Ancillary Services Acquisition Model: considering market interactions in policy design

TL;DR

This paper introduces an open-source agent-based model, ASAM, to evaluate complex interactions in electricity market design for ancillary services, aiding policy development amid sector changes.

Contribution

It develops a novel, open-source agent-based model combining MESA and PyPSA to analyze ancillary services and market interactions with detailed algorithms and performance indicators.

Findings

Order types in redispatch markets have minimal risks for market parties.

Large orders can cause ramping issues and additional imbalance trading.

All-or-none redispatch design may lead to over-procurement and induced imbalances.

Abstract

A rapidly changing electricity sector requires adjusted and new ancillary services, which enable the secure and reliable operation of the electricity system. However, assessments and policy advice regarding ancillary services and market design lack methods to evaluate the complex interaction of markets and services. Therefore, this paper contributes an open-source agent-based model to test design options for ancillary services and electricity markets. The Ancillary Services Acquisition Model (ASAM) combines the agent-based modeling framework MESA with the toolbox Python for Power System Analysis (PyPSA). The model provides various design parameters per market and agent-specific strategies as well as detailed clearing algorithms for the day-ahead market, intra-day continuous trading, redispatch, and imbalances. Moreover, ASAM includes numerous policy performance indicators, including a novel price mark-up indicator and novel redispatch performance indicators. A stylized simulation scenario verified and validated the model and addressed a redispatch design question. The results displayed the following implications from order types in redispatch markets with multi-period all-or-none design: (1) The order design provides few risks for market parties, as orders are fully cleared. (2) Large orders may lead to dispatch ramps before and after the delivery period and may cause ramp-risk mark-ups as well as additional trading of imbalances on intra-day. (3) All-or-none design in a liquid situation leads to the over-procurement of redispatch by the grid operator, as orders cannot be partially activated. Moreover, it is likely that the grid operator induces imbalances to the system by incomplete redispatch activation (i.e. upward and downward redispatch volumes are not equal).