1

A New Congested Flowgate Ranking Strategy In MISO Market Efficiency Planning Study Rui Bo, Senior Member, IEEE, Liangying Hecker, Yang Gu, Member, IEEE, James Okullo, Jordan Bakke, and Ming Ni, Senior Member, IEEE Abstract — In MISO Market Efficiency Planning Study, a critical task is to pick from congested flowgates the ones that have the most potential economic benefits. A new ranking strategy, Estimated Potential Benefit (EPB), is proposed. To compare the performance of the new proposed ranking strategy with commonly used ones, ranking correctness index is introduced as a quantitative measure. EPB not only has the most comparable physical meaning to the actual potential benefit, but also has been verified to outperform other ranking strategies, especially when only the very top ranked flowgates are to be picked. Index Terms—power markets, congested flowgates, flowgate ranking, economic benefit, cost allocation.

I. INTRODUCTION

F

EDERAL Energy Regulatory Commission (FERC) issued order 1000 in 2011 to amend the transmission planning and cost allocation requirements established in Order No. 890 [1]. Public utilities and transmission planning entities are required to participate in a cooperative manner the regional and inter-regional transmission planning that addresses transmission needs efficiently and cost-effectively. As the first regional transmission organization (RTO) approved by FERC, Midwest Independent Transmission System Operator (MISO) has been proactive in developing economic transmission planning procedures while ensuring reliability standards are met. Over the years, MISO has implemented a value-based transmission planning process which aims at developing transmission expansion plans that provides least cost energy delivery to the customers and meets reliability, economic and policy needs [2]. One application of the process is the Joint Coordinated System Plan (JCSP) [3, 4], the first inter-regional collaborative planning effort among much of the Eastern Interconnection including MISO, SPP, PJM, and TVA. In MISO, this value-based planning process is employed in Market Efficiency Planning Study (MEPS) in 2012. Expanded from the former Top Congested Flowgate Study (TCFS) [5], MEPS seeks to identify and evaluate transmission project/portfolio solutions more broadly within the MISO footprint and on the seams, to enhance market efficiency and produce greater economic benefits. MEPS consists of two integral parts, Top Congested Flowgate Analysis to identify flowgate specific mitigation solutions, and Congestion Relief

Analysis to identify longer-term transmission needs and guide development of larger scale regional transmission projects/portfolios that provide synergy of benefits. The purpose of the Top Congested Flowgate Analysis in MEPS is to identify and prioritize highly congested flowgates within the MISO market footprint, and explore cross-border seams efficiency enhancement opportunities in coordination with neighboring regions. Flowgates that have been congested in the market and projected to be heavily congested in the future are of most concern. Among them, those that have the most potential economic benefits from congestion mitigation are of particular interest. The potential economic benefit can be obtained via out-year production cost model simulations. However, due to the large number of congested flowgates and limited computation and staff resource, it is impractical and ineffective to fully evaluate each of the congested flowgates. Therefore, a key component of the analysis is to employ a proper flowgate ranking strategy to determine which flowgates have the highest potential benefit from congestion mitigation. Among flowgate ranking strategies, binding hours, shadow prices and congestion cost are most commonly used and accepted. Previous analysis has introduced various measures to compare and evaluate the performance of these conventional ranking strategies [6]. Through revealing the relationship between shadow price and potential economic benefits, this paper proposes a new flowgate ranking strategy, along with enhanced ranking performance measure. The new ranking strategy has been shown to outperform conventional strategies, especially when the ranking is used to pick a limited number of flowgates. The new strategy is being employed in MISO MEPS study. The paper is organized as follows. Section II reviews the congested flowgate ranking principles and conventional ranking strategies. In Section III, a new congested flowgate ranking strategy, Estimated Potential Benefit (EPB), is proposed. In Section IV, the performance of the new ranking strategy is evaluated in comparison to those of conventional ranking strategies. A new performance measure, ranking correctness index, is proposed to quantify the performance of ranking strategies in a fairer manner. Discussion and conclusions are presented in Section V. II. CONVENTIONAL RANKING STRATEGIES

Rui Bo, Liangying Hecker, Yang Gu, James Okullo and Jordan Bakke are with Midwest Independent Transmission System Operator (MISO), St. Paul, MN 55108, USA. Contact: [email protected], +1-651-632-8447 (R. Bo). Ming Ni is with State Grid Electric Power Research Institute of China. Disclaimer: The views expressed in this paper are solely of the authors and do not necessarily represent those of MISO.

A. Congested Flowgate Ranking and General Principles The purpose of congested flowgates ranking is to use a ranking strategy to produce a ranking and use the ranking to select a subset of flowgates which are expected to have the

2 most potential economic benefits from congestion mitigation, so that further efforts can be devoted to these flowgates only. In MISO market efficiency studies, economic benefit of a transmission project is defined as the reduced adjusted production cost by putting in this project. The reduced adjusted production cost is the difference in adjusted production cost of two simulations: one for base case where the project is not in place, and one for change case where the project is in place. As the congested flowgate ranking has to be performed before the focused list of flowgates is finalized and MISO stakeholders can propose transmission project to mitigate the congestions, the details of the transmission project are yet unknown at the stage of congested flowgate ranking. Therefore, potential economic benefit of a flowgate is evaluated instead, which represents the reduction of adjusted production cost from congestion relief by relaxing the thermal limit of the flowgate. Generally, the higher the potential economic benefit is, the higher the economic benefit is expected. For a congested flowgate, its potential economic benefit can be obtained from two simulations: one for base case where the flowgate thermal limit is enforced; and one for change chase where the flowgate thermal limit is unconstrained. Even though all flowgates share the same base case, a change case simulation is needed for each congested flowgate. Note that the simulations are computational and time intensive because each simulation is performed using hourly chronological security constrained unit commitment and economic dispatch (SCUC/SCED) for 8760 hours of a year. The computation may be prohibitive when evaluating the potential economic benefits for a large number of congested flowgates. Therefore, it is generally ineffective and impractical to obtain the potential economic benefits for each and every congested flowgate. To avoid such exhaustive simulations, ranking strategies are employed to help select limited number of flowgates for further analysis. It would be ideal if a ranking strategy can rank flowgates in the same order as if it were ranked by actual potential economic benefits. It is however not very likely. Different strategies utilize different information and therefore may produce different rankings. A general principle for ranking strategies is to utilize readily available information in base case so that no exhaustive simulations for change cases are required. B. Commonly Used Ranking Strategies Ranking strategies that are used most commonly include binding hours, shadow price, and congestion cost. This information is available in base case simulation results and normally aggregated to annual total number. Total binding hours (BH) is the total number of hours a flowgate is binding at its lower or upper limit for a period of time, typically one year. It reflects the frequency of congestion. Total shadow price (SP) is the sum of hourly shadow price for a period of time. As shadow price is equivalent to the

reduction in system production cost resulting from 1MW increase of thermal rating on a flowgate, total shadow price can indicate severity of congestion in terms of blocked economic opportunity. Total congestion cost (CC) is the sum of hourly congestion cost for a period of time. Hourly congestion cost is the multiplication of hourly shadow price and hourly flow magnitude on the flowgate. By its definition, total congestion cost reflects the differences in total load payment total and generation revenue. All these three ranking strategies have been widely used in ranking congested flowgates in industrial practices. The ranks produced by using these ranking strategies are generally acceptable compared to the actual rank obtained by using potential economic benefit. Nonetheless, the ranking strategies themselves have no direct link with potential economic benefit. Therefore, the strategies are used only for ranking purpose, not for deriving the actual value of potential economic benefit. It naturally brings up a question: if we can devise a ranking strategy that by itself has much closer relationship with the potential economic benefit, would it result in a rank that is closer to the actual rank? This question motivated the development of the new ranking strategy that is proposed in section III. III. A NEW RANKING STRATEGY --- ESTIMATED POTENTIAL BENEFIT (EPB) A. Actual Potential Benefit (APB) and Estimated Potential Benefit (EPB) In Fig.1, the blue solid staircase curve illustrates the flowgate shadow price versus flowgate rating increase. Before any rating increase, this flowgate is binding and has a nonzero shadow price, which is the base case shadow price. When the rating on this flowgate increases, its shadow price may remain the same before rating increase reaches certain amount. When the rating goes beyond that amount, the shadow price may drop to a smaller value. This pattern may repeat till the shadow price goes down to zero, when rating increase is sufficiently large so that the flowgate is not binding any more. The rating increase from base case to the situation where the flowgate just becomes unbinding is maximum flow change. When the flowgate rating is further increased, flow on this flowgate will not grow and remain the same.

3 Fig. 1. Flowgate shadow price versus flowgate rating increase

These interesting characteristics of the flowgate shadow price curve are present due to two distinct features of the dispatch model. In typical MISO economic studies, power losses are reflected in the load forecast, and therefore the SCED is a lossless dispatch model. In addition, the generator cost curve will be piece-wise linearized so that commercial linear programming (LP) solver can be utilized for solving SCED. Based on these features, two lemmas and one proposition will be presented as follows. Detailed proof is omitted due to limit of paper length. Lemma 1: Flowgate shadow price does not increase when the flowgate rating is increased. When the rating of a flowgate is being increased, its shadow price will remain the same or decreased. However, it is not necessarily true for the shadow price of other flowgates, which may actually increase. Lemma 2: Actual potential benefit (APB) of relieving the congestion on a flowgate equals the area enclosed by the shadow price versus rating increase curve. It should be noted that, even though the shadow prices on other flowgates change as well, the calculation of actual potential benefit (APB) involves just the shadow price of the flowgate whose rating is being increased. It is because the shadow prices changes on the studied flowgate and other flowgates are both the reflection of the changes in binding constraints including marginal unit and congestions. Definition 1: Estimated Potential Benefit (EPB) of relieving the congestion on a flowgate is maximum flow change times base case shadow price. In Fig.1, EPB is the area enclosed by the red dashed line. Note that, if flow decreases from base case to change case, maximum flow change will be 0MW. Proposition 1: EPB is greater or equal to APB. In Fig.1, the rectangular area enclosed by the red dashed line is greater than the shaded area. That is, EPB is the upper bound or over-estimation of APB. B. Annual Total EPB Definition 1 defines the calculation of hourly EPB. For ranking purpose, the hourly EPB needs to be aggregated to annual total EPB. By definition, the annual total EPB should be defined as follows: 8760

Annual Total EPB = ∑ ( ∆f m−i * µ i )

(1)

i =1

Where ∆f m −i is the maximum flow change from base case to change case for hour i;

µ i is

the flowgate shadow price for

hour i in base case. ∆f m −i is defined as

 f new−i − f i . when f new−i >= f i ∆f m−i =  when f new−i < f i 0,

Where f new−i is the flowgate flow at hour i in the change case; f i is the flowgate rating in the base case. In planning studies, flowgate rating is not dynamically changing for each hour and typically differs only by summer/winter period and system-in-tact/contingency scenarios. In current commercial simulation software for hourly chronological SCED, maximum flow change for each hour (

∆f m −i ) is not part of the general optimization solution. Additional parametric analysis is needed to calculate ∆f m −i . Reference [6] introduced an approach that can calculate

∆f m −i using simplex-like method [7] or variable substitution method [8]. These methods, however, require development of additional functionalities in current commercial simulation software. Before it is implemented into the commercial software, we want to take an approach that utilizes only information that is currently available in the software. A brute force approach would be to relax the flowgate rating and perform a change case simulation to get the new hourly flow. Then, one may consider the difference between hourly flow in base case and change case to be the maximum flow change for the hours ( ∆f m −i ). This approach requires one simulation for the change case of each flowgate, which is to be avoided according to the general principles for ranking strategies in section II-A. Fortunately, we can take an effective alternate to the above brute-force approach. Relax all the ratings of all congested flowgates together and perform one change case simulation, and resulting hourly flow on each flowgate has been verified to be close to the one if the flowgate rating is relaxed individually. This is particularly true if the congested flowgates are electrically distant from each other. In the situation where some flowgates are electrically close, they can be divided into several groups and the flowgate ratings of each group can be relaxed simultaneously so that only a few change case simulations are needed. C. Revised Formula for Annual Total EPB If the simulation in each hour is independent on those in other hours, Equation (1) is the ideal definition for annual total EPB. However, the inter-hour dependency of simulation results may make equation (1) an inaccurate estimate, particularly an underestimate. For any hour the flowgate is binding in the base case, we expect to see positive hourly EPB for that hour when the flowgate rating is relaxed. For any hour the flowgate is not binding (i.e.,

µi

=0), it is expected that maximum flow

change ∆f m −i =0 and the hourly EPB will be zero. However, the intuitions are not always true in the actual hourly chronological simulation. In hourly chronological SCED simulation, dispatch results in one hour will be carried over to the next hour and affects the dispatch results for that hour. This hourly interdependency

4 results in counter-intuitive congestion pattern changes between base case and change case where the flowgate rating is relaxed. For instance, if a flowgate is congested in a specific hour in the base case, and when the flowgate constraint is relaxed in the change case, the flowgate flow is expected to be higher in the same hour in the change case and therefore the EPB for that hour is expected to be non-zero. However, it happens that the flowgate flow at that hour becomes smaller than the original rating, due to the carry-over dispatch from previous hours. The resulting maximum flow change is 0MW and EPB for this hour is $0. Another example is, a flowgate is not binding in base case, ie, base case flowgate flow is smaller than original rating and the shadow price is $0/MWh. When the flowgate rating is increased in the change case, the flowgate flow is expected to remain the same and EPB is expected to be $0. In hourly chronological simulation, the flowgate flow in the change case may actually go above the original rating for that hour. However, the resulting EPB is still $0 due to the flowgate shadow price in base case being $0/MWh. The above examples illustrate the fact that hourly economic benefit may not align with hourly shadow price, and it may shift or redistribute among hours because of the chronological carry-over of dispatch (and/or unit commitment). Therefore, in order to more appropriately capture the economic benefits that do not align with the shadow price due to inter-hour dependency, the annual total EPB is calculated using revised formula as follows: 8760

Annual Total EPB = ( ∑ ∆f m−i ) * µ avg

(2)

i =1

Where

µ avg is the average shadow price.

Equation (2) implies that, for every MW of flow increase from the original rating, its corresponding economic benefit is priced at the average shadow price (i.e., total shadow price divided by total binding hours).

Congestion Cost (CC)

shadow price times MW flow on the flowgate

$

Estimated Potential Benefits (EPB)

flowgate shadow price times maximum flow change

$

reflects the difference between load payment and gen revenue reflects the estimate of the actual potential benefit

B. Performance Evaluation Methods To evaluate the performance of ranking strategy, the produced rankings need to be compared with the actual rank. As none of the ranking strategy can produce a rank that is identical to the actual rank, quantitative measures need to be developed and utilized to compare the performance of different ranking strategies. A few measures including ranking correlation, ranking correctness count and ranking correctness percentage have been presented in [6]. These measures however do not take into account the order and relative significance of the flowgates that are picked using the ranking strategies. For instance, when we use two ranking strategies to pick top 10 flowgates respectively and both strategies pick 6 flowgates that are in the actual top 10 list (and 4 other flowgates not in the actual top 10 list), then the ranking correctness count is 6 and ranking correctness percentage is 60% for both strategies. The two strategies would be considered to have the same performance in produced rankings. However, if one ranking strategy picks flowgates #1~#6 in the actual list while the other ranking strategy picks flowgate #5~#10, it is apparent one strategy performs better than the other as it is always desired to select flowgates that are ranked as high in the actual list as possible. To capture that effect and ensure fairer comparison among ranking strategies, this paper proposes a different approach, ranking correctness index, to evaluate and compare the performance or ranking strategies. C. Ranking Correctness Index

IV. EVALUATION OF RANKING STRATEGIES A. Comparison of Ranking Strategies The three commonly used ranking strategies introduced in Section II-B and the proposed new ranking strategy are reviewed in Table I for comparison purpose. TABLE I COMPARISON OF FOUR RANKING STRATEGIES

Strategy

Binding Hours (BH)

Shadow Prices (SP)

Definition

number of hours in a year the flowgate is binding reduced production cost for 1MW increase of flowgate thermal rating

Unit

Implication

Hrs

reflects the frequency of congestion

$/MWH

reflects the severity of congestion

Firstly, a weight is assigned to each of the flowgates to indicate its significance. The weight of a flowgate is defined as its actual potential benefit (APB) versus the total APB for all flowgates. The higher the APB of a flowgate is, the higher weight the flowgate gets. Table II illustrates an example of flowgate weight assignment. In this example, there are totally 4 flowgates, which are sorted in descending order by APB. It should be noted that APB refers to the actual MISO APC Savings in MISO economic studies. It is obtained from exhaustive simulations for each of the flowgates, and it is only needed for evaluating performance of ranking strategies. Therefore, no exhaustive simulation is needed once the best ranking strategy is determined. Last column of Table II shows the rolling summation of flowgate weights. For example, when we want to select the top two flowgates, the weight rolling summation is the total weights for the first two flowgates, i.e., 0.5+0.4=0.9. It is actually the maximum possible rolling summation because the flowgate is sorted in descending order by APB.

5 Second, we use ranking strategy to rank flowgates, and then look up Table II for the corresponding weight of each flowgate. The weight rolling summation is subsequently calculated, which will differ from the maximum weight rolling summation in Table II, because the flowgate order is typically different from the ideal order in Table II. Last, ranking correctness index is calculated, which is defined as the ratio of weight rolling summation to maximum weight rolling summation. An example of ranking correctness index is illustrated in Table III. Ideally, if a ranking strategy produces the same flowgate order as that is identified using APB, the ranking correctness index will always be 100%. Note that the value of ranking correctness index changes with the number of flowgates of interest, which indicates the trend of performance change of ranking strategies. TABLE II EXAMPLE OF FLOWGATE WEIGHT ASSIGNMENT

Flowgate Ranked by APB A

APB ($M)

Flowgate Weight

Maximum Weight Rolling Summation 0.500

10

0.500

B

8

0.400

0.900

C

1.5

0.075

0.975

D

0.5

0.025

1.000

Total:

20

1

TABLE III EXAMPLE OF RANKING CORRECTNESS INDEX

Rank by a Strategy 1

A

0.500

Weight Rolling Sum 0.500

2

C

0.075

0.575

0.639

3

D

0.025

0.600

0.615

4

B

0.400

1.000

1.000

Flowgate

Weight

Ranking Correctness Index 1.000

D. Evaluation of Ranking Strategies on MISO System In 2012 MISO Market Efficiency Planning Study, all the four ranking strategies were evaluated using ranking correctness index. All MISO internal and seams congested flowgates projected for 2022 Business As Usual future scenario were included in the analysis. Fig.2 depicts the ranking correctness index curve for each of the four strategies. It shows that, when a limited number of flowgates (such as up to 9 flowgates, i.e., 12% of total flowgates) is to be selected out of the total 75 flowgates, Total EPB is the best ranking strategy as it consistently outperforms the other three strategies in terms of ranking correctness index. Total Binding Hours is the worst performer among the four. When sufficiently large number of flowgates are to be picked (such as 20 or greater), all four ranking strategies perform closely to each other. Overall, Total EPB has the highest average correctness index. In practice, only a small number of the congested flowgates will be selected for further analysis, and therefore Total EPB tends to be the best ranking strategy.

Fig. 2. Ranking correctness index versus number of flowgates being selected

V. DISCUSSIONS AND CONCLUSIONS The proposed new ranking strategy, Estimated Potential Benefit (EPB), not only has the most comparable physical meaning to the actual potential benefit, but also has been verified in MISO Market Efficiency Planning Study that it outperforms other three commonly used ranking strategies. Ranking correctness index is proposed to facilitate the quantitative comparison among different ranking strategies. In addition to being used as a ranking strategy, EPB may also serve a purpose in transmission project screening. For instance, suppose transmission owners or stakeholders submit a transmission upgrade project to MISO to evaluate its eligibility for MISO Market Efficiency Project, which requires economic benefit to cost ratio to be 1.25 or higher. Suppose the total project cost is $6M and annual fixed charge rate is 18%. Then, the APB should be no less than $1.35M (=1.25*6*18%). If EPB is $1M, it means the project will likely not meet the benefit to cost threshold since total EPB is often an overestimation of total APB. As the EPBs for all flowgates can be obtained through just one or only a few simulations, it can enable the screening process when lots of projects are to be evaluated for their eligibility for Market Efficiency Project. VI. REFERENCES [1] [2]

[3]

[4]

[5]

[6]

[7]

[8]

FERC 1000, http://www.ferc.gov/industries/electric/indus-act/transplan.asp, accessed in Nov. 2012. Liangying Hecker, Rui Bo, Dale Osborn, John Lawhorn, "Regional and Inter-regional Transmission Planning and Cost Allocation", Proceedings of 2012 IEEE PES General Meeting, San Diego, California, USA, 2012. John Lawhorn, Dale Osborn, Jay Caspary, et al, “The View from the Top,” IEEE Power and Energy Magazine, vol. 7, no. 6, pp. 76-88, November/December 2009. "Joint Coordinated System Plan", https://www.midwestiso.org/Library/Repository/Study/JCSP/JCSP_Rep ort_Volume_1.pdf, accessed in Nov. 2012. "Market Efficiency Analysis --- 2011 Top Congested Flowgate Study", https://www.midwestiso.org/Library/Repository/Study/MTEP/2011%20 Market%20Efficiency%20Analysis.pdf, accessed in Nov. 2012. Rui Bo, Ming Ni, Yang Gu, "Congested Flowgates Ranking Analysis And A Potential New Approach", Proceedings of 2012 IEEE PES Transmission and Distribution Conference & Exposition, Orlando, Florida, USA, 2012 Fangxing Li and Rui Bo, “Congestion and Price Prediction under Load Variation,” IEEE Trans. on Power Systems, vol. 24, no. 2, pp. 911-922, May 2009. Rui Bo and Fangxing Li, "Efficient Estimation of Critical Load Levels Using Variable Substitution Method," IEEE Transactions on Power Systems, vol. 26, no. 4, pp. 2472-2482, November 2011.