Mobile Edition
Print
Get the Mag
Weekly UpdatesINTRODUCTION
Deregulation has significantly changed the structure of power markets, including their operation, management, and planning processes, not only in the United States but also in other countries. Today's transmission networks do not sufficiently support the competition of generators, causing congestion of transmission lines. Despite the need for more transmission, the lack of transmission investment in North America has been documented by several researchers (Hirst and Kirby, 2001: Hyman, 1999: Shahidehpour, 2004). In restructured markets, users are paying increasing congestion costs. Independent system operators (ISOs) report constant growth in congestion costs. Figure 1 shows the total congestion charges by year as reported by the Pennsylvania-New Jersey-Maryland (PJM) ISO and the congestion as a percentage of the total electricity costs. The congestion charges at the PJM Interconnection grew from $65 million in 1999 to $1,603 million in 2006. During this period, the percentage of total congestion cost of the total cost of energy at PJM was 8.29% on the average (PJM, 2006). These figures underscore the value and the need for economics-based investments in transmission lines. Transmission expansion planning aimed at reducing these congestion costs is a challenge for both system operators (SO) and market participants.
[FIGURE 1 OMITTED]
Transmission investments that are not required for the enhancement of system reliability are defined as economic investments (Joskow, 2005a). From the social welfare perspective, an economic transmission investment is justified if the total cost of the congestion relieved by the investment is higher than the cost of the investment itself. However, it is difficult to compare these two amounts since the congestion cost, an operational expense, occurs at every dispatch, while the transmission investment cost, a capital expense, is allocated at the onset of the economic life of the project.
In traditional transmission expansion models the economic effect of congestion is usually neglected. Traditional models seek the minimum investment cost for a feasible peak-load dispatch without considering the generator costs explicitly (see Romero et al., 2002). This approach needs to be updated for today's deregulated markets because when the economic effects are taken into account, investment decisions should be based on a comparison of the equivalent costs of the investment and the savings that will be realized as a result of the investment. Consequently, a peak-load analysis approach fails to provide sufficient information about the cost of congestion in the system.
In addition to the reliability-based transmission planning, economics-based transmission planning is necessary to alleviate the excess cost of congestion. The congestion cost has become an important factor affecting electricity prices in the new deregulated markets (see PJM, 2006; Joskow, 2005b). Several approaches have been proposed to plan these economics-based transmission investments (see Wong et al., 1999; Buygi et al., 2004; Shrestha and Fonseka, 2004). The PJM ISO has included projects that are intended to relieve the persistent congestion in its regional transmission expansion planning procedures (Joskow, 2005a). However, a common framework for economics-based transmission expansion planning (E-TEP) has not yet been established.
In this article, our purpose is to build such a decision framework to help analyze and compare transmission expansion plans in power systems. This framework introduces a multi-period analysis of the system and considers minimization of investment and congestion costs. Using average load profiles for every period, equivalent economic values of operational and investment costs at the end of the construction period are calculated. The investment decisions are made by comparing the investment and the operational costs throughout the planning period. This framework makes it easy to identify, understand, and validate the necessary transmission investments.
Under this proposed framework, we develop three E-TEP approaches and show how the effects in the resulting expansion plans can be reviewed under the framework. First, we use mathematical programming and model E-TEP as a mixed-integer nonlinear problem. This model minimizes the sum of total investment cost and total redispatch cost, by considering the least-cost dispatch of the generators during the planning timeframe. This more sophisticated approach would be the best for large and subject to loop flow power networks. Second, a heuristic approach is used to plan market-based transmission investments. In this approach, the candidate transmission lines are determined based on the price differences between the nodes and the best investment plan chosen within a budget. This relatively simple approach would result in best solutions for simple networks. We show how this approach might cause more expensive solutions when the power network is more complicated. Third, the cheapest way to relieve congestion in the power system is found. The same candidate transmission lines are used as those determined by the market-based approach. Once these three investment plans are obtained, they are compared with the do-nothing alternative using the measures defined in this article.
The remainder of this article is organized as follows: The following section describes the proposed decision framework for transmission expansion planning. Next, the economics-based transmission expansion planning approaches are presented. The first part in this section covers the mathematical model for transmission expansion planning in deregulated markets and the second part covers the market-based approach for transmission expansion planning. In the other sections, the measures used to compare the investment plans are presented and a 179-bus power system to compare three alternative investment plans for this power grid is introduced. The article concludes with a summary of the findings.
DECISION FRAME WORK
A real power centralized economic dispatch model without losses is used to estimate the operation costs that will be incurred out to the planning horizon after the expansion has been implemented. For markets where generators can carry out self-dispatch, the model would still be suitable. When generators are sell-dispatched in an uncongested power network, the generation dispatch quantities might not be equal to the ones obtained from a centralized dispatch model. However, considering that utilities would intend to reduce their generation costs by purchasing cheaper energy from other utilities, the resulting dispatch of the market's units is expected to be close to the economic dispatch. On the other hand, when the network is congested, the independent system operator would have to resolve the congestion by redispatching the units using a centralized dispatch model. Since the objective of the investment is to reduce the total congestion cost over the operational phase of the investment, the centralized model would give a good estimate of this cost under the assumed simplifications of the model.
The shadow prices of this economic dispatch model are referred as the locational marginal prices (LMP), and the participants in the markets make payments based on these prices. If the transmission lines are congested during a dispatch, LMPs vary across the system. Under this structure, the power systems literature commonly applies two measures of congestion: redispatch cost and congestion rent. The redispatch cost refers to the system's cost due to congestion, namely the difference between the total generation cost without transmission constraints and the total generation cost with transmission constraints. In some publications, the term redispatch cost is also referred to as out-of-merit generation cost, cost of constraints, or congestion cost. The congestion rent refers to the difference between the total payment that the load requires and the total payment that the generators receive; this is also called merchandising surplus or congestion cost. To clarify, the terms redispatch cost and congestion rent, which are both costs of congestion, are specific to the definitions provided. Using the umbrella term congestion cost in lieu of either term is sometimes inaccurate, especially without clarification. To avoid confusion, we will not use the term congestion cost from here on but will refer specifically to redispatch cost or congestion rent.
[FIGURE 2 OMITTED]
The decision framework used in this article is decomposed into two phases: the construction phase and the operation phase, shown in Figure 2. Once the decision concerning the investment plan is taken, the construction period begins. During this period, all of the investment costs occur. When the planned transmission lines are ready to operate, the operation phase will start. A planning horizon is determined to take into consideration the savings of the investment. This approach allows us both to consider the least-cost dispatch options and to make an effective economic analysis during the entire timeframe out as far as the planning horizon.
In this approach, comparing the operating cost that is incurred every hour of the day/week/month/year to the investment "capital" cost calculated over the economic life of the system is critical. A well-defined coefficient is needed for the operational costs to make these two amounts comparable. Therefore, if the investment cost is calculated in present-value terms for a given planning timeframe, the operating costs should also be in present-value terms for the same timeframe. By using the proposed decision framework, this coefficient can be calculated easily and it will have a well-defined economic interpretation.
Since the dispatch of the generators is performed every hour, or 8,760 times a year, the occurrence of the operational cash flows can be assumed to be continuous and the present value of these values should be calculated with continuous compounding. Under these assumptions, the comparison coefficients can be treated as compounding coefficients, as shown in Equation (1), for each period p of the year y.
FEED