Review and analysis of investment decision making algorithms in long-term agent-based electric power system simulation models

Abstract

Long-term electric power system planning models are frequently used to provide policy support in the context of the ongoing transition towards a low-carbon electric power system. In a liberalized market, this transition relies on generation company investment decisions. These decisions are shaped by both economic and behavioral factors. Agent-based modeling allows the incorporation of both factors in the description of the investment decision making process. Nevertheless, there are several challenges associated with the design of agent-based models such as the definition of the model structure and its lack of transparency. In this study, we aim to increase the transparency of investment decision making algorithms by shedding light on how implicit assumptions of the price projection methods used in these algorithms impact model results. To achieve this goal, we developed a core long-term agent-based model to assess different investment decision making algorithms from the literature and we introduced a novel price projection method based on optimization modeling. Our results show that investment decisions vary enormously depending on the assumptions and parameters used in investment decision making algorithms. We also found that our proposed price projection method is robust to parametric deviations. Thus, the proposed investment decision making algorithm enables agent-based modelers to mitigate the potential impacts of hidden implicit assumptions.

Simulating institutional heterogeneity in sustainability science

“…Among the model combinations presented here, agent-based IAMs have simulated heterogeneous agents (e.g., individual goals, bounded-rationality, imperfect foresight) in macro-energy systems ( 30 , 45 ). Hybrid EEO/ABM models have examined short-term (e.g., market bidding) as well as long-term (e.g., planning and transition) decisions by energy firms, for example, by embedding a detailed optimization in a larger ABM considering heterogeneous firm characteristics and preferences ( 72 , 73 ). These hybrid models provide an avenue to improve the institutional and behavioral realism of the modeled decisions, while also creating new analytical challenges with respect to computational complexity and tractability.…”

Assessing the relative impacts of maximum investment rate and temporal detail in capacity expansion models applied to power systems

“…1.2. Reviewing the impact of maximum investment rates While they can be a point of interest in best practices papers [2] and papers with a strong review element [43,44], maximum investment rates have comparatively received far less attention from model or case study focussed papers 4 . Investment constraints such as technology growth rates or market shares are typically used as calibration techniques to ensure models propose credible investment pathways.…”

Identification of Weak Buses for Optimal Load Shedding Using Differential Evolution

“…The genetic algorithm is a method used to search for the optimal solution by simulating the natural evolutionary process. It involves encoding, population initialization, fitness function, genetic operators, and elite reservation strategy [23], which can be summarized as follows:…”