Modelling to Generate Continuous Alternatives: Enabling Real-Time Feasible Portfolio Generation in Convex Planning Models

Michael Lau, Xin Wang, Neha Patankar, Jesse D. JenkinsarXiv 2024 · DOI · arXiv

Introduction

Macro-energy systems (MES) models are critical for long-term energy planning, typically optimized to find the "least-cost" solution. However, real-world decision-making involves competing priorities—such as land use, job creation, and local pollution—that are difficult to capture in a single objective function. While "Modelling to Generate Alternatives" (MGA) techniques identify a set of "near-optimal" solutions within a specified budget slack, they are computationally expensive and provide only discrete data points. This paper introduces Modelling to Generate Continuous Alternatives (MGCA), a post-processing framework that transforms these discrete model runs into a continuous, searchable space of feasible energy portfolios, enabling real-time stakeholder interaction.

Methodology: The MGCA Framework

The MGCA framework leverages the mathematical properties of convex optimization to bridge the gap between slow, high-fidelity simulations and fast, interactive exploration. The process involves four key stages:

  1. Dimensionality Reduction: Thousands of operational variables (e.g., hourly dispatch) are removed, retaining only strategic "capacity" decisions and key metrics like total cost and emissions. This reduces the problem size by 3–5 orders of magnitude.
  2. Convex Hull Exploration: Since the underlying planning models are convex, any linear combination of feasible solutions is also feasible. MGCA treats original MGA solutions as vertices of a convex hull. The core "Aha" moment is that the full model is not required to explore the feasible space once these "corners" have been identified.
  3. Exploratory Optimization: Instead of re-solving the massive original problem, MGCA solves a small optimization problem over the weights of these vertices. This allows for "live" searching for portfolios that meet new, user-defined constraints and enables budget flexibility on the fly (e.g., deriving a 5% slack space from a 10% slack set).
  4. Metric Interpolation: Linear algebra estimates the costs and emissions of interpolated portfolios without requiring full hourly simulations.
The MGCA framework illustrates how feasible spaces (A) are mapped to capacity variables and metrics. It shows interpolation for affine capacity metrics (B), operational costs (C), and non-cost affine operational metrics (D).

Validation and Results

The authors validated MGCA using the GenX open-source capacity expansion model applied to a 3-zone representation of the ISO New England (ISONE) power system. The model featured 8,760-hourly resolution and linearized unit commitment.

The results demonstrate a significant performance leap:

  • Computational Speed: Generating a new interior solution took approximately 0.6 ms, a million-fold speedup compared to the ~600 seconds required for a full model run.
  • Interactive Trade-offs: An 11-point Pareto frontier for cost vs. emissions was generated in just 0.09 seconds.
  • Accuracy: Capacity metrics were exact. System-wide operational costs were overestimated by only ~3% on average, while emissions estimates remained within a 10% error margin. Importantly, interpolated solutions remain physically feasible when plugged back into the full-scale simulator.

Discussion and Conclusion

MGCA shifts the energy planning workflow from batch processing to real-time interaction. By navigating the interior of the feasible space using simple linear algebra, researchers can provide instant feedback to stakeholders during workshops. While the method currently relies on model convexity, future work will focus on integrating MGCA into interactive dashboards and developing "down-scaling" methods to help decide exactly where new infrastructure should be built at the local level.

Summary updated

Abstract

Decarbonization provides new opportunities to plan energy systems for improved health, resilience, equity, and environmental outcomes, but challenges in siting and social acceptance of transition goals and targets threaten progress. Modelling to Generate Alternatives (MGA) provides an optimization method for capturing many near-cost-optimal system configurations, and can provide insights into the tradeoffs between objectives and flexibility available in the system. However, MGA is currently limited in interactive applicability to these problems due to a lack of methods for allowing users to explore near-optimal feasible spaces. In this work we describe Modelling to Generate Continuous Alternatives (MGCA), a novel post-processing algorithm for convex planning problems which enables users to rapidly generate new interior solutions, incorporate new constraints, and solve within the space with convex objectives. MGCA begins with a dimensionality reduction to capacity decisions and metric values. We then take advantage of convex combinations to generate interior points by allowing user weight specification and encoding convex combinations in an optimization problem with user-defined additional constraints and objective. Dimensionality reduction enables this problem to solve in tenths of a second, suitable for analysis in interactive settings. We discuss the interpolation of capacity and operational metric values, finding capacity metrics can be perfectly interpolated while operational metrics remain within the feasible range of the points used to create them. We demonstrate interpolated solutions can be exported and re-solved with an economic dispatch model to provide operational metric values consistent with least-cost decision-making and show interpolated metric values are generally within 10% of the optimal value.
BibTeX
@misc{lau2024modelling,
  author = {Michael Lau and Xin Wang and Neha Patankar and Jesse D. Jenkins},
  title = {{Modelling to Generate Continuous Alternatives: Enabling Real-Time Feasible Portfolio Generation in Convex Planning Models}},
  year = {2024},
  eprint = {2411.16887},
  archivePrefix = {arXiv},
  url = {https://arxiv.org/abs/2411.16887},
}
Figure