PREreview del Development and Validation of a Process Model and Open-Source Process Simulator for Microalgae-Based Tertiary Phosphorus Recovery

Summary

This study presents PM2 (Phototrophic-Mixotrophic Process Model), an updated lumped pathway metabolic model for microalgae integrating both photoautotrophic and heterotrophic metabolisms, implemented within QSDsan — an open-source Python-based platform for quantitative sustainable design. The model is calibrated and validated against batch kinetic assays and 45 days of continuous online monitoring data from the full-scale EcoRecover system (568 m³·d⁻¹), achieving effluent phosphorus prediction within 0.02–0.04 mg-P·L⁻¹. The work addresses a genuine gap: the absence of validated, simulator-ready mechanistic models has been a recognized barrier to industrial adoption of microalgae-based tertiary treatment. This contribution is timely, methodologically sound, and of significant practical relevance.

Strengths

1. Genuine novelty and practical relevance. Bridging the gap between metabolic modeling and deployable process simulation is a meaningful contribution. The integration of PM2 into QSDsan as a reusable Process module, combined with a real full-scale validation dataset, makes this more than an academic modeling exercise — it provides a direct tool for engineers and utilities considering microalgae-based phosphorus recovery.

2. Full-scale validation with rich temporal resolution. Validation against 45 days of continuous 5-minute interval monitoring data from a commercial-scale facility (EcoRecover, Clearas Water Recovery) is a considerable strength. Most comparable microalgae modeling studies rely on lab-scale or pilot-scale data; full-scale validation substantially increases confidence in the model's transferability to real operational conditions.

3. Rigorous uncertainty and sensitivity analysis. The use of global sensitivity analysis (GSA) to prioritize parameters for calibration is methodologically appropriate and well-reported. Connecting GSA outputs directly to calibration decisions strengthens the credibility of the calibrated parameter set and avoids the common pitfall of arbitrary manual tuning.

4. Inclusion of diel light/temperature cycling. Providing dynamic light intensity and temperature inputs with diel variation as model drivers — rather than assuming steady-state environmental conditions — is a notable improvement over many existing models and better reflects real operational behavior of outdoor or semi-outdoor photobioreactors.

5. Open-source ethos and reproducibility. Implementing the simulator within QSDsan and making the code openly accessible is strongly aligned with good scientific practice. This significantly lowers the barrier for other researchers and engineers to adopt, adapt, or benchmark against this work.

Weaknesses and Points Requiring Clarification

1. Single-site validation limits generalizability claims. The model is calibrated and validated exclusively on the EcoRecover system at one facility. While the full-scale nature of this dataset is valuable, the community structure, reactor geometry, climate, and operational strategy of this system may differ substantially from other microalgae-based tertiary treatment configurations (e.g., raceway ponds, tubular PBRs, high-rate algal ponds). The authors should more explicitly discuss the expected boundaries of model transferability and what re-calibration would be required for different system types.

2. Lumped community modeling assumptions need more scrutiny. PM2 treats the microbial community as a lumped, single-organism representation. The preprint acknowledges this is a simplification of a mixed microbial community, but the discussion of how community shifts (e.g., seasonal succession, bacterial contamination events, cyanobacterial blooms) could affect model performance is limited. Reviewers and practitioners would benefit from a more thorough treatment of when the lumped assumption is expected to break down.

3. Nitrogen dynamics are largely secondary. The focus on phosphorus prediction is entirely justified given the study objectives, but the preprint gives relatively limited attention to nitrogen dynamics and their coupling with phosphorus uptake. Given that nitrogen availability is a key driver of algal luxury phosphorus uptake, a more explicit discussion of how nitrogen model components were treated — and whether nitrogen prediction accuracy was evaluated independently — would strengthen the work.

4. Calibration data independence could be clearer. It is not entirely clear from the preprint text whether the batch kinetic assay data and the continuous EcoRecover monitoring data are fully independent datasets, or whether some batch data informed parameter estimation that was then tested on the full-scale data. A clearer explicit statement of the calibration/validation data split and any risk of data leakage across the two would improve methodological transparency.

5. Practical guidance for practitioners is underdeveloped. The paper is strong on model mechanics and validation metrics, but relatively thin on practical guidance: How should a water utility or engineering firm set up this simulator for a new site? What minimum monitoring infrastructure is needed? What are the critical parameters most requiring site-specific data? A concise "practical implementation" section or even a dedicated paragraph in the discussion would substantially increase the work's impact with its intended audience.

Recommendations

1. Add a dedicated section or paragraph explicitly scoping the model's domain of applicability — what reactor types, climate regimes, and operational conditions it has been validated for, and what adaptations would be needed outside those conditions.

2. Expand the discussion on mixed community dynamics and their interaction with the lumped modeling assumption, including how users should interpret prediction deterioration if community composition shifts significantly.

3. Clarify the data independence between batch calibration and full-scale validation more explicitly in the Methods.

4. Consider including a brief worked example or decision tree for prospective users on how to implement the simulator for a new site — even as supplementary material — to maximize uptake by practitioners.

5. Minor: ensure that all acronyms (WRRF, PBR, GSA, etc.) are defined at first use in the main text, and that figure captions are self-contained for readers who navigate the figures independently.

Overall Assessment

This is a high-quality preprint making a substantive contribution to the field of microalgae process engineering. The combination of a mechanistically updated model, rigorous uncertainty analysis, and full-scale validation within an openly accessible simulation platform is a genuine step forward. The main limitations relate to single-site generalizability and the depth of guidance provided for practical adoption — both addressable in revision. The work is recommended for publication with minor to moderate revisions, and the open-source implementation deserves particular recognition as a model for the field.

Note: This preprint appears to have been subsequently published in npj Clean Water (2025). This review is based on the ChemRxiv preprint version (v2) and is submitted as an open peer review record.

Competing interests

The author declares that they have no competing interests.

Use of Artificial Intelligence (AI)

The author declares that they did not use generative AI to come up with new ideas for their review.