an entire region or metro area. Transmission lines use very tall poles or towers, have long wire spans between poles, and usually traverse fairly straight paths across large distances. They do not tend to follow roads. Typical transmission voltages include 345 kV, 115 kV, and 69 kV. VELCO (Vermont Electric Power Company) - VELCO is a transmission company wholly owned by Vermonts distribution companies, and responsible for the planning, construction, and operation of Vermonts transmission system. Voltage - Voltage in an electric transmission or distribution system is much like water pressure in a system of pipes. If the pressure is too low, the pipes cannot carry enough water to satisfy the needs of those connected to them. If the voltage is too low, the electric system cannot carry enough electricity to satisfy the needs of those connected to it.

38

Appendix A An Explanation of QV and PV Analyses

39

Understanding QV characteristics (reactive power margin) QV characteristics or QV curves are mathematical functions used to describe the relationship between a transmission systems reactive power demand, its reactive power supply, and its operating voltage. Q refers to quadrature power, also known as reactive power. V refers to voltage. Figure 1 is a typical or generic QV demand curve for a transmission line, with reactive power on the vertical axis and voltage on the horizontal axis. This function depicts the reactive power demanded by the system, at a certain bus, to maintain a given voltage. In the right-hand region, the curve displays classic voltage response (i.e. more injected reactive power corresponds to higher voltage). In the left-hand region, the depressed voltage results in very high line current (assuming that constant-power loads are being fed). In this region, more reactive power injection will only depress voltage further, as the linearly-increasing current causes exponentially-increasing reactive power losses, which begin to overwhelm the system.

The QV demand curve may be overlaid with a QV supply curve, as in Figure 2. This additional curve depicts reactive power supply available, at a certain bus, as a function of voltage. If shunt capacitors are dispatched at or electrically near to this bus, the supply curve will be positively-sloped as indicated in the figure, due to the voltage dependency of the capacitor. If no shunt capacitors are nearby, the supply curve will essentially be a flat horizontal line at the zero value. The two points where the two curves intersect represent states of equilibrium between reactive power supply and demand. Theoretically, the system may operate at either point. In practice however, the left-hand point is dynamically unstable and the right-hand point is dynamically stable. Physical operation at the unstable equilibrium point is analogous to placing a round marble on a sharp peak, where even a tiny perturbation will cause it to permanently roll away from its origin, as in Figure 3. In contrast, physical operation at the stable (right-hand) point is analogous to placing a marble in a V-shaped valley, where perturbations will cause it to temporarily move about, but invariably return to its origin.
40

Loadflow programs do not recognize this distinction, as they are based on mathematical equations intended to represent steady-state operation, not dynamic operation. Consequently, they may occasionally stumble upon a solution (i.e. an equilibrium point) that is dynamically unstable, and therefore operationally unsustainable. Engineers conducting loadflow simulations must learn to recognize and address such occurrences. Of great significance is the distance on the reactive power axis between the stable operating point and the minimum value of the demand curve, as in Figure 4. This quantity is referred to as the reactive power margin of the system, and is one measure (at least locally) of the systems robustness or strength12. If the reactive power supply is diminished for some reason, this margin will shrink and the voltage will decline as the equilibrium point moves leftward and downward as in Figure 5. If the supply is diminished to the point that there is no longer an intersection between the supply and demand curves, then the system will experience a voltage collapse, leading to a blackout. QV analysis is simply the quantification of these curves, and of the reactive margin, as a means of assessing the robustness of systems that may be suffering from reactive power or voltage problems. This quantification is achieved through special loadflow modeling techniques that force differing voltage levels at a given bus (the control bus) by injecting differing amounts of reactive power at that same bus. More than one such bus may be used, either simultaneously or individually, and dispersed about the area of interest.

Notably, if the minimum value of the demand curve occurs below 0.90 P.U. voltage then the reactive power margin is taken as the distance between the stable operating point and the demand value at 0.90 P.U. The reason for this practice is that the reactive power margin is really a safety margin, all or most of which may be lost with acceptable consequences. If the minimum demand value falls below 0.90 P.U., a loss of reactive power to this value will not necessarily cause voltage collapse, but will cause unacceptably low voltage. Therefore, in such cases, the margin is based on the minimallyacceptable voltage in order to avoid overstating its benefit. 41

12

A sufficiently large reactive margin is one of the requirements (but not the only one) for a robust transmission system. However, there are no widely recognized criteria that define quantitatively what the margin should be. Generally, planning engineers establish these criteria on an ad hoc basis, taking into account preliminary QV results, system connectivity, system impedances, voltage class, relative burden (load) of the area under study, and engineering judgment. Note that QV curves may change shape and position due to load variation and equipment failures (contingencies), and due to automated or supervisory actions such as distribution transformer/regulator tap changes, network switching, and generation redispatch. However, the specific effects of each of these factors are quite complex, and beyond the scope of this discussion. Last, although QV curves are useful visual aids for presentations or training, the actual graphing of QV curves is usually dispensed with, in favor of simply reporting their critical parameters. These include the location of the control bus, its operating voltage, its voltage at minimum reactive power, and its reactive power margin. Understanding PV characteristics (real power margin) PV characteristics or PV curves are mathematical functions used to describe the relationship between a transmission systems real power transfer, and its operating voltage. P refers to real power. V refers to voltage. PV curves are somewhat more easily explained and understood than QV curves. Figure 6 is a typical or generic PV curve for a transmission line, with voltage on the vertical axis and load or power transfer on the horizontal axis. This function depicts the voltage at a certain bus (or the worst voltage in a defined area), for a given amount of load or power transfer carried by the line. The downward curvature of the function as loading increases is due to the non-linearity of the equations governing voltage drop. Like the QV demand curve, the PV curve has a

42

stable region, as well as an unstable region (dotted portion of curve) that may emerge in computer simulations, but cannot exist in actual steady-state operation.

Analogous to the reactive margin in QV curves is the real power margin in PV curves. This is the distance on the real power axis between the actual operating point of the system, and, either the maximum power value of the curve itself, or the power value at which the curve reaches unacceptably low voltage (typically 0.90 per unit). Which definition is used depends on whether the maximum value of power transfer is still within the acceptable voltage range. Real power margin is another measure of transmission system robustness. As with reactive power margin, planning engineers generally establish the criterion for real power margin on an ad hoc basis, although it is generally in the range of 5-10%. Truly robust transmission systems must have both sufficient real power margin and sufficient reactive power margin. Although there are other measures of overall transmission robustness (including such things as angular stability margins), these two measurements, in combination, provide a very clear picture of the reactive power and voltage robustness of transmission systems, at least within a locally-defined electrical area. This current Southern Loop / Brattleboro Area Reliability Study has assessed both of these measures of strength and offers appropriate recommendations to keep them adequate. In addition, the study incorporates more conventional analytical techniques and criteria, such as thermal and voltage limits.

43

Appendix B Stability Discussion of Synchronous Condensers at Stratton

44

Generally speaking, there are two chief concerns with respect to angular stability when contemplating the connection of new rotating machines to an electrical transmission or distribution system. The first of these two concerns is the possibility that the proposed rotating machine(s) may case adverse impact to existing apparatus, in the form of angular instability. This necessarily requires that such apparatus be in reasonable electrical proximity to the proposed installation, in order for an adverse impact to potentially exist. Furthermore, it requires that any such apparatus be rotational in design, as nonrotating equipment has no mechanical angle to speak of and is therefore not subject to angular problems. Finally, these rotating machines must be of AC synchronous design, not AC induction or DC design, because AC induction and DC machines are essentially impervious to machine angle problems such as transient or oscillatory disturbances (although they may still be vulnerable to voltage excursion or power quality problems, which are different concerns altogether). The second of these two concerns is the possibility that the proposed rotating machine(s) may itself be vulnerable to angular instability, given the unique operating environment into which it is being placed. The first concern, explained above, is of no consequence on the Southern Loop because there are no other synchronous machines within reasonable proximity of Stratton, such as those associated with hydro generators or specialized manufacturing facilities, which might be vulnerable. The second concern, also explained above, is generally negligible in the case of synchronous condensers because they have no mechanical coupling and therefore no shaft torque, unlike synchronous generators. It is conceivable that a synchronous condenser could experience a slippage of one or more poles during a protracted and severe transient disturbance, but would be expected to quickly return to stable operation after the disturbance ended. A loss of synchronism would only be possible if the disturbance was so protracted that it resulted in a significant reduction in machine speed prior to system recovery. Despite the unlikelihood of possible stability problems, in 2003 VELCO performed a brief stability analysis of a generic 30 Mvar synchronous condenser connected to the 46 kV system at Stratton. This analysis found no adverse impacts or undue system responses as a result of adding the two 15 Mvar synchronous condensers at Stratton.

45

Appendix C Basic Probability Theory

46

INTRODUCTION Probability is a broad and complex topic within the field of mathematics. Although the study of probability can be quite challenging, the average utility engineer needs only to master a few of its basics to perform most "probabilistic" engineering studies. This paper is intended to briefly introduce these basic concepts, through discussion and example calculations. BASIC PROBABILITY EQUATION Although recurring events may occur at random intervals, and for random durations, the average of these values may be calculated from historical records. These averages may then be used in various probability calculations. Basic probability equation: Prob. of being in an event state = [ave. freq.][ave. duration] total time Examples: Event 1 ave. freq. = 5/yr. ave. duration = 20 hrs. P1(Prob. of being in Event 1 state) = [5][20] = 100 8760 8760 = .0114 = 1.14% That is, because the total yearly average duration of the Event 1 state is 100 hours and because there are 8760 hours in one year, it is expected that the Event 1 state would exist for 100/8760 or 1.14% of the time. Thus, the probability of being in the Event 1 state, at any given time, is .0114. Event 2 ave. freq. = 7/yr. ave. duration = 10 hrs. P2(Prob. of being in Event 2 state) = [7][10] = 70 8760 8760 = .00799 = .799% Although Event 2 is slightly more frequent than Event 1, its probability is actually lower, because its average duration is only half that of Event 1. That is, Event 2 averages only 70 hours of annual event time, as opposed to 100 hours for Event 1. Thus, the probability of being in the Event 2 state is lower. Another useful quantity is "mean time between events". It is the average time between the successive starting points ("initiations") of some recurring event. Returning to our Event 1 example: Event 1 ave. freq. = 5/yr Mean time between events = 8760 = 175 5 47

Thus, mean time between events may be calculated by simply taking the reciprocal of the average frequency. COINCIDENCE OF MULTIPLE EVENTS Often, we are interested in the occurrence of multiple coincident events, that is, two or more events that have overlapping durations. Overlapping equipment failures are a typical case in point. Such occurrences may be either dependent or independent. In the dependent case, the occurrence of one event alters the probability of the occurrence of another. As an example, consider a transmission system with two parallel transmission lines, A and B. If line A fails, line B's loading will increase, causing greater conductor sag, which will make line B more failure-prone. Thus, the occurrence of Event A (failure of line A) has a direct effect on the likelihood of Event B (failure of line B). In the independent case, the occurrence of one event has no effect on the occurrence of another. As an example, consider the successive tossing of two coins, A and B. The occurrence of a "tails" result on coin A has no effect on the probability of a "tails" result on coin B. In reality, the degree of dependence between different events varies; if the degree of dependence is low enough, independence may be assumed as an approximation. As an example, consider two transmission lines A and B, which are widely separated. Loadflow simulations may demonstrate that the loss of line B changes the loading on line A, but only by an insignificant amount. Thus, these two events may be assumed to be independent. The remainder of this discussion will assume the independence of multiple events. Accordingly, the probability of being in a coincident event state, is simply the product of the probabilities of the underlying events. Returning to our Event 1 and Event 2 examples from page 1, recall that: P1 = .0114, P2 = .00799 thus, P1,2 (Prob. of being in a coincident Event 1, Event 2 state) = [P1][P2] = [.0114][.00799] = .0000911 In fact, the coincidence of Event 1 and Event 2 may be thought of as being a separate and distinct event system (Event 1,2) with its own average frequency, average duration, and mean time between (coincident) events. These quantities may be calculated from those we have already identified. We will proceed with the calculation of the average frequency of Event 1,2. Given that the frequency of Event 2 is 7 initiations per year, we may expect considerably fewer than this during the 100 hours/yr. that we are exposed to Event 1. Specifically, we would expect only 100/8760 as many initiations, or: 100/8760 X 7 = .08 That is, on average, we expect .08 initiations of Event 2, while Event 1 is already in progress, each 48

year. Conversely, we must also calculate the average annual number of initiations of Event 1, while Event 2 is already in progress: 70/8760 X 5 = .04 Thus, the total (coincident) average frequency for Event 1,2 is: .04/yr. + .08/yr. = .12/yr The mean time between events is simply the reciprocal, or 1 coincident event every 8.3 years, on average. It is interesting to note at this point, that an event may be assumed to have zero duration. When average event durations are very short relative to mean times between them, and relative to the durations and mean times of recognized coincident events, they may be approximated as being zero. If one or the other of our example events was treated this way, then one or the other of our two frequency terms (.04 or .08) would have a zero value instead. Given that both terms in our particular example make substantial contributions to the total coincident frequency, such an approximation would be ill-advised. Moreover, if both events were assumed to have zero durations, the total coincident frequency would be zero also. In essence, this is because differing events of zero duration have no overlapping spans, regardless of their individual average frequencies. The steps taken above, to calculate the total (coincident) average frequency for 2 events, are conveniently combined in the following equation: Total (coincident) Average Frequency

= [ave. freq.event 1][ave. freq.event 2][ave. dur.event 1 + ave. dur.event 2] total time
Having identified probability and average frequency values for Event 1,2, we may now calculate the average (coincident) duration of Event 1,2. We begin with the basic probability equation from page 1: P1,2 = [ave. freq.1,2][ave. duration1,2] total time Algebraic manipulation yields: ave. duration1,2 = [P1,2][total time] ave. freq.1,2 = [.0000911][8760] .12 = 6.65 hours In other words, the (rather infrequent) occurrences of Event 1 overlapping with Event 2, have an average duration of 6.65 hours. Interestingly, if the two events in question have equal average durations, the average duration of a coincident event will be 1/2 that of either of the individual average event durations, regardless of average frequencies. This is because wherever the two events coincide, their minimum possible 49

overlap is zero and their maximum possible overlap is the average duration of either one. Because the two events are independent, their coincidence shows no preference for any particular overlap duration. Thus, averaged over many coincidences, the duration gravitates toward the middle of the two possible extremes, or 1/2. As an exercise, the reader may wish to recalculate the average duration for Event 1,2, assuming that Event 1 now has the same average duration as Event 2, or 10 hours. (Don't forget to recalculate both P1,2 and ave. freq.1,2 ). As expected, the average duration of coincident events will now be 5 hours instead of 6.65 hours. CUMULATIVE PROBABILITY Although we have explained how to calculate the probability of being in a given event state at any point in time, a related question is, "What is the probability of experiencing one or more events within a given period of time?" This requires the calculation of cumulative probability. Consider Event 1 from our earlier examples; its average frequency was 5 events/year, or a mean time between events of 8760/5 or 1752 hours. Does this imply that after waiting 1752 hours we should necessarily expect an event to have initiated? No, because the frequency and duration values cited are only averages, not the parameters of a well-behaved periodic function. In fact, for any finite period of time, there is a chance, however small, that an event will still not have initiated. After an infinite time, however, we must assume that at least one event will have initiated; and so the probability of this having occurred is unity. Of great significance in cumulative probability calculations is the fact that average durations of events are of no consequence; only average frequencies are important. This is because we are only concerned with the initiation of an event at some point in the future and not with the time following that initiation. Accordingly, we always assume that at time t = 0 (the beginning of the time frame in which we are interested), we are looking ahead to the next event and are not concerned with past events. Any past event whose duration is of such length as to overlap with the current time frame, is rightly associated with the previous time frame where it was initiated. Similarly, any event which is initiated in the present time frame is associated only with this time frame, although its termination may occur in a later time frame. To reiterate, in cumulative probability calculations, durations are ignored. The function associated with the cumulative probability of events occurring randomly in time, is: F(t) = e-t Where = average frequency of the event (a constant) and t = time (a variable) The cumulative probability of having one or more initiations of the event is equal to the area under this function, for any given time span (see Fig. C1).

50

probability function (area under curve at t = infinity is 1.0)

F(t)

time Figure C1

This area is calculated by taking the integral of the above function; the Addendum to this appendix outlines the actual integration steps.

51

The area then is defined by the function: area = 1 - 1 et Note that at t = infinity, area = 1, which corresponds with our earlier premise that after an infinite time, an event must have occurred, and that its cumulative probability is therefore unity. Note also, that much of the area under the curve is accumulated within a relatively short time. This corresponds to the fact that after a long but finite time, it is very likely that at least one event will have occurred, but is still not a certainty. Also, the probability that no event will have occurred is proportional to the remaining area between t and infinity. Recalling that the coincidence of two events may be treated as a single event system, let us now calculate the probability of experiencing at least one event (Event 1,2 from our earlier examples) in a one year period. As calculated previously, Event 1,2 had an average frequency of .12/yr. Substituting the time and frequency values into the area function above yields: area = 1 - 1 e[.12][1] = .113 Thus, the probability of experiencing one or more (coincident) events within 1 year equals .113. The probability of experiencing no (coincident) events within 1 year is: 1 - .113 = .887 If we wish to know the probability of experiencing one or more such events within 10 years, there are several alternative calculation methods. The most obvious is simply to substitute the appropriate values into the area equation as before: area = 1 - 1 e[.12][10] = .699 Thus, the probability of experiencing one or more events within 10 years equals .699. Another method is to use the calculated values for the one year period and extrapolate. That is, if we know that the probability of not experiencing an event in one year is .887, then logically, the probability of "surviving" 2 years without an event is: [.887][.887] = [.887]2 = .787 In fact, we may take the cumulative probability value for any given time period and raise it to the power of the number of such time periods for which we wish to calculate a longerterm cumulative probability. Thus, for a 10-year period, the probability of having no event is: [.887]10 = .301 The probability of experiencing one or more events within 10 years is, therefore:

52

1 - .301 = .699 Note that this agrees precisely with our earlier calculation using the area equation. There is one important caveat, however. Notice that our extrapolation calculations focused on the probability of not experiencing events. Thus, we were calculating an everdeclining probability value for longer and longer time spans. We then subtracted from unity to calculate the alternate probability of experiencing one or more events. We could not have directly calculated this value as it is an increasing probability, and raising its probability values to higher powers would incorrectly yield lower, not higher, values. Thus, the key when extrapolating, is to calculate the declining probability, and subtract from 1 if the alternate probability is desired. Addendum We wish to find the definite integral of the function F(t) = e-t . This function is of the basic form eudu. The standard integration formula for such functions is:
u u e du = e + C

(this formula may be found in most integral calculus text books)

Therefore, we denote the definite integral of the desired function to be of the form:
t -t e 0

Following the standard integration formula yields:

t -t -t -0 t t e = (-e ) - (-e ) = (-1/ e ) - (-1) = 1-1/e 0

Therefore, the definite integral of F(t) = e-t is 1-1/et

53

Appendix D Why Chester-Londonderry No Longer Makes Sense

54

A Discussion of the Justification for Eliminating the Chester-Londonderry Solution Option for the Southern Loop ASC Effort, Docket # 6805
Prepared by CVPS T&D Planning April 18, 2003

At the March 26, 2003 ASC meeting, the possibility of a new Chester-Londonderry 46 kV line was discussed as a means of rectifying existing reliability problems on the Southern Loop 46 kV system in Southern Vermont. CVPS presented information at that meeting supporting its position that this solution option was probably not cost-justified, in comparison to other available options. Accordingly, CVPS representatives suggested that the Chester-Londonderry solution option be eliminated from further consideration in CVPS on-going Southern Loop Reliability Study. Before making this decision, the other members of the collaborative felt that they needed further information and requested that CVPS provide specific addenda. This report is the response to that request. Table D1 summarizes the results of loadflow simulations that quantify the reliability attributes of two possible solution options. These are (one) a new 46 kV Chester-

55

Londonderry line and (two) the installation of two 15 Mvar synchronous condensers at the Stratton 46 kV substation. Although other solution options are also in the running, this two-fold comparison will be sufficient to justify the elimination of the ChesterLondonderry option. The table is organized with independent parameters (i.e. inputs to the loadflow) in columns A-I, and dependent parameters (i.e. outputs from the loadflow) in columns J-R. Column S provides comments and other information. Specifically, column A denotes the loadflow case name, column B denotes system load level, column C denotes the percent annual duration at which the load is at or below the indicated load level, column D denotes the Southern Loop and Brattleboro area capacitor dispatch, columns E and F denote shed loads (none in any of these cases), column G denotes the tested contingency, column H denotes the time frame (all are steady-state in these cases), columns J-N provide voltage data, columns O-Q provide reactive power margins at various control buses, and column R provides real power margin. Appendix A provides a detailed explanation of reactive and real power margins as well as an explanation of the analytical techniques used to quantify them. Some readers may be unfamiliar with the term steady-state. This refers to a pre or post-contingency condition in which all system transients, short term responses, and long term responses have died out. This is as opposed to T0+, which refers to a post-contingency condition in which the assumed time frame is after the contingency and resultant breaker tripping, but prior to such longer-term responses as automatic transformer tap movement, automatic or supervisory capacitor dispatch, supervisory switching and reconfiguration, and generation re-dispatch. These later actions are intended to improve system performance, and typically do so. Therefore, in order to evaluate the worst possible condition that the system must survive, however temporary in nature, post-contingency simulations are usually performed at T0+, unless otherwise stated. The first three entries (i.e. first three rows of data) provide the results of non-contingency simulations at present-day peak winter load. The first entry is the present-day system without any reinforcement. Although the voltages on the Southern Loop are adequate, the reactive power margin at or near Stratton is insufficient to meet criteria (at least 6 Mvar), and the real power margin is just able to meet criteria (5%). The addition of the synchronous condensers in the second entry results in a dramatic increase in all available margins, and provides for a fully-compliant and robust system. In contrast, the addition of the Chester-Londonderry line provides only modest increases in the margins, although the system is still fully compliant in terms of its performance criteria. Entries four, five and six provide the results of contingency simulations at 70% of the present-day winter peak load, but are otherwise the same in terms of their independent parameters. The contingency presented is the loss of Woodford-South Shaftsbury. Other contingencies not reported in this table, have been run, and these provide different specific numbers, but similar comparative conclusions. The unimproved system simulation results in a non-convergent case, indicative of a voltage collapse (i.e. a

56

system failure resulting in a blackout of unknown scope). The synchronous condenser option fares much better, with all margins and other criteria satisfied. The ChesterLondonderry option is unable to satisfy the criterion for reactive power margin at South Shaftsbury and only barely satisfies it at Stratton. You may recall from the ASC meeting on March 26th, that the capital costs of the synchronous condenser installation and Chester-Londonderry options were estimated to be $3 million and $4 million respectively. This gives the synchronous condenser option a capital cost savings of $1.0 million over the Chester-Londonderry option. A brief analysis of the loss differences between these two options concluded that the 20-year present worth of future loss savings for the Chester-Londonderry option over the synchronous condenser option was approximately $0.5 million, leaving the synchronous condenser installation with a net present value cost savings of about $0.5 million (1.0 - 0.5 = 0.5). Therefore, given that the Chester-Londonderry option is more costly and of less benefit than one of the other options that is still in the running, it is clearly not cost-justified, and should be dropped from consideration. Moreover, the Chester-Londonderry option requires significant new ROW whereas the synchronous condenser option requires only a modest increase in the substations footprint.

57

D
Dummerston 2.7 Mvar 12.5kV 2.7 Mvar 12.5kV 1.8 Mvar 12.5kV 2.7 Mvar 12.5kV 2.7 Mvar 12.5kV 5.4 Mvar 46kV 5.4 Mvar 46kV 10.8 Mvar 46kV Londonderry 5.4 Mvar 46kV 10.8 Mvar 46kV 5.4 Mvar 46kV 5.4 Mvar 46kV 5.4 Mvar 69kV

Table D1 - Comparison of Chester-Londonderry and Synchronous Condenser Options

Rawsonville

Manchester

Shaftsbury

Woodford

Arlington

Jamaica

Bromley

Stratton

A
entry # 1 2 3 4 5 6 case name SLOOP-100%-calibrated SLOOP-B2 SLOOP-C8 SLOOP-B23 SLOOP-B24 SLOOP-C8aaaa

B
load level (ignoring any shed load) 100% of winter peak 100% of winter peak 100% of winter peak 70% of winter peak 70% of winter peak 70%of winter peak

C
% annual duration of equal or lesser loads 100% 100% 100% 88.3% 88.3% 88.3%

Brudies

Vernon

Vernon

E
deliberately shed interruptable load zero zero zero zero zero zero

F
deliberately shed non-interruptable load zero zero zero zero zero zero

G
contingency none (a basecase) none (a basecase) none (a basecase) Woodford-S Shaftsbury 46 kV Woodford-S Shaftsbury 46 kV Woodford-S Shaftsbury 46 kV

H
time frame steady state steady state steady state steady state steady state steady state

I
added reinforcements on 46 kV system none 30 Mvar sync condenser at Stratton 3 Chester-Londonderry 46 kV line none 30 Mvar sync condenser at Stratton 3 Chester-Londonderry 46 kV line

J
S Shaft 46 kV voltage (P.U.) 1.02 1.04 1.01

K
Stratton 46kV voltage (P.U.) 0.94 1.00 0.95

L
W Dumm 46kV voltage (P.U.) 0.98 1.00 0.99

M
lowest CVPS 46 kV voltage (P.U.)1 0.94 @ Stratton 0.99 @ Bromley 0.95 @ Stratton

N
highest CVPS 46 kV voltage (P.U.)1 1.04 @ Woodford 1.05 @ Woodford 1.03 @ Woodford

O
reactive power margin for QV control bus at S Shaft (Mvar) 19.1 49.1 16.7 -1.0

P
reactive power margin for QV control bus at Stratton (Mvar) 2.9 32.9 7.0 -1.1 28.9 6.8

Q
reactive power margin for QV control bus at W Dumm (Mvar) 11.8 25.6 18.9 -1.7 17.6 16.5

R
real power margin (Mw)11 5.7 (5% of peak) 46.3 (41% of peak) 17.0 (15% of peak) negative <11.3 (<10% of peak)12 12.0 (11% of peak)

S
comments reactive power margin is positive but inadequate10

Southern Vermont capacitor dispatch Southwestern VT Southeastern VT

Chester 46 kV bus voltage drops from 0.972 to 0.966 w/ tie

Non Convergent Case 1.01 0.96 1.00 0.98 0.98 0.99 0.97 @ Jamaica 0.96 @ S Shaft 1.04 @ Woodford 1.06 @ Woodford

9.2 2.5

real power margin would still meet minimum criteria11 reactive power margin is positive but inadequate10

1 3

West of Vernon Rd (i.e. Southern Loop system) Actually 15 Mvar x 2 units, both of which would be located at the Stratton 46 kV bus The precise value that constitutes "adequate" reactive power margin is a judgement call. A typical standard is to require sufficient margin to withstand any single contingency coincident with loss of the largest reactive power source within the locali reactive power source (that is actually dispatched) is any one of the 3 local 5.4 Mvar capacitors . Given the voltage dependancy of the these sources (shunt caps), the loss of a nominally-rated 5.4 Mvar bank may result in a further reduction in local reasonable to slightly increase the required margin at Stratton to 6 Mvar (5.4 Mvar x 1.1). The largest reactive power sources electrically close to West Dummerston are the two local 2.7 Mvar shunt caps, therefore, the required reactive power margin reactive power source close to S Shaftsbury is the 5.4 mvar cap at Manchester, therefore, the required reactive power margin at South Shaftsbury is 6 Mvar (5.4 Mvar x 1.1). Note that the larger 10.8 Mvar caps at Woodford Road and Vernon Road are igno As with reactive power margin, "adequate" real power margin is a judgement call - a typical standard is to require adequate voltage (greater than or equal to 0.90 P.U.) up to 105% or 110% of peak loading (that is, a 5% or a 10% margin). The standard c We know the real power margin is < 10% of peak because the 80% case is N.C., but its exact value is unclear due to the large difference in load dispersal between the 70% and 80% load cases. By simply scaling the 70% case's load upward, the margin appe

10

11 12

Appendix E A New Probabilistic Reliability Guideline Based on the Equal Slope Criterion

59

A foundation for, and the establishment of a probabilistic reliability guideline for the Southern Loop
L. R. Kirby April 2005

Entities that build transmission facilities comprising a part of the New England bulk transmission network have an inherent obligation to meet the minimum reliability standards recognized by the various owners and operators of that system. Failure to meet such standards may not only jeopardize the entitys own transmission system, but those of the interconnected entities as well. The tightly woven connectivity of higher-voltage bulk networks is something of a double-edged sword. On the one hand, if unanimously operated to prudent standards and with ample margins, they are robust throughout. On the other hand, if any part is operated to lesser standards or with poor margins, the resulting risk is imposed not just locally, but over a broad area. NPCCs recognized standards for reliability on bulk power transmission networks in the Northeastern US are deterministic in nature, and are generally known as N-1 or N-2. These terms refer to the ability of a system to continue operating acceptably with one less (or two less) than the normal number of system components (i.e. lines, transformers, and generators). This requirement extends all the way up to the peak load of the system or subsystem in question. However, the NPCC standards also allow for a level of reliability lower than the N-1 or N-2 standard on lower-voltage systems, in cases where the failure of that system does not jeopardize the bulk system13. Typically, the failure of lower-voltage local networks (or radials) served by a higher-voltage bulk network, does not jeopardize the bulk network itself, but only those load centers served by the lower voltage system(s). Therefore, entities operating or building such systems are not necessarily required to adhere to the N-1 or N-2 standards, as is the case for the bulk transmission system. The appropriate level of reliability on such underlying systems is instead, determined by relative economic costs and benefits. Logically then, the appropriate level of reliability for such systems is not a fixed parameter (as with deterministic standards) but is a function of cost. This leads naturally to a discussion of a typical cost-benefit characteristic (see Figure 1). The law of diminishing returns dictates that the characteristic typically rises rapidly at first, then gradually flattens out. Solution options near the extreme left have relatively low costs and relatively high benefit-to-cost ratios (i.e. the slope of the characteristic) but minimal overall benefits. Solution options near the extreme right have higher overall benefits but very high cost and poor benefit-to-cost ratios. Deterministic standards are often so rigorous as to result in solution options that fall within this undesirable high-cost region. More desirable solutions, with a reasonable cost/benefit tradeoff tend to exist near the knee of the curve as indicated in Figure 2. CVPS strives to achieve a reasonable balance between total benefit levels and benefit-to-cost ratios. In other words, the Company seeks solutions that tend to fall near the middle of the cost-benefit tradeoff curve. At CVPS, this cost/benefit philosophy is often referred to as a most for less approach, that is, most of the benefit of a deterministic standard but at substantially less cost. In terms of utility reliability standards, the approach which best achieves this balance is the probabilistic approach, which is flexible and may be cost-based, as opposed to the deterministic
Section 2.1 Design Criteria of the NPCCs Basic Criteria for Design and Operation of Interconnected Power Systems document states It is recognized that the Basic Criteria are not necessarily applicable to those elements that are not a part of the bulk power system or in the portions of a member system where instability or overloads will not jeopardize the reliability of the bulk power system.
13

60

approach, which is rigid and technically-based. The probabilistic approach to reliability is inherently more cost-effective than the deterministic approach because it concentrates capital resources on the greatest threats until the cost becomes prohibitive, whereas the deterministic approach basically assumes that capital resources are limitless and will be expended until the N-1 or N-2 standard is satisfied. Furthermore, although the deterministic approach does recognize cost in choosing an option from among those that meet the reliability standard, it does not recognize cost as a determining factor in the development of the standard itself14. Thus far, we have established a probabilistic reliability philosophy (i.e. most for less) but have yet to develop a quantitative standard or guideline based on that philosophy. The development of such a guideline begins with an examination of a typical load duration curve. Figure 3 is a generic load duration curve for an area or subsystem that may be defined by a closed interface. A load duration curve is simply a graphical representation of the range of loads that are experienced by an area or subsystem over a period of time, typically one year. The parameters of load and duration may be expressed as percentages (up to 100%), or as absolute values (e.g. megawatts on the vertical axis and hours on the horizontal axis). Note that the slope of the curve is related to the duration spent at or near a given load value - the flatter the slope, the greater the amount of time spent at or near that load. Load duration curves are widely used by planning engineers to understand systems load behavior, and to determine the probabilities of being within specific load ranges. Before embarking on an explanation of a probabilistic standard or guideline, it may be useful to first explain a deterministic standard using the load duration curve. Figure 4 provides the same load duration curve as in Figure 3 but with the minimum load-serving capability required to meet an N-1 deterministic standard, indicated by the horizontal green line. Note that it intersects the curve at 100% of peak load, in other words, it requires that the system operate acceptably for any first contingency up to peak load. Now we may expand our discussion by considering once again the diminishing return on investment that is inherent in the cost benefit characteristic. The reasons for the law of diminishing returns are varied, but in the case of transmission planning, one of its leading causes may be observed in a typical load duration curve. When approaching its left-hand extreme, the load duration curve accelerates dramatically upward, culminating in a near-vertical termination. The electrical infrastructure required to satisfy the N-1 criterion all the way to peak load may be quite expensive, but the duration of time spent near this peak (and therefore the probable exposure to a critical contingency) increases much more slowly than the increase in load itself. Therefore, the reliability benefit of meeting the N-1 criterion on this last upward stretch of the load duration curve quickly diminishes, contributing greatly to the characteristic non-linearity of the cost-benefit curve. An effective probabilistic reliability guideline must recognize this important relationship. Therefore, we should avoid requirements for load-serving capability that extend all the way to peak load, but we should also avoid overall lax requirements that result in few reliability benefits. Accordingly, it makes sense to require first-contingency coverage for the load levels in the shoulder region of the load duration curve (i.e. the broad middle region between the upward and
3

Note that a deterministic reliability standard (specifically known as N-1) was used in the assessment of each Southern Loop DR option, so as to provide equal reliability performance, thereby permitting an equitable comparison of differing capital costs, losses, aesthetics, and other attributes. The N-1 standard serves as a convenient yardstick for such comparisons, but its use in this limited way should not be viewed as a general endorsement of its use for the Southern Loop. To the contrary, the probabilistic method, mentioned above, is endorsed as the appropriate reliability approach for non-bulk transmission network planning.

61

downward curvatures) where the % increase in load almost always occurs more slowly than the % increase in duration. In this region, reasonable gains in reliability (through reductions in probable contingency exposure) may be achieved by capital investment15. But as the load duration curve begins to shift toward vertical near the extreme left-hand side, the incentive for continued investment quickly diminishes. At what point does this relationship break even, that is, at what point are we effectively indifferent to the expenditure of the next increment of capital to achieve the next increment of load-serving capability? The answer will differ depending on the specifics of the system under consideration, the criticality of the load it serves, and other factors. However, a logical guideline would be the point at which the rate of change of load (on a percent basis) begins to exceed the rate of change of duration (on a percent basis). Any significant movement to the left of this point on the load duration curve will lead to a rapidly-diminishing return on investment. Yet, such a guideline will still capture all of the shoulder region of the curve where the percent change in load is usually lower than the percent change in duration, and where reliability investment tends to be gainful. Figure 5 presents this concept graphically, using the same load duration curve as in Figures 3 and 4. A diagonal line connecting the peak load at zero duration to the minimum load at full duration has a slope equal to one (on a percent basis), and thus all along this line the rate of change of load percentage is equal to the rate of change of duration percentage. This slope may then be shifted until it is tangent to the curved portion of the left-hand side of the load duration curve. This point of tangency defines the required load-serving capability of the system during the worst contingency, as well as the acceptable level of exposure duration (i.e. the percent of time that the system would not be able to withstand the contingency without load-shedding). In our generic example, these values are 80.5% of peak load and 7.5% duration, respectively. The proposed probabilistic reliability guideline may therefore be summarized as follows: The (minimum) reliability guideline for non-bulk network transmission is such that the system or subsystem in question must be able to survive the worst single contingency with minimally acceptable performance, for the load level at which the systems marginal load percentage is equal to its marginal duration percentage, within its load duration characteristic. Generally there are two such points on any load duration characteristic. The right-hand point is trivial; the left-hand point is that to which this guideline applies. This probability-based reliability guideline may be referred to as the equal slope criterion for short. It must be recognized that capital investments in utility infrastructure are not always scalable, but may instead be lumpy, in the parlance of utility planners. This means that achieving the equal slope criterion with perfect precision may not be practical or even possible in some cases. This may leave the utility planner having to decide between overshooting or undershooting the prescribed load-serving capability, with the resulting upward or downward deviation yielding a less than desirable cost-benefit tradeoff either way. Nonetheless, a rational guideline that may be difficult to achieve is better than no guideline at all, and at least provides a frame of reference for making the decision. It is suggested however, that the guideline be treated as a desired minimum,
15

If contingency coverage of even the shoulder hours is deemed to be too expensive, then the next lower reliability standard for networked transmission systems is effectively N-0, also known as all lines in service. Although used in some parts of the world, US utilities and regulatory authorities are generally reluctant to accept this standard, viewing it as too lax for transmission networks.

62

as noted within the parentheses above. This resolves the difficulty of deciding whether to overshoot or undershoot the criterion, and errs on the side of overshooting it in cases where it cannot be achieved with exactitude. Until this point, the discussion has focused on the development of a probabilistic reliability guideline, with generic examples of its application. But what of its application to the Southern Loop specifically? Figure 6 presents the present-day load duration curve for a closed interface around the Southern Loop, including Bennington and Brattleboro. Note that the software developed for this assessment cannot draw a perfectly smooth curve, as revealed by a close examination of Figure 6. However, an ancillary assessment was done with a load duration curve hand-drawn by a French curve to enhance its smoothness. The application of the equal slope criterion to this curve results in a (minimum) post-contingency load-serving requirement of 73.5% of the present-day peak load, or 85 Mw. This corresponds to a desired exposure duration of no more than 8.8%, or 771 hours per year.

63

64

65

66

67

68

69

Appendix F When and Where Load Growth Modeling in Southern Vermont

70

INTRODUCTION Southern Vermont load is a mixture of atypical load patterns, diverse but clustered customer types, and varied commercial development, making its growth assessment quite challenging. Accordingly, there are two chief parameters of load growth that must be addressed: when the load grows and where the load grows. And a very important thing to understand is that these two parameters must be recognized simultaneously within a single, unified model. In other words, we cannot merely be concerned with when or where the load grows as independent considerations; we must be concerned with when and where the load grows, because the interdependency of these two factors is critical to determining the performance of each option as growth occurs. WHEN THE LOAD GROWS The when parameter, is not with respect to seasonal, weekly, or daily cycles, but rather, with respect to annual load duration (i.e. When the load is at the higher end of the LDC, or at the intermediate portions, or at the lower end of the LDC, how fast it is growing?). Quite often, the load in geographically proximate areas may be assumed to grow in relatively uniform fashion, at all points on the LDC. This is generally true in areas where there is no peak-shifting DSM program, so that the load factor tends to be stable over time. In such cases, a single percentage growth value may sometimes be applied to all load levels for as many years as one wishes to forecast, as a reasonable approximation. With peak-shifting DSM however (which has been present for some years on the Southern Loop), the shape of the LDC is deliberately manipulated via rate structure, special contracts, load-control technologies, and other means. This fact, coupled with the atypical mix of loads and load behaviors, means that Southern Vermont load growth varies as a function of position on the LDC (when), and this fact must be recognized in the simulation and analysis of load growth. Consequently, an automated LDC processor was developed to meet this need. This is simply a spreadsheet-based graphing program. It permits any LDC to be approximated by 10 connected data points that may be grown independently (based on historic trends or whatever logic one wishes to apply), resulting in a more accurate forecasted LDC than one that is simply grown uniformly. Furthermore, it calculates exposure duration (if critical load level is known) and LDC load factor. Any or all of these 10 discrete load values may then be individually simulated in loadflow. See Figure 1. WHERE THE LOAD GROWS Clearly, certain customers growth will contribute more to system unreliability than others because of their physical location. The presence of differing growth rates at these differing locations only compounds the problem. How should we recognize this distinction in our load models? Furthermore, should we be looking at just the B12 + B2 loads and load growth, or at the wider area that includes Brattleboro and Bennington? A solution to this dilemma has been developed, that borrows conceptually from finite element analysis. Finite element analysis is a technique used by mechanical engineers to simulate structural bending in generally rigid but slightly flexible structures, such as car frames or jet fuselages. The basic premise of the finite element technique is that a single large structure cannot be depended on to behave as perfectly rigid, but it may instead be treated as a finite collection of perfectly rigid sub-structures that are connected, and that allow for some bending motion between
71

them but no bending motion within them. The engineer must select the number of elements such that it breaks the structure into enough pieces to be a good approximation of reality, but not so many as to impose undue complexity and computational burden. One clever trick used by structural engineers is to choose the elements and the boundaries between them in a way that recognizes more rigid areas (such as steel beams) treating them as single elements, and less rigid areas (such as the welds between the beams) treating them as boundaries between the chosen elements. This helps to minimize the number of elements needed for a good approximation, as opposed to simply breaking the structure up into many elements of equal size and random boundary location. In a parallel sense, we know that we cannot rely on the load growth of the whole Southern Vermont system to behave uniformly, but we can certainly identify a finite number of load elements whose growth will be reasonably uniform unto themselves. Where are these clusters of uniformly-growing load, and where are the natural break points between them (i.e. where are the beams and where are the welds)? After some consideration, it was decided that the system may be broken in 5 load elements that should have fairly uniform load behavior and growth rates within each, but not necessarily between them. These 5 elements are, in no particular order: 1. The non B-2 Woodford load (South Bennington, Pownal, North Bennington, etc) 2. The Brattleboro area load (B-67) 3. The Stratton load (both circuits, because they are both sensitive to resort activity and because they are electrically proximate) 4. The Bromley load (similar to Stratton, but different enough, big enough, and electrically far enough from Stratton to warrant its own load element) 5. The non-Stratton/non-Bromley Southern Loop load TYING IT ALL TOGETHER Just as the mechanical elements must be combined to represent the total structure in a finite element analysis, so must these 5 load elements be combined to represent the total load in the southern Vermont load growth analysis. In other words, the individual load elements must always add up to the total load as represented by the load duration curve. This requirement is what makes the when and where parameters work simultaneously within a single unified model as alluded to in the introduction of this appendix. More specifically, each one of the 5 load elements will be individually represented at each one of the 10 different positions on the ungrown load duration curve, for a total of 50 degrees of freedom in the complete load model matrix. Each of the 5 load elements, at each of the 10 duration positions will then be grown using historic records of its specific growth (or based on some other rationale), but with the caveat that the total of the 5 must be equal to the total load value at the specific duration value of the LDC for the whole system. This limiting condition may require some growth rate adjustments due to poor coincidence of specific data or metering error, in which the individual load element growth rates must be scaled up or down to meet the total load, while maintaining their relative rate differences. The resulting 50 data points may be then represented with 10 loadflow models (each one containing 5 distinct load elements in Southern Vermont). Based on the discussion above, this would require 10 load flow models for each solution option, in order to completely simulate the operating performance of each, following a given period of load growth (say 5 or 10 years). However, the resulting large group of required loadflows may be substantially reduced if we recognize that we are not really interested in the specific performance of every option at every load level, but rather, we are only interested in the one limiting load value
72

where each option is just barely able to meet deterministic performance criteria16 (voltage, for this region). We can do so by using iterative loadflow runs to quickly zero in on this critical load. This limiting load value, combined with our capability to represent grown LDCs as multi-point functions with multiple growth rates, allows us to efficiently determine the exposure duration of each option, for any given contingency, after a specified period of load growth occurs. Although we already know how effective each option is at improving the system reliability today, this analysis tells us how effective each option is at maintaining that improvement over time.
Southern Loop Total Area Load Duration Curve, 2000 / 01 (from the original calibrated source cases)
120.0 hour 1 is 100% load, represented by ave. load in hour 53 is 90% load, represented by ave. load in hours 100.0 hour 561 is 80% load, represented by ave. load in hours hour 2051 is 70% load, represented by ave load in hours 80.0 hour 6026 is 60% load, represented by ave. load in hours Load (% of peak) hour 9646 is 50% load, represented by ave load in hours 784160.0

40.0 hour 13436 is 40% load, represented by ave. load in hours hour 16125 is 30% load, represented by ave. load in hours 20.0 hour 17407 is 20% load, represented by ave. load in hours

0.0 100.0

Figure 1

Duration (annual %)

Peak Load of Base Curve (MW) = Exposure Duration of Base Curve (%) = Load Factor of Base Curve (%) =

100.0 100.0 52.3

Peak Load of Grown Curve (MW) = Exposure Duration of Grown Curve (%) = Load Factor of Grown Curve (%) =

100.0 100.0 52.3

Disclaimer: The values in this graph reflect the actual change in duration-based load growth from 1997/98 to 2002/03

This single load value, applied to the LDC, determines the exposure duration for the contingency under study, which in turn makes possible a comparison of the failure probabilities of various options for that contingency. 73

16

Appendix G Overcompensation of Network Systems and its Relationship to the Southern Loop

74

Figure 1 shows a generic PV (Power versus Voltage) characteristic for an uncompensated network (i.e. transmission or subtransmission) system, meaning that it is not reliant on shunt capacitors to support its voltage. As the systems power burden increases, its voltage decreases in a non-linear, but fairly gradual fashion. Given a sufficient increase in power loading or transfer, the voltage will drop to a minimally acceptable operating value (i.e. 90% of nominal if coincident with a contingency, or 95% if not). Beyond this point, operators would normally take action to restore adequate voltage, up to and including deliberate load-shedding. If no action was taken and power transfer continued to increase, the voltage would drop further until the rightmost tip of the PV characteristic was reached. At this point the voltage would decay rapidly (i.e. collapse), culminating in a blackout of the area proximate to the overburdened system. The amount of load or power transfer between the point at which operators would be expected to act, and the point of voltage collapse, constitutes a margin of safety as indicated in Figure 1.

Figure 2 shows the same uncompensated characteristic depicted as a dotted curve, with a heavily compensated characteristic as the solid curve above it. Note the saw tooth shape of the heavily compensated characteristic. Each vertical discontinuity shows the effect of adding another shunt capacitor. At each discontinuity, enough shunt capacitance is added to return the systems voltage to unity (100% of nominal). Due to the physics of voltage drop, the trajectory of the compensated characteristic after each capacitor addition, mimics that of the uncompensated trajectory below it, but with a higher starting point. Because these two parallel trajectories become increasingly steep, the distance between each successive discontinuity diminishes. Simply put, the additional capacitors permit higher power transfers with acceptable voltage, but provide diminishing returns with each successive addition. In the example in Figure 2, a total of four capacitors are added sequentially, corresponding to the four discontinuities in the heavily compensated characteristic. By the time that the fourth capacitor has been added, the slope of the trajectory is rapidly becoming steeper, and is approaching the limit where it attains a vertical slope. Consequently, at the point
75

where the voltage reaches its minimally acceptable value, the safety margin has shrunk to a small fraction of its former size. An operator waiting to act until the voltage reached this minimally acceptable value would be perilously close to system collapse, and would likely be unaware of the danger.

Note that shunt compensation can be pushed even farther, to a point where the safety margin is reduced to zero. This happens when the vertical tip of the PV characteristic (where voltage collapse initiates) is positioned at or above the point where the voltage reaches a minimally acceptable value. In such cases, the system may collapse even while the voltage is still within an apparently healthy range. Systems that display such behavior are described as being overcompensated, and are vulnerable to a sudden and unexpected voltage collapse. Loadflow studies of the Southern Loop indicate that it is presently at or near the point of overcompensation. Entry 38 in Table 6 in the main report shows that if the Woodford-South Shaftsbury 46 kV line segment is kept out of service for a significant period of time due to a fault, the capacitors on the Southern Loop may be re-dispatched so as to provide 88% of nominal voltage (or better, depending on the location) at 68% of peak load. This is very nearly in compliance with post-contingency voltage criteria. Yet, the real power margin (i.e. the safety margin) for these conditions was found to be zero (see entry 38, column R) meaning that even a tiny increment of added load would cause mathematical non-convergence of the loadflow, which, in this particular case may be interpreted as a portent of voltage collapse. A slight decrease in load would undoubtedly permit the voltage to rise from 88% to 90% (the minimally acceptable level) thereby providing a razor-thin load margin between this purportedly safe voltage limit and outright voltage collapse. This data is compelling evidence of overcompensation. Any further reliance on shunt capacitors is ill-advised and in fact, the present situation already warrants remediation, with the goal of reducing the Southern Loops reliance on shunt compensation.
76

Appendix H The Effect of Low Load Growth but High Load Factor Growth on the Southern Loop

77

Figure 1 depicts a generic load duration curve (LDC), which is simply a graphical representation of a given systems load levels as a function of the duration of time that each load level persists. The shape of this example LDC is quite typical, with the gently sloping shoulder hours near the middle (so named because of their resemblance to a human shoulder) and a pronounced upward sweep at the left side and downward sweep at the right side. The total width from left to right represents one year (8760 hrs).

Figure 2 depicts the same LDC but with a second one superimposed over the first to show the effect (generically) of an increasing load factor (perhaps due to DSM efforts) but with a static peak load.

Let us assume that the power system in question cannot maintain adequate postcontingency voltage beyond 70% of its peak load. One might well imagine that in the
78

absence of peak load growth, as is the case here, the system would be able to avoid increasing reliability problems, but in fact that is not always true. The duration of time that the Figure 1 LDC is exposed to loads beyond its postcontingency load-serving capability, is proportional to the distance between the vertical axis and the vertical marker corresponding to 70% load. This duration is referred to as the exposure duration, meaning the annual duration of time that the system is exposed to the possibility of a contingency that it could not withstand without violating one or more operating criteria (i.e. voltage limits, equipment loading limits, etc). Such violations would likely force the operators to take rapid and drastic corrective measures, possibly including deliberate load-shedding. A belated response could lead to widespread voltage collapse. Note that although the peak load is the same in the second curve, the exposure duration has effectively doubled because of the rise in load factor. Therefore the exposure to reliability problems has doubled as well. This relationship explains, in part, why the reliability problems on the Southern Loop have escalated substantially despite the modest load growth in recent years. The Southern Loop is only capable of handling contingencies up to a certain intermediate load level, and its growing load factor makes the consequent exposure worse as time goes on. The strong tendency of the Southern Loop toward an increasing load factor means that the shape of its load duration curve is changing over time, particularly that of its intermediate loads. This fact, coupled with this systems inability to handle certain contingencies beyond intermediate load levels, means that any attempt to forecast and simulate its load growth must recognize the differing growth rates along the load duration curve, not just the peak growth rate. This concern, in part, lead to the use of the When and Where load forecasting method, described in Appendix F When and Where load growth modeling in Southern Vermont.

79

Appendix I Derivation of Outage Probabilities For the Southern Loop

80

The reader is encouraged to begin by reading Appendix C Basic Probability Theory in order to better understand the mathematics and terminology that will be used in this appendix. However it is not essential. We begin our derivation by noting that we are interested in the frequency of outages that may cause loss of load, but not their duration. This is because we are quantifying the probability of an event occurring within a specified period, not the probability of being in the event state, which is a different parameter altogether. Appendix C explains this distinction. Given our interest only in the frequency of events (i.e. contingencies) that cause loss of load, we can make certain simplifying assumptions. It is well-established that transformer outages, breaker failures, and bus faults all tend to be relatively low-frequency (though relatively high-duration) contingencies. Accordingly, they may be ignored because they contribute little to the total frequency of critical events or contingencies. Line outages on the other hand, have relatively high frequencies and must therefore be included in our calculations. Table 1 presents the relevant line outage statistics necessary to quantify the desired 50/90 probabilities.
Table 1 - Southern Loop Sustained Line Outage Probabilities
line segment Woodford - S. Shaftsbury 46 kV S. Shaftsbury - Arlington 46 kV Arlington - Wallace 46 kV Wallace - Manchester 46 kV Manchester - Bromley 46 kV Bromley - Stratton 46 kV Stratton - Rawsonville 46 kv Rawsonville - Londonderry 46 kV Londonderry - E. Jamaica 46 kV E Jamaica - W. Dummerston 46 kV W. Dummerston - N. Brattleboro 46 kV N. Brattleboro - Vernon Road 46 kV Vernon - VY - Keene 115 kV (N-186)
1

exposed line length (miles)1 5.17 7.73 10.24 0.15 3.68 5.52 4.60 0.01 7.01 12.82 7.49 2.10 25.00

average fault frequency per year 0.222 0.332 0.440 0.006 0.158 0.237 0.198 0.000 0.301 0.551 0.322 0.090 0.150

simulated threshold load for a severe steady state outcome 75% 75% 75% 75% 75% 95% 80% 85% 75% 75% 75% 70% 65%

P.U. duration at or above threshold load 0.083 0.083 0.083 0.083 0.083 0.001 0.034 0.017 0.083 0.083 0.083 0.137 0.239 total

predicted fault frequency at or above threshold load, per year 0.018 0.028 0.037 0.001 0.013 0.000 0.007 0.000 0.025 0.046 0.027 0.012 0.036 0.249

excluding radial line taps

It is assumed that each segment of the line is independent of the others in terms of the probabilities of its fault or failure. These individual line segments all have switches with which operators may remotely de-energize and isolate them, but not all segments are individually protected by breakers. Therefore, line faults may temporarily take out more than one of the segments listed in the table, but the bad segment would quickly be isolated and the others restored to service through supervisory switching actions. As a simplifying assumption, these temporary extensions of the outage are deemed to be of little consequence to overall reliability and are therefore ignored. This approach also dovetails with our SERVE standards which ignore outages of < 5 minutes . Accordingly, the title of the table includes the word sustained to describe the line outage statistics.
81

The predicted fault frequency of any given line segment is simply a product of its length and the historic average fault frequency per unit length for line faults on the CVPS system (0.0429 faults per mile per year for 46 kV lines). These quantities are presented in the 2nd and 3rd columns of the table. Of course faults occurring at less than a certain critical load are of no consequence because they do not result in post-contingency violations (typically undervoltage) of sufficient severity to warrant load-shedding or to trigger voltage collapse. So for each line segment there is an associated threshold load that will cause a severe outcome presented in the 4th column. The 5th column is based on the 4th and is the annual duration (in perunit) that the system is at or above this threshold load. So the predicted fault frequency (2nd column x 3rd column) is then multiplied by the per-unit duration at or above threshold load (5th column) to calculate the final parameter called predicted fault frequency at or above the threshold load per year. This is a line-lengthweighted and load-duration-weighted statistic that may then be added together for the entire Southern Loop system. Note that the ever-critical N-186 115 kV line has also been included in the total, with its own statistics appropriate to 115 kV lines. The final combined number (0.249) is the predicted frequency per year that some line outage will occur in the Southern Loop area that is bad enough to cause loss of load. This value, while useful, is not very intuitive, so we will now convert it into 50/90 probabilities to make it more meaningful. The cumulative probability function relevant to our discussion is:
F(t) = e-t Where = average frequency of the event (a constant) and t = time (a variable)

Specifically, we need to find the integral of this function, which is:

Per unit probability = 1 - 1 et

Substituting in the 0.249 frequency value into the equation, along with the desired load loss probability of 50% yields:
0.50 = 1 - 1 e(0.249) t

Algebraic manipulation may then be used to solve for t as follows :

0.50 -1 = - 1 e(0.249) t

VELCO provided the 115 kV outage statistics. 82

- 0.50 = - 1 e(0.249) t 0.50 = 1 e(0.249) t (0.5)e(0.249) t = 1 e(0.249) t = 2 (0.249 t) = ln(2) t = ln(2) 0.249 t = 2.78 t = 3 years, rounded to the nearest year

Similar calculations show that the 90% probability is achieved within 10 years. Some qualifications and reminders are in order at this point. This technique does not attempt to quantify the amount of load lost, nor the duration of this loss, but only the probability of losing at least some load within a given span of time. Furthermore, the probabilities are presented with the caveat at least, meaning that the actual probabilities of load loss are possibly higher than is presented here. This is because, as noted earlier, certain failure mechanisms have been ignored, such as transformer outages, breaker failures, and bus faults, although these are typically of much lower frequency than line failures. Also, no load growth is assumed for the 3 year and 10 year durations that were derived for the 50% and 90% probability values. All of these factors would increase (though it is believed only modestly) the probabilities of load loss.

83

Appendix J Strategic Solution Option Descriptions

84

Strategic Solution Option #1 (A,B,S) - A new synchronous condenser installation at a new Stratton 46 kV substation. A new Vermont Yankee to West Dummerston 115 kV transmission line on the existing VELCO 345 kV line right-of-way between Vermont Yankee and Coolidge. A new West Dummerston to Coolidge 115 kV transmission line on the same right-of-way. A new 115/46 kV substation at West Dummerston. Strategic Solution Option #2 (A,B,C,S) - A new synchronous condenser installation at a new Stratton 46 kV substation. A new Vermont Yankee to West Dummerston 115 kV transmission line on the existing VELCO 345 kV line right-of-way between Vermont Yankee and Coolidge. A new West Dummerston to Coolidge 115 kV transmission line on the same right-of-way. A new 115/46 kV substation at West Dummerston. A new West Dummerston to Stratton 115 kV line on the existing West Dummerston to Stratton 46 kV right-of-way. A new 115/46 kV substation at Stratton. Strategic Solution Option #3 (A,B,C,D) - A new Vermont Yankee to West Dummerston 115 kV transmission line on the existing VELCO 345 kV line right-of-way between Vermont Yankee and Coolidge. A new West Dummerston to Coolidge 115 kV transmission line on the same right-of-way. A new 115/46 kV substation at West Dummerston. A new West Dummerston to Stratton 115 kV line on the existing West Dummerston to Stratton 46 kV right-of-way. A new 115/46 kV substation at Stratton. A new Stratton to Woodford Road (Bennington) 115 kV transmission line on the existing Stratton to Woodford Road 46 kV right-of-way. Strategic Solution Option #4 (A,B,C,D,S) - A new synchronous condenser installation at a new Stratton 46 kV substation. A new Vermont Yankee to West Dummerston 115 kV transmission line on the existing VELCO 345 kV line right-of-way between Vermont Yankee and Coolidge. A new West Dummerston to Coolidge 115 kV transmission line on the same right-of-way. A new 115/46 kV substation at West Dummerston. A new West Dummerston to Stratton 115 kV line on the existing West Dummerston to Stratton 46 kV right-of-way. A new 115/46 kV substation at Stratton. A new Stratton to Woodford Road (Bennington) 115 kV transmission line on the existing Stratton to Woodford Road 46 kV right-of-way. Strategic Solution Option #5 (A,C,D,S) - A new synchronous condenser installation at a new Stratton 46 kV substation. A new Vermont Yankee to West Dummerston 115 kV transmission line on the existing VELCO 345 kV line right-of-way between Vermont Yankee and Coolidge. A new 115/46 kV substation at West Dummerston. A new West Dummerston to Stratton 115 kV line on the existing West Dummerston to Stratton 46 kV right-of-way. A new 115/46 kV substation at Stratton. A new Stratton to Woodford Road (Bennington) 115 kV transmission line on the existing Stratton to Woodford Road 46 kV right-ofway. Strategic Solution Option #6 (A,T4,S) - A new synchronous condenser installation at a new Stratton 46 kV substation. A new Vermont Yankee to West Dummerston 115 kV transmission line on the existing VELCO 345 kV line right-of-way between Vermont Yankee and Coolidge. A new 115/46 kV substation at West Dummerston. A new backup 345/115 kV transformer located at the Vermont Yankee 115 kV substation. Strategic Solution Option #7, initial phase (A,B,S, 25 Mw of DG) - A new synchronous condenser installation at a new Stratton 46 kV substation. A new Vermont Yankee to West Dummerston 115 kV transmission line on the existing VELCO 345 kV line right-of-way between Vermont Yankee and Coolidge. A new West Dummerston to Coolidge 115 kV transmission line on the same right-of-way. A new 115/46 kV substation at West Dummerston. 20 Mw of distributed generation at Stratton and 5 Mw of distributed generation at Bromley. Strategic Solution Option #7, final phase (A,B,S, 19 Mw of DSM, 21 Mw of DG) - A new synchronous condenser installation at a new Stratton 46 kV substation. A new Vermont Yankee to West Dummerston 115 kV transmission line on the existing VELCO 345 kV line right-of-way between Vermont Yankee and Coolidge. A new West Dummerston to Coolidge 115 kV transmission line on the same right-of-way. A

85

new 115/46 kV substation at West Dummerston. 19 Mw of dispersed demand side management. 21 Mw of distributed generation at Stratton. Strategic Solution Option #8, initial phase (A,T4,S, 25 Mw of DG) - A new synchronous condenser installation at a new Stratton 46 kV substation. A new Vermont Yankee to West Dummerston 115 kV transmission line on the existing VELCO 345 kV line right-of-way between Vermont Yankee and Coolidge. A new 115/46 kV substation at West Dummerston. A new backup 345/115 kV transformer located in the Vermont Yankee 115 kV substation. 20 Mw of distributed generation at Stratton and 5 Mw of distributed generation at Bromley. Strategic Solution Option #8, final phase (A,T4,S, 19 Mw of DSM, 21 Mw of DG) - A new synchronous condenser installation at a new Stratton 46 kV substation. A new Vermont Yankee to West Dummerston 115 kV transmission line on the existing VELCO 345 kV line right-of-way between Vermont Yankee and Coolidge. A new 115/46 kV substation at West Dummerston. A new backup 345/115 kV transformer located in the Vermont Yankee 115 kV substation. 19 Mw of dispersed demand side management. 21 Mw of distributed generation at Stratton. Strategic Solution Option #9, initial phase (S, 42 Mw of DG) - A new synchronous condenser installation at a new Stratton 46 kV substation. 21 Mw of distributed generation at Stratton, 6 Mw of distributed generation at Bromley and 15 Mw of distributed generation at Brattleboro. Strategic Solution Option #9, final phase (S, 26 Mw of DSM, 38 Mw of DG) - A new synchronous condenser installation at a new Stratton 46 kV substation. 26 Mw of dispersed demand side management. 31 Mw of distributed generation at Stratton and 7 Mw of distributed generation at Bromley. Strategic Solution Option #10, initial phase (80 mw of DG) - 21 Mw of distributed generation at Stratton, 6 Mw of distributed generation at Bromley, 16 Mw of distributed generation at Brattleboro, 17 Mw of distributed generation on the 46 kV loop west of Stratton, 20 Mw of distributed generation on the 46 kV loop east of Stratton. Strategic Solution Option #10, final phase (105 mw of DG) - 31 Mw of distributed generation at Stratton, 9 Mw of distributed generation at Bromley, 18 Mw of distributed generation at Brattleboro, 21 Mw of distributed generation on the 46 kV loop west of Stratton, 26 Mw of distributed generation on the 46 kV loop east of Stratton.

7.2 Mw @ Bennington, 5.4 Mw @ west loop, 0.4 Mw @ Bromley, 1.6 Mw @ Stratton, 4.2 Mw @ east loop, ZERO Mw @ Brattleboro (because it is not needed with a strong West Dummerston source) 7.2 Mw @ Bennington, 5.4 Mw @ west loop, 0.4 Mw @ Bromley, 1.6 Mw @ Stratton, 4.2 Mw @ east loop, 6.9 Mw @ Brattleboro

86