Day-Ahead Reserve Capacity Procurement based on Mixed-Integer Bilevel Linear Programming

Day-Ahead Reserve Capacity Procurement based on Mixed-Integer Bilevel Linear Programming

Introduction

A. Introduction to the electricaty market in Nordic countries

1. Nord Pool

This market is shared by the four Nordic countries

This market closed at 12:00 before theday of operation

2. the Regulation Market

This market is the reserve market for trading FRR-M

This market at 45min before operation

The providers of the procured reserve capacity have the mandatory obligation toprovide regulating power bids in the regulation market

3. Availabiltity Market

This market closes at 09:30 before the day of operation

TSO is the only buyer of this market

Note: The total regulating power bids submitted from Denmark (Availability Market) to the common regulation market consist of 2 aspects

Mandatory regulating power bids

Voluntary regulating power bids

B. Problems need to be considered in the dispatch of power

Problem 1: Location of the reserve

In case that the capacoty of the interconnection is fully utilizered or unabailable

Problem 2: The day-ahead forecast error of the system wind power production

C. Contribution of this paper

Approach using the market model based on unit commitment (UC) to determine the volume of reserve capacity to be procured in the day-ahead availability market in Denmark

Model the sequential clearing of the reserve availability market and energy market to account for the forecast error of system wind power and demand

Day-ahead Reserve Capacity Procurement Using Mixed-integer Bilevel Linear Programming

A. Diagram of bilevel formulation

Upper level problem

TSO determines the amount of reserve capacity to procure and the corresponding EENS

Lower level problem

The clearing of energy market is simulated by a UC formulation

Why choose the UC formulation ?

1. UC can provide the information of spare capacity of generators

Note: This spare capacity can be treated as the voluntary regulating power bids

2. The voluntary bids may also include generators that are not selected in the energy market (not considered in this paper)

3. Bidding curves are not available

Possible solution to this problem: forecast the bidding curves

The soultion may introduce another unceratinty

B. Mathematical formulation

Upper level problem of BLPP

minimize the operating cost

Cost of pruchase reserve capacity

Cost of loss of load

The load loss due to generators outage (Not considered in this paper)

System wind forecast error

Demand forcaset error

Objective function of the upper level problem

The cost of reserve capacity for up and down regulation

The cost of EENS due to forecast errors

Lower level problem to model the energy market clearing process

Objective function of the lower level problem

Constraints

Power balance in spot market

Energy sale limits in spot market

Start up cost

Other standard constraints for generation units

C. Calcluation of the Energy not Served

Formulation of the energy not served (ENS) due to the forecasting errorof the net demand

Formulation of the available up-regulating reserve capacity in thesystem

D. Solution algorithm

Case Study and Result

A. Case and data description

B. No reserve capacity procurement

This could be the case that if the reserve capacity price is much higher than the VOLL

With reserve capacity procurement

Case 1: with no procurement of reserve capacity

Case 2: 10 MW reserve capacity is procured from Unit 5

Case 2 instead increases the EENS than that of Case 1

font color=\"#231f20\" size=\"2\

This implies that the total available reserve capacity does not necessarily increase when procuring morereserve capacity.

Case 3 & Case 4: font color=\"#231f20\" size=\"2\

Reason: font color=\"#231f20\" size=\"2\

子主题

Case 5: an additional 10 MW reserve capacity procured from Unit 6 compared with Case 4

The EENS remains the same and does not decrease compared with VCae 4

Case 6: the additional 10 MW reserve capacity is producedfrom Unit 2 instead

The EENS become zero

Conclusions

Modelled the sequential clearing of the reserve availability market and energy market as a stochastic MILP-BLPP problem accounting for the forecast error of system wind power and demand.

Procuring more reserve capacity does not necessarily increase the total available reserve capacity or decrease the EENS if it reduces the total online generation capacity by pushing a unit with a large capacity offline.