Pith. sign in

REVIEW 4 major objections 5 minor 18 references

Equilibrio de carga para transformadores de distribucion electrica mejorando la calidad de servicio en fin de linea

T0 review · 4 major / 5 minor · reviewed 2026-08-15 · deepseek-v4-flash

Pith's one-line read A genetic algorithm can identify the fewest customer phase changes that improve both phase balance and end-of-line voltage.

desk verdict A sensible but incomplete conference short-paper: the GA objective is coherent and the authors are candid about the toy test case, but the missing results section leaves the central claim unsupported. read the letter →

arxiv 2505.09235 v2 pith:U5MGA33L submitted 2025-05-14 cs.NE

classification cs.NE
keywords phasebalancingloadre-phasinggeneticalgorithmlow-voltagedistributionvoltagedropservicequalitysmartmeterstransformer
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

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

The reading

Unbalanced loads on a three-phase distribution circuit waste energy, stress equipment, and pull down voltage at the far end of the line. The paper proposes a genetic algorithm whose search space is the assignment of each customer to one of the three phases; each candidate plan is scored by how much it improves phase balance, how much it reduces the end-of-line voltage drop, and how few customers have to be reconnected. The authors say this yields the optimal set of re-phasing moves, so a utility with smart-meter data can turn the result into a concrete low-cost field action. The method is validated on a simplified 60-customer, single-circuit network with realistic transformer and cable parameters, under the assumption that every customer draws the same constant load.

What carries the argument

The carrying mechanism is the genetic algorithm plus its fitness function: $f(S) = 1 - \frac{r B/B_{\max} + t PV/PV_{\max} + v N/N_{\max}}{r+t+v}$. Here $B$ is the phase-imbalance index over the analysis period, $PV$ is the maximum voltage drop at the final node, $N$ is the number of customers whose phase changes, and $B_{\max}, PV_{\max}, N_{\max}$ are expected maxima that normalize the three objectives. The coefficients $r,t,v$ make the fitness a convex combination, so the operator can decide whether balance, voltage, or few changes matters more. The genetic algorithm's chromosome is the phase label (1, 2, or 3) for each customer, and the search returns the assignment with the highest fitness.

What would settle it

Run the algorithm on the same 60-customer network and 112233 connection scheme, but replace the fixed 200 W loads with a measured 24-hour smart-meter profile from a real low-voltage circuit; then apply the recommended phase reassignment. If any hour of the day shows the end-of-line voltage drop or the imbalance index worse than the original connection, the claim that the method improves both balance and voltage under real conditions is false.

Watch

Extended reading notes

Core claim

The central claim is that a single weighted fitness function can guide a genetic algorithm to phase-reassignment plans that improve both load balance and voltage quality at the end nodes while minimizing the number of changes. Each solution $S$ is an assignment of every customer to a phase; the fitness $f(S)$ combines the period imbalance index $B$, the worst end-of-line voltage drop $PV$, and the number of phase changes $N$, each normalized by expected maxima, with weights the operator chooses. Maximizing this fitness selects plans that trade the three objectives off in a controllable way. Evaluated with a power-flow simulation on a 60-customer circuit built from real low-voltage network characteristics, the algorithm identifies which loads to move, to which phase, and how many visits the work requires.

Load-bearing premise

All 60 customers are modeled as drawing exactly 200 W active power and 10% of that value in reactive power for the whole period, so the recommended re-phasing is tuned to a constant-load world; with fluctuating real loads it could fail to improve balance or voltage, or even make them worse.

Editorial extensions

If this is right

  • A utility can turn the algorithm's output directly into a work order: the list of customers whose phase connection changes, and the target phase for each.
  • Operators can tune the three weights to favor service quality (balance and voltage) or cost (fewer field changes) before running the search.
  • The same weighted objective can be applied to any transformer circuit once its $B_{\max}$, $PV_{\max}$, and $N_{\max}$ constants are set to match that circuit's expected ranges.
  • Balancing is achieved by reconnecting existing service drops rather than adding capacitors or filters, so the main cost is the number of physical changes, which is what the algorithm minimizes.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • A rolling implementation is a natural extension: re-run the search on each day's smart-meter profiles and re-phase only when the optimized plan changes enough to justify the visits.
  • The normalized weighted objective makes it easy to add other operational goals, such as limiting neutral current or cutting resistive losses, by inserting one more normalized term into the fitness function.
  • Robustness to mislabeled phases is untested; adding random noise to the assumed phase of a fraction of customers would show how much error the planning step tolerates.
  • The static-load evaluation covers one network shape; testing the same fitness function on longer feeders, multiple circuits on one transformer, or networks with distributed generation would show whether the method scales.
Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

4 major / 5 minor

Summary. The manuscript proposes a genetic-algorithm-based method for reassigning single-phase customers across the three phases of a low-voltage distribution transformer. Three objectives are defined: phase imbalance (Eq. 2), maximum voltage drop over the analysis period (Eq. 4), and the number of customer phase changes (Eq. 5), combined into a scalar fitness function (Eq. 6). The algorithm is tested in a PandaPower model of a simple radial feeder with 60 residential customers in Tucumán, initially with constant 200 W active and 20 var reactive loads per customer. The available text stops before presenting any post-optimization results or comparisons with the initial network configuration.

Significance. If fully validated, the approach would be practically useful for distribution utilities with smart meters, because the change-count term directly minimizes field work and the formulation is easy to adapt. The objective definitions are coherent and the use of an open-source power-flow tool is a strength. However, the current manuscript provides no quantitative evidence of improvement, and the single-point constant-load test is acknowledged by the authors as only a theoretical test case. The practical significance of the method therefore cannot be assessed from the presented results.

major comments (4)
  1. [Validación del algoritmo] The manuscript does not contain a results section: after defining the test network and the 112233 connection scheme, the text stops before reporting the optimized values of B, PV, and N or the pre-optimization baseline. Consequently, the abstract's claim that the genetic algorithm 'optimally identifies' which loads should be reassigned is not supported by any quantitative outcome in the submitted version.
  2. [Validación del algoritmo] Equations 2 through 6 are evaluated only at a single operating point: every customer draws a constant 200 W with 10% reactive power, a value the authors themselves describe as 'adecuados como caso teórico de prueba.' Real residential loads fluctuate over time, and an optimum at one loading point can worsen imbalance or voltage at other times. The paper should include at least a time-series or multi-scenario test (for example, peak versus off-peak loading) to demonstrate robustness; without this, the operational claim is not established.
  3. [Combinación de objetivos, Eq. (6)] The fitness function depends on hand-chosen normalization constants Bmax=100, PVmax=10, and Nmax=50, and on weights r, t, v that are set a priori. The manuscript provides no sensitivity analysis or empirical rationale tying these values to the Tucumán network data, so any reported solution would be conditional on arbitrary parameter choices. A sensitivity study, or a comparison with a Pareto-front multi-objective formulation, is needed to support the optimality claim.
  4. [Validación del algoritmo] The paper does not compare the genetic algorithm's solution against the original connection scheme or against a simpler baseline such as a greedy phase-balancing heuristic. Because the fitness function is maximized by construction, an improvement in f(S) alone is not evidence of practical benefit; a before/after table reporting B, PV, N and, ideally, neutral current or losses is required.
minor comments (5)
  1. [General] The manuscript contains corrupted accents and replacement glyphs in several places, including the title ('L´ÕQHD') and some equations; a careful proofreading pass is needed.
  2. [Ecuación 4] Equation 4 uses the maximum over t but does not explicitly define the index range n or the meaning of the subscript in 'PV = max{PVt}_n'; the sample interval and period should be stated.
  3. [Validación del algoritmo] The test network parameters list transformer rating, cable type, and geometry, but the conductor impedance values used in the PandaPower model are not provided; these should be included for reproducibility.
  4. [Figuras] Figure 5 is referenced in the text but does not appear in the submitted extract; the authors should ensure that all figures are included and correctly numbered in the final version.
  5. [Referencias] Several references are incomplete (for example, [4] and [15]) and [15] is a TechRxiv preprint; the authors should complete the bibliographic data and, where possible, cite peer-reviewed versions.

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity found: the GA fitness landscape is an objective, not a derived prediction, and the self-citations are not load-bearing.

full rationale

The paper's central contribution is an optimization formulation, not a derived prediction. Equation 6 defines a mono-objective fitness f(S)=1 - (r B/Bmax + t PV/PVmax + v N/Nmax)/(r+t+v), where B, PV, and N are objective metrics computed from the power-flow model and the proposed re-phasing plan (Eqs. 2-5). A genetic algorithm maximizing this fitness necessarily improves those objectives on the modeled scenario; that is the designed behavior of an optimizer, not a circular transfer of inputs into outputs. No quantity is defined in terms of the result it is supposed to predict, no fitted parameter is renamed as a prediction, and no uniqueness theorem or substantive ansatz is imported from the authors' prior work. The two self-citations ([11], [12]) appear as background on phase reassignment and niching genetic algorithms, and the proposed method does not depend on them for its validity. The constant 200 W / 20 var load assumption is explicitly labeled a 'caso teórico de prueba' and is a limitation of the evaluation, not a circularity: it weakens generalizability but does not make the derivation self-referential. Because the paper does not claim to derive external quantities from definitions that already contain the target result, no circular step is exhibited.

Assumptions & free parameters 4 free parameters · 3 assumptions · 0 invented entities

The central claim rests on hand-chosen normalization constants and objective weights, a static-load domain assumption, and the representativeness of the simple test feeder. No new physical entities are introduced.

free parameters (4)
  • Objective weights r, t, v = not specified (assigned a priori)
    Weights control relative importance of imbalance, voltage drop, and number of changes in Eq. 6; chosen by the authors, no sensitivity analysis shown.
  • Bmax = 100
    Hand-chosen normalization bound for the imbalance index in Eq. 6.
  • PVmax = 10
    Hand-chosen normalization bound for voltage drop in Eq. 6.
  • Nmax = 50
    Hand-chosen normalization bound for number of phase changes in Eq. 6.
assumptions (3)
  • domain assumption All customers in the analyzed network have constant active and reactive power (200 W and 10% reactive) over the analyzed period.
    Validation uses static loads; real loads vary over time, so optimal phase assignments may differ.
  • domain assumption A single radial circuit with 60 customers at 10 m spacing and a 250 kVA transformer is representative of real low-voltage networks in Tucumán.
    The paper extrapolates from this simple test network to real operational networks.
  • domain assumption PandaPower's power-flow model accurately represents the behavior of the modeled low-voltage network.
    All simulation results depend on the correctness of the power-flow tool.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Equilibrio de carga para transformadores de distribucion electrica mejorando la calidad de servicio en fin de linea." pith.science (2026). https://pith.science/paper/U5MGA33L

@misc{pith2026250509235,
  author       = {Pith},
  title        = {Pith review of: Equilibrio de carga para transformadores de distribucion electrica mejorando la calidad de servicio en fin de linea},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/U5MGA33L}},
  note         = {Machine review of arXiv:2505.09235}
}
read the original abstract

The distribution of electrical energy faces global challenges, such as increasing demand, the integration of distributed generation, high energy losses, and the need to improve service quality. In particular, load imbalance-where loads are not evenly distributed across the circuit phase-can reduce efficiency, shorten equipment lifespan, and increase susceptibility to service interruptions. While methods that involve shifting loads from one phase to another can be costly, they are effective when smart meters are available and implemented efficiently. This work proposes the use of genetic algorithms to optimally identify which loads should be reassigned in order to improve both phase balance and voltage quality at the end nodes of the network while minimizing the number of required changes. The algorithm was evaluated through simulations using PandaPower, a power flow analysis tool, modeling simple networks based on real-world characteristics of the electrical system in Tucuman.

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

18 extracted references · 17 canonical work pages

  1. [9]

    A practical approach for distribution network load balancing by optimal re-phasing of single phase customers using discrete genetic algorithm

    Omid Homaee, Arsalan Najafi, Mohammad Dehghanian, Mehdi Attar, and Hamid Falaghi. A practical approach for distribution network load balancing by optimal re-phasing of single phase customers using discrete genetic algorithm. International Transactions on Electrical Energy Systems, 29(5):e2834, 2019

  2. [11]

    Phase Reassignment for Load Balance in Low- Voltage Dis- tribution Networks

    Victor A Jimenez, Diego F Lizondo, and Adrian L E Will. Phase Reassignment for Load Balance in Low- Voltage Dis- tribution Networks. International Journal of Electrical Po- wer & Energy Systems , 107691, November 2021

  3. [3]

    Application of the Hurricane-Based Optimization Algorithm to the Phase-Balancing Problem in Three-Phase Asymmetric Networks

    Jose Luis Cruz-Reyes, Sergio Steven Salcedo-Marcelo, and Oscar Danilo Montoya. Application of the Hurricane-Based Optimization Algorithm to the Phase-Balancing Problem in Three-Phase Asymmetric Networks. Computers, 11(3):43, March 2022

  4. [4]

    ‹•‡¿‘• †‡ ‹•–”‹„— ‹‘ † ‘ ƒ„އ Preensamblado, March 2015

    Empresa de Distribucio´n de Electricidad de Tu- cuma´n (EDET). ‹•‡¿‘• †‡ ‹•–”‹„— ‹‘ † ‘ ƒ„އ Preensamblado, March 2015

  5. [5]

    Search and optimiza- tion by metaheuristics

    Ke-Lin Du, MNS Swamy, and others. Search and optimiza- tion by metaheuristics . Springer, 2016

  6. [6]

    Im- pact of Load Unbalance on Low Voltage Network Losses

    J Nuno Fidalgo, Carlos Moreira, and Rafael Cavalheiro. Im- pact of Load Unbalance on Low Voltage Network Losses. In 2019 IEEE Milan PowerTech , pages 1 5. IEEE, 2019

  7. [7]

    Galan and Ole J

    Severino F. Galan and Ole J. Mengshoel. Generalized crow- ding for genetic algorithms. In Proceedings of the 12th an- nual conference on Genetic and evolutionary computation , GECC ïsrÆ ’ƒ‰‡• yyw 782, New York, NY, USA, 2010. ACM

  8. [8]

    An Efficient Method to Optimal Phase Load Balancing in Low Voltage Distribution Network

    Gheorghe Grigoras, Mihai Gavrilas, Bogdan- Constantin Neagu, Ovidiu Ivanov, Ion Tristiu, and Constantin Bulac. An Efficient Method to Optimal Phase Load Balancing in Low Voltage Distribution Network. In 2019 International Con- ference on ENERGY and ENVIRONMENT (CIEM) , pages 323 327, Timisoara, Romania, October 2019. IEEE

Show all 18 references
  1. [10]

    M. Y. Huang, C. S. Chen, C. H. Lin, M. s. Kang, H. J. Chuang, and C. W. Huang. Three-phase balancing of dis- tribution feeders using immune algorithm. IET Generation, Transmission Distribution, 2(3):383 392, May 2008

  2. [12]

    A new data- driven method based on Niching Genetic Algorithms for phase and substation identification

    Victor Adrian Jimenez and Adrian Will. A new data- driven method based on Niching Genetic Algorithms for phase and substation identification. Electric Power Systems Research, 199:107434, 2021

  3. [13]

    N. V. Korovkin, Q. S. Vu, and R. A. Yazenin. A method for minimization of unbalanced mode in three-phase power systems. In 2016 IEEE NW Russia Young Researchers in Electrical and Electronic Engineering Conference (EICon- RusNW) , pages 611 614, February 2016. ISSN: null

  4. [14]

    Review on Power Quality Issues and Its Solution For Better Electrical Reliability In Power System

    Rubi Kumari, Nirban Chakraborty, and Shreya Shree Das. Review on Power Quality Issues and Its Solution For Better Electrical Reliability In Power System. In Proceedings of 1 0 Memorias 12vo. Congreso Nacional de Ingeniería Informática – Sistemas de Información International Co...

  5. [15]

    Review of Dis- tribution Network Phase Unbalance: Scale, Causes, Conse- quences, Solutions, and Future Research Direction

    Kang Ma, Lurui Fang, and Wangwei Kong. Review of Dis- tribution Network Phase Unbalance: Scale, Causes, Conse- quences, Solutions, and Future Research Direction. TechR- xiv, 2020

  6. [16]

    Niching methods for genetic algorithms

    Samir W Mahfoud. Niching methods for genetic algorithms. Urbana, 51(95001):62 94, 1995

  7. [17]

    Swapna Mansani and R. Y. Udaykumay. An optimal pha- se balancing technique for unbalanced three-phase secondary distribution systems. In 2016 IEEE 7th Power India Interna- tional Conference (PIICON) , pages 1 6, Bikaner, Rajasthan, India, November 2016. IEEE

  8. [18]

    Motors and generators, MG-1

    NEMA. Motors and generators, MG-1. National Electri- cal Manufacturers Association , 1998

  9. [19]

    pandapower - an Open Sour- ce Python Tool for Convenient Modeling, Analysis and Op- timization of Electric Power Systems

    Leon Thurner, Alexander Scheidler, Florian Scha¨fer, Jan- Hendrik Menke, Julian Dollichon, Friederike Meier, Steffen Meinecke, and Martin Braun. pandapower - an Open Sour- ce Python Tool for Convenient Modeling, Analysis and Op- timization of Electric Power Systems. IEEE Trans...

  10. [20]

    A PSO Algorithm for Phase Load Balancing in Low Voltage Distribution Networks

    Nicusor Toma, Ovidiu Ivanov, Bogdan Neagu, and Mihai Gavrila. A PSO Algorithm for Phase Load Balancing in Low Voltage Distribution Networks. In International Conference and Exposition on Electrical And Power Engineering (EPE), pages 857 862, Iasi, October 2018. IEEE

Pith tools

Reviewed August 15, 2026 · model on record in the stance chip above.