pith. sign in

arxiv: 2604.18926 · v2 · pith:VQ5UHYNSnew · submitted 2026-04-21 · 🧮 math.OC · cs.SY· eess.SY

High-Fidelity Capacity Expansion Planning for Puerto Rico's Electric Power System

Elizabeth Glista , Tomas Valencia Zuluaga , Amelia Musselman , Minda Monteagudo , Juliette Franzman , Jean-Paul Watson This is my paper

Pith reviewed 2026-05-10 02:58 UTC · model grok-4.3

classification 🧮 math.OC cs.SYeess.SY
keywords capacity expansion planningPuerto Rico power systemcombined cycle capacitypower system reliabilitystochastic optimizationunit commitmenttransmission modelinggenerator retirement
0
0 comments X

The pith

Puerto Rico requires at least 1.5 GW of new combined-cycle gas capacity beyond planned projects to maintain grid reliability.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

This paper develops a high-resolution optimization model for planning generation and storage investments in Puerto Rico's electric power system. The model jointly decides which new plants to build, which old ones to retire, and how to run the system hour by hour while respecting engineering limits and fuel constraints. It incorporates stochastic scenarios for load, renewable output, and the high outage rates of existing thermal units. The results show that an optimal portfolio adds at least 1.5 GW of H-class combined-cycle capacity mainly to replace unreliable legacy generators rather than to serve new demand. These additions remove modeled load shedding and restore strong reserve margins across stressed conditions.

Core claim

The study finds that an optimal portfolio includes at least 1.5 GW of new H-class combined cycle capacity beyond planned projects. These additions are needed mainly to replace unreliable legacy thermal units rather than to serve new load. The new combined cycle units eliminate modeled bulk-system load shedding and restore a strong reserve margin, even under stressed load and outage conditions.

What carries the argument

A stochastic capacity expansion model that co-optimizes new generation and storage investments with thermal retirements, using nodal transmission at 38 kV and above, hourly chronological operations, explicit unit commitment with ramping and startup costs, system-wide fuel constraints, and scenarios for load, renewables, and outages.

If this is right

  • The least-cost plan calls for new combined-cycle capacity even when future load growth is modest.
  • Planned projects alone leave the system exposed to load shedding under high-outage scenarios.
  • Relaxation of near-term renewable targets allows the model to select thermal replacements that improve reliability.
  • System-wide fuel supply limits influence the scale and location of new thermal additions.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • The same level of operational detail could be applied to other island systems facing aging thermal fleets.
  • Refining the outage rate assumptions with recent operational data would produce updated capacity recommendations.
  • Pairing the new combined-cycle units with additional storage might further reduce fuel consumption while preserving reliability.

Load-bearing premise

The input data from LUMA, PREPA, DOE, and public sources, together with the assumed high forced outage rates of legacy units and the stochastic scenarios, accurately represent real-world conditions and future uncertainties.

What would settle it

If measured forced outage rates of legacy thermal units are substantially lower than the modeled values, or if planned projects deliver higher reliability than assumed, then the modeled requirement for at least 1.5 GW of additional combined-cycle capacity would no longer hold.

Figures

Figures reproduced from arXiv: 2604.18926 by Amelia Musselman, Elizabeth Glista, Jean-Paul Watson, Juliette Franzman, Minda Monteagudo, Tomas Valencia Zuluaga.

Figure 4
Figure 4. Figure 4: Sensitivities of first-stage decisions to different baseline fleets and fuel supplies. For these cases, we saw [PITH_FULL_IMAGE:figures/full_fig_p023_4.png] view at source ↗
Figure 5
Figure 5. Figure 5: Sensitivities of first-stage decisions to future load scenarios A, B, and C. For these (F2) cases, when we [PITH_FULL_IMAGE:figures/full_fig_p023_5.png] view at source ↗
Figure 9
Figure 9. Figure 9: Comparison of capacity factor data for different solar PV and wind plants in Puerto Rico’s power system. The [PITH_FULL_IMAGE:figures/full_fig_p037_9.png] view at source ↗
read the original abstract

This study presents a mathematical optimization framework and analysis to inform practical long-term investment planning in Puerto Rico's electric power system. We utilize a high-resolution capacity expansion planning model to identify least-cost generation and storage investments that improve reliability. The model co-optimizes new investments with thermal retirements and includes detailed dispatch, unit commitment, fuel selection, storage operation, engineering limits, system constraints, fuel supply limits, and load balance. Key advances over prior studies on Puerto Rico's system include: (i) Nodal transmission representation at 38 kV and above; (ii) hourly chronological simulation for representative days; (iii) explicit unit commitment for existing and new thermal units with realistic ramping, minimum up and down times, and startup costs; (iv) system-wide fuel supply constraints; and (v) operational scenarios reflecting load variability, renewable availability, and high forced outage rates in legacy units. Using data from LUMA, the Puerto Rico Electric Power Authority (PREPA), U.S. Department of Energy, and public sources, the study builds representative Puerto Rico systems for 2024 and 2030, with the latter including planned generation and storage projects. It tests scenarios with different future load levels, fuel supply assumptions, planned additions, and allowed technologies. Under the study assumptions, least-cost portfolios that meet reliability targets require about 1.5 GW or more of new H-class combined cycle capacity, in addition to planned projects. These additions mainly replace unreliable legacy thermal units rather than serve new load. The new CC investments eliminate modeled load shedding in the bulk system and restore a robust reserve margin, even under stressed load and outage conditions.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

2 major / 3 minor

Summary. The paper develops a high-resolution capacity expansion optimization model for Puerto Rico's electric power system that co-optimizes new generation/storage investments with thermal unit retirements. It incorporates nodal transmission (38 kV+), hourly chronological dispatch with unit commitment (ramping, min up/down times, startup costs), system-wide fuel constraints, and stochastic scenarios for load, renewables, and outages. Using 2024/2030 test systems built from LUMA/PREPA/DOE/public data and including planned projects, the analysis concludes that at least 1.5 GW of new H-class combined-cycle capacity beyond planned projects is required in the optimal portfolio, primarily to replace unreliable legacy thermal units, eliminate bulk-system load shedding, and restore reserve margins under stressed conditions.

Significance. If the input assumptions hold, the work provides a methodologically advanced least-cost planning tool for a real-world system undergoing energy transition, with explicit co-optimization of retirements and investments plus stochastic reliability modeling that goes beyond typical long-term studies. Credit is due for the detailed engineering constraints, use of real data sources, and quantitative result on the 1.5 GW figure; these elements make the framework potentially useful for policy if validated.

major comments (2)
  1. [Abstract and §5] Abstract and §5 (Results): The central claim that the 1.5 GW of new H-class CC is needed 'mainly to replace unreliable legacy thermal units rather than to serve new load' is load-bearing for the headline result, yet the manuscript does not appear to include a decomposition or sensitivity run isolating the contribution of legacy forced outage rates versus load growth or renewable variability. Without this (e.g., a table comparing optimal portfolios under baseline vs. reduced outage rates), the attribution remains an interpretation rather than a demonstrated output of the co-optimization.
  2. [§4.2 and §6] §4.2 (Model formulation) and §6 (Scenarios): The stochastic scenarios for outages are described as reflecting 'the high forced outage rates of legacy units,' but no explicit validation against historical LUMA/PREPA outage data or sensitivity table on these rates is referenced. Because the abstract states that the new CC units 'eliminate modeled bulk-system load shedding' under these rates, a load-bearing robustness check is required to confirm the 1.5 GW figure does not shift materially under plausible lower rates.
minor comments (3)
  1. [Table 2] Table 2 (2030 test system): Planned projects are listed but the exact MW breakdown of 'beyond planned' additions is not cross-referenced to the optimization output table; adding a column for incremental capacity would improve traceability.
  2. [§3.1] Notation in §3.1: The distinction between 'representative days' and full-year stochastic sampling is introduced but the mapping from scenarios to representative days is not shown in an equation or pseudocode; a small diagram or equation would clarify.
  3. [References] References: Several DOE and LUMA reports cited in the data section lack DOIs or access dates; standardizing the bibliography would aid reproducibility.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for the constructive feedback on our capacity expansion model for Puerto Rico. The comments highlight opportunities to strengthen the attribution of results and the robustness of outage modeling. We address each point below and will revise the manuscript accordingly.

read point-by-point responses

  1. Referee: [Abstract and §5] The central claim that the 1.5 GW of new H-class CC is needed 'mainly to replace unreliable legacy thermal units rather than to serve new load' is load-bearing for the headline result, yet the manuscript does not appear to include a decomposition or sensitivity run isolating the contribution of legacy forced outage rates versus load growth or renewable variability. Without this (e.g., a table comparing optimal portfolios under baseline vs. reduced outage rates), the attribution remains an interpretation rather than a demonstrated output of the co-optimization.

    Authors: We agree that the current attribution relies on interpretation of the co-optimization under baseline conditions. To address this directly, we will add a new sensitivity analysis in revised §5 (and update the abstract if needed) that compares optimal portfolios under baseline legacy outage rates versus reduced rates (e.g., 50% lower to simulate improved maintenance). This will include a table quantifying differences in new CC capacity, load shedding, and reserve margins, isolating the reliability-driven need. The core 1.5 GW finding is expected to hold but will now be demonstrated rather than interpreted. revision: yes

  2. Referee: [§4.2 and §6] The stochastic scenarios for outages are described as reflecting 'the high forced outage rates of legacy units,' but no explicit validation against historical LUMA/PREPA outage data or sensitivity table on these rates is referenced. Because the abstract states that the new CC units 'eliminate modeled bulk-system load shedding' under these rates, a load-bearing robustness check is required to confirm the 1.5 GW figure does not shift materially under plausible lower rates.

    Authors: The outage rates in §4.2 and §6 are based on aggregated data from PREPA/LUMA reports and comparable Caribbean systems as cited in the data sources section. We did not include a dedicated historical validation table in the original version. We will revise §4.2 to add explicit references to the source data and include a sensitivity table in §6 varying the rates (baseline, -30%, -50%). This will confirm robustness of the 1.5 GW requirement and load-shedding elimination. If lower rates reduce the need, we will report the threshold explicitly. revision: partial

Circularity Check

0 steps flagged

Optimization model derives portfolio results from external data and constraints with no circular reduction

full rationale

The paper constructs a high-resolution capacity expansion optimization model that co-optimizes generation/storage investments and thermal retirements subject to nodal transmission, hourly unit commitment, fuel constraints, and stochastic scenarios for load/renewables/outages. All inputs (data from LUMA/PREPA/DOE/public sources, forced outage rates, engineering limits) are stated as exogenous; the reported optimal need for ≥1.5 GW new H-class combined cycle capacity is produced as the solver output under those inputs rather than being defined in terms of itself or obtained by fitting a parameter to a related quantity and relabeling it a prediction. No equations, self-citations, or imported uniqueness theorems reduce the central claim to the inputs by construction.

Axiom & Free-Parameter Ledger

2 free parameters · 3 axioms · 0 invented entities

The central claim rests on standard power-system optimization assumptions and external data whose specific numerical values and validation are not detailed in the abstract.

free parameters (2)
  • Future load growth and fuel supply parameters
    Vary across planning scenarios but exact numerical values and sources are not stated in the abstract.
  • Forced outage rates for legacy thermal units
    Described as high and central to the reliability assessment; specific rates drawn from data but not enumerated.
axioms (3)
  • domain assumption Hourly chronological operations on representative days adequately represent annual system behavior
    Invoked to model dispatch, unit commitment, and storage operations.
  • domain assumption Stochastic scenarios capture the joint variation of load, renewable availability, and forced outages
    Used to evaluate reliability under stressed conditions.
  • domain assumption Unit commitment constraints with realistic ramping, minimum up/down times, and startup costs correctly model thermal generator flexibility
    Applied to both existing and new thermal units.

pith-pipeline@v0.9.0 · 5629 in / 1703 out tokens · 58351 ms · 2026-05-10T02:58:05.448792+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.