pith. sign in

arxiv: 2604.08403 · v1 · submitted 2026-04-09 · 📡 eess.SY · cs.SY

Data-Driven Power Flow for Radial Distribution Networks with Sparse Real-Time Data

Oleksii Molodchyk , Omid Mokhtari , Samuel Chevalier , Mads R. Almassalkhi , Timm Faulwasser This is my paper

Pith reviewed 2026-05-10 17:12 UTC · model grok-4.3

classification 📡 eess.SY cs.SY
keywords data-driven power flowradial distribution networkssparse measurementssensor placementDistFlow modelbehavioral approachvoltage prediction
0
0 comments X

The pith

A data-driven power flow method solves voltage predictions in radial distribution networks using offline data and only 25 percent real-time sensor locations.

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

The paper establishes that power flow calculations for real-time control in radial grids can remain accurate even when real-time measurements cover only a quarter of all possible locations. It does so by combining a behavioral systems approach with the DistFlow equations and pre-computed historical data to reconstruct the missing measurements. A sensor-placement routine based on successive network reductions selects the most informative locations under a fixed budget, preserving the radial structure. On standard test feeders the resulting voltage magnitude errors stay below 0.001 per unit. This removes the need for full network observability that conventional state-estimation methods require.

Core claim

The DDPF algorithm recovers accurate voltage magnitudes from a sparse set of real-time measurements by solving a behavioral data-driven model that is constrained by the DistFlow equations and calibrated on historical operating points; the sensor locations are chosen via an optimal-reduction procedure that accounts for the radial topology and a given measurement budget.

What carries the argument

The sensor-placement problem formulated as an optimal network reduction that selects a minimal set of measurement buses while preserving radial connectivity and DistFlow consistency.

If this is right

  • Real-time voltage control becomes feasible on distribution feeders without installing sensors at every bus.
  • The same framework can be used to certify the minimum number of sensors needed to keep prediction error below a chosen threshold.
  • The method directly supplies the input needed by model-predictive or feedback controllers that act on voltage magnitudes.
  • Because only radial topology is exploited, the approach applies to the majority of existing medium-voltage distribution systems.

Where Pith is reading between the lines

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

  • The same reduction-based placement logic could be tested on unbalanced or meshed networks by replacing DistFlow with a more general power-flow surrogate.
  • If historical data drifts, periodic re-training or online adaptation of the behavioral model would be required to maintain accuracy.
  • The 0.001 p.u. error bound achieved at 25 percent coverage suggests that many existing distribution SCADA systems already contain enough sensors for this style of data-driven reconstruction.

Load-bearing premise

Historical data collected under full observability remains statistically representative of the operating conditions that occur once the sensor set is thinned.

What would settle it

Run the DDPF predictor on a radial feeder whose load and generation statistics have shifted substantially from the historical training set and check whether voltage magnitude errors exceed 0.001 p.u. at more than a few buses.

Figures

Figures reproduced from arXiv: 2604.08403 by Mads R. Almassalkhi, Oleksii Molodchyk, Omid Mokhtari, Samuel Chevalier, Timm Faulwasser.

Figure 1
Figure 1. Figure 1: Different stages of the sensor placement process based on historical [PITH_FULL_IMAGE:figures/full_fig_p005_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: A summary of the considered DistFlow representations. [PITH_FULL_IMAGE:figures/full_fig_p006_2.png] view at source ↗
Figure 4
Figure 4. Figure 4: Maximum voltage magnitude errors over the test dataset and over [PITH_FULL_IMAGE:figures/full_fig_p007_4.png] view at source ↗
Figure 5
Figure 5. Figure 5: Voltage magnitude errors of DDPF in terms of differences to the [PITH_FULL_IMAGE:figures/full_fig_p007_5.png] view at source ↗
read the original abstract

Real-time control of distribution networks requires accurate information about the system state. In practice, however, such information is difficult to obtain because real-time measurements are available only at a limited number of locations. This paper proposes a novel data-driven power flow (DDPF) framework for balanced radial distribution networks. The proposed algorithm combines the behavioral approach with the DistFlow model and leverages offline historical data to solve power flow problems using only a limited set of real-time measurements. To design DDPF under sparse measurement conditions, we develop a sensor placement problem based on optimal network reductions. This allows us to determine sensor locations subject to a predefined sensor budget and to explicitly account for the radial nature of distribution networks. Unlike approaches that rely on full observability, the proposed framework is designed for practical distribution grids with sparse measurement availability. This enables data-driven power flow for real-time operation while reducing the number of required sensors. On several test cases, the proposed DDPF algorithm could demonstrate accurate voltage magnitude predictions, with a maximum error less than 0.001 p.u., with as little as 25% of total locations equipped with sensors.

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

3 major / 2 minor

Summary. The paper proposes a data-driven power flow (DDPF) framework for balanced radial distribution networks that combines the behavioral approach with the DistFlow model. It leverages offline historical data to solve power flow problems using only a limited set of real-time measurements, with sensor locations determined by an optimal network reduction problem that respects the radial topology. The framework is claimed to achieve voltage magnitude predictions with maximum error less than 0.001 p.u. on several test cases using as little as 25% of locations equipped with sensors.

Significance. If the central claim holds under proper out-of-distribution testing and without data leakage between historical learning and validation, the approach could substantially lower the sensor density required for accurate real-time state estimation in distribution grids, addressing a key practical barrier to observability in large-scale radial networks.

major comments (3)
  1. [Abstract] Abstract: the reported maximum voltage error below 0.001 p.u. is given without accompanying error bars, standard deviations across multiple runs, or sensitivity analysis with respect to the choice and volume of historical data, which is load-bearing for the sparse-sensor claim.
  2. [Methodology] Methodology section on behavioral model integration: the description does not specify how the behavioral parameters are identified from offline data (e.g., least-squares, optimization constraints) or whether the same data windows are later reused for validation, raising the risk of circular evaluation.
  3. [Numerical results] Numerical results and test-case description: no explicit statement or table indicates whether evaluation scenarios include temporal shifts, load/renewable distribution changes, or out-of-sample operating points relative to the historical training set; without this, the accuracy does not demonstrate robustness for unseen real-time conditions.
minor comments (2)
  1. [Notation] Notation: ensure that voltage and power variables are defined consistently between the behavioral representation and the DistFlow equations to avoid ambiguity in the combined model.
  2. [Figures] Figures: the sensor-placement diagrams would benefit from explicit labeling of the reduced network nodes and the corresponding measurement locations for each budget level.

Simulated Author's Rebuttal

3 responses · 0 unresolved

We thank the referee for the constructive comments on our manuscript. We address each major comment point by point below, indicating the revisions we will incorporate to improve clarity and rigor.

read point-by-point responses

  1. Referee: [Abstract] Abstract: the reported maximum voltage error below 0.001 p.u. is given without accompanying error bars, standard deviations across multiple runs, or sensitivity analysis with respect to the choice and volume of historical data, which is load-bearing for the sparse-sensor claim.

    Authors: We agree that additional statistical measures would better support the sparse-sensor claim. In the revised manuscript, we will augment the numerical results section with error bars, standard deviations across runs, and a sensitivity analysis on historical data volume and choice. The abstract will be updated to reference these supporting analyses while preserving its concise form. revision: yes

  2. Referee: [Methodology] Methodology section on behavioral model integration: the description does not specify how the behavioral parameters are identified from offline data (e.g., least-squares, optimization constraints) or whether the same data windows are later reused for validation, raising the risk of circular evaluation.

    Authors: The behavioral parameters are identified via a constrained least-squares optimization on the historical trajectory data that incorporates the DistFlow equations. We will revise the methodology section to explicitly detail this identification procedure, including the optimization formulation and constraints. We will also add a statement confirming that validation employs temporally disjoint data windows separate from those used for parameter identification, eliminating any circularity. revision: yes

  3. Referee: [Numerical results] Numerical results and test-case description: no explicit statement or table indicates whether evaluation scenarios include temporal shifts, load/renewable distribution changes, or out-of-sample operating points relative to the historical training set; without this, the accuracy does not demonstrate robustness for unseen real-time conditions.

    Authors: Our test cases already incorporate load and renewable generation profiles with temporal shifts and variations distinct from the historical training data. To make this explicit, we will add a dedicated paragraph and summary table in the numerical results section describing the out-of-sample characteristics of the evaluation scenarios relative to the training set, thereby demonstrating robustness to unseen real-time conditions. revision: yes

Circularity Check

0 steps flagged

No significant circularity in derivation chain

full rationale

The paper's core derivation combines the behavioral approach (an external systems-theoretic framework) with the standard DistFlow equations to form a data-driven power flow solver. Offline historical data is used to identify the behavioral model, after which sparse real-time measurements are fed into the resulting DDPF equations to produce voltage predictions. This is a standard supervised learning + model-based inference pipeline rather than a self-referential loop. No equation reduces to its own fitted parameters by construction, no uniqueness theorem is imported from the authors' prior work, and the sensor-placement optimization is formulated as a separate combinatorial problem whose objective is independent of the subsequent voltage-error metric. The reported test-case accuracy (<0.001 p.u.) is therefore an empirical outcome, not a tautological restatement of the training data. The derivation remains self-contained against external benchmarks.

Axiom & Free-Parameter Ledger

0 free parameters · 1 axioms · 0 invented entities

Only the abstract is available, so the ledger is necessarily incomplete; the main domain assumption is radial balanced topology and the existence of representative historical data.

axioms (1)
  • domain assumption The distribution network is balanced and radial.
    Explicitly stated as the setting for which the DDPF framework is developed.

pith-pipeline@v0.9.0 · 5517 in / 1172 out tokens · 47358 ms · 2026-05-10T17:12:33.986464+00:00 · methodology

discussion (0)

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

Reference graph

Works this paper leans on

22 extracted references · 22 canonical work pages

  1. [1]

    Monitoringbericht 2025,

    Bundesnetzagentur and Bundeskartellamt, “Monitoringbericht 2025,” tech. rep., Bundesnetzagentur and Bundeskartellamt, Bonn, 2025

    work page 2025

  2. [2]

    Data-Driven Control Based on the Behavioral Approach: From Theory to Applications in Power Systems,

    I. Markovsky, L. Huang, and F. D ¨orfler, “Data-Driven Control Based on the Behavioral Approach: From Theory to Applications in Power Systems,”IEEE Control Systems, vol. 43, no. 5, pp. 28–68, 2023

    work page 2023

  3. [3]

    Data-Based Predictive Control Based V oltage Control in Active Distribution Networks,

    Q. Li, Y . Zhu, Z. Tang, Y . Liu, and Y . Shi, “Data-Based Predictive Control Based V oltage Control in Active Distribution Networks,” Electronics, vol. 14, no. 21, p. 4211, 2025

    work page 2025

  4. [4]

    Data-Driven Opti- mal Power Flow: A Behavioral Systems Approach,

    S. Otzen, H. M. H. Wolf, and C. A. Hans, “Data-Driven Opti- mal Power Flow: A Behavioral Systems Approach,”arXiv preprint arXiv:2509.25120, no. arXiv:2509.25120, 2025

    work page arXiv 2025

  5. [5]

    A note on persistency of excitation,

    J. C. Willems, P. Rapisarda, I. Markovsky, and B. L. De Moor, “A note on persistency of excitation,”Systems & Control Letters, vol. 54, no. 4, pp. 325–329, 2005

    work page 2005

  6. [6]

    Towards Data-Driven Multi-Stage OPF,

    O. Molodchyk, P. Schmitz, A. Engelmann, K. Worthmann, and T. Faulwasser, “Towards Data-Driven Multi-Stage OPF,” in2025 IEEE Kiel PowerTech, (Kiel, Germany), pp. 1–6, IEEE, 2025

    work page 2025

  7. [7]

    Optimal capacitor placement on radial distri- bution systems,

    M. Baran and F. Wu, “Optimal capacitor placement on radial distri- bution systems,”IEEE Transactions on Power Delivery, vol. 4, no. 1, pp. 725–734, 1989

    work page 1989

  8. [8]

    Radial Distribution Load Flow Using Conic Programming,

    R. Jabr, “Radial Distribution Load Flow Using Conic Programming,” IEEE Transactions on Power Systems, vol. 21, no. 3, pp. 1458–1459, 2006

    work page 2006

  9. [9]

    Identifiability in the Behavioral Setting,

    I. Markovsky and F. D ¨orfler, “Identifiability in the Behavioral Setting,” IEEE Transactions on Automatic Control, vol. 68, no. 3, pp. 1667–1677, 2023

    work page 2023

  10. [10]

    A Survey of Relaxations and Approximations of the Power Flow Equations,

    D. K. Molzahn and I. A. Hiskens, “A Survey of Relaxations and Approximations of the Power Flow Equations,”Foundations and Trends® in Electric Energy Systems, vol. 4, no. 1-2, pp. 1–221, 2019

    work page 2019

  11. [11]

    Inverter V AR control for distribution systems with renewables,

    M. Farivar, C. R. Clarke, S. H. Low, and K. M. Chandy, “Inverter V AR control for distribution systems with renewables,” in2011 IEEE Interna- tional Conference on Smart Grid Communications (SmartGridComm), (Brussels, Belgium), pp. 457–462, IEEE, 2011

    work page 2011

  12. [12]

    Branch Flow Model: Relaxations and Convexification—Part I,

    M. Farivar and S. H. Low, “Branch Flow Model: Relaxations and Convexification—Part I,”IEEE Transactions on Power Systems, vol. 28, no. 3, pp. 2554–2564, 2013

    work page 2013

  13. [13]

    Branch Flow Model: Relaxations and Convexification—Part II,

    M. Farivar and S. H. Low, “Branch Flow Model: Relaxations and Convexification—Part II,”IEEE Transactions on Power Systems, vol. 28, no. 3, pp. 2565–2572, 2013

    work page 2013

  14. [14]

    Exact Convex Relaxation of Optimal Power Flow in Radial Networks,

    L. Gan, N. Li, U. Topcu, and S. H. Low, “Exact Convex Relaxation of Optimal Power Flow in Radial Networks,”IEEE Transactions on Automatic Control, vol. 60, no. 1, pp. 72–87, 2015

    work page 2015

  15. [15]

    Linear Tracking MPC for Nonlinear Systems—Part II: The Data-Driven Case,

    J. Berberich, J. Kohler, M. A. Muller, and F. Allgower, “Linear Tracking MPC for Nonlinear Systems—Part II: The Data-Driven Case,”IEEE Transactions on Automatic Control, vol. 67, no. 9, pp. 4406–4421, 2022

    work page 2022

  16. [16]

    Towards Optimal Kron-based Reduction Of Networks (Opti-KRON) for the Electric Power Grid,

    S. Chevalier and M. R. Almassalkhi, “Towards Optimal Kron-based Reduction Of Networks (Opti-KRON) for the Electric Power Grid,” in 2022 IEEE 61st Conference on Decision and Control (CDC), (Cancun, Mexico), pp. 5713–5718, IEEE, 2022

    work page 2022

  17. [17]

    Structure-preserving optimal kron-based reduction of radial distribution networks,

    O. Mokhtari, S. Chevalier, and M. Almassalkhi, “Structure-preserving optimal kron-based reduction of radial distribution networks,”arXiv preprint arXiv:2508.15006, 2025

    work page arXiv 2025

  18. [18]

    Optimal Kron- based reduction of networks (Opti-KRON) for three-phase distribution feeders,

    O. Mokhtari, S. Chevalier, and M. Almassalkhi, “Optimal Kron- based reduction of networks (Opti-KRON) for three-phase distribution feeders,”arXiv preprint arXiv:2510.19608, 2025

    work page arXiv 2025

  19. [19]

    MAT- POWER: Steady-State Operations, Planning, and Analysis Tools for Power Systems Research and Education,

    R. D. Zimmerman, C. E. Murillo-S ´anchez, and R. J. Thomas, “MAT- POWER: Steady-State Operations, Planning, and Analysis Tools for Power Systems Research and Education,”IEEE Transactions Power Systems, vol. 26, no. 1, pp. 12–19, 2011

    work page 2011

  20. [20]

    On the implementation of an interior- point filter line-search algorithm for large-scale nonlinear programming,

    A. W ¨achter and L. T. Biegler, “On the implementation of an interior- point filter line-search algorithm for large-scale nonlinear programming,” Mathematical Programming, vol. 106, no. 1, pp. 25–57, 2006

    work page 2006

  21. [21]

    MOSEK ApS,MOSEK Optimizer API for Julia Release 11.0.25, 2019

    work page 2019

  22. [22]

    Stan- dardlastprofile Strom,

    BDEW Bundesverband der Energie- und Wasserwirtschaft e.V ., “Stan- dardlastprofile Strom,” 2025. Accessed: March 28, 2026

    work page 2025