pith. sign in

arxiv: 2110.02590 · v1 · submitted 2021-10-06 · 🧮 math.OC

A Feasible Reduced Space Method for Real-Time Optimal Power Flow

Fran\c{c}ois Pacaud , Daniel Adrian Maldonado , Sungho Shin , Michel Schanen , Mihai Anitescu This is my paper

Pith reviewed 2026-05-24 12:52 UTC · model grok-4.3

classification 🧮 math.OC
keywords optimal power flowreduced space methodfeasible path algorithmAugmented Lagrangianreal-time optimizationGPU accelerationsecond-order methodpower systems
0
0 comments X

The pith

A reduced-space second-order method solves optimal power flow while staying feasible at every iteration.

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

The paper presents a feasible-path algorithm for the optimal power flow problem that operates directly in the space defined by the power flow equations. It augments an earlier reduced-space approach with second-order derivatives and uses Augmented Lagrangian penalties only for the operational inequality constraints. Because the physical constraints are satisfied exactly at every step, the iterates never leave the feasible set, which matters for real-time settings where a solution must be available on short notice. The authors demonstrate that graphics processing units make the required Hessian computations fast enough for both static and time-varying cases.

Core claim

The algorithm augments the Dommel-Tinney reduced-space formulation with second-order information, enforces the power flow equations exactly by working in the induced manifold, and treats operational limits through Augmented Lagrangian penalty terms; the resulting iterates remain feasible with respect to the original power flow equations throughout the solution process.

What carries the argument

The reduced space induced by the power flow equations, in which only inequality constraints remain and are handled by Augmented Lagrangian penalties.

If this is right

  • The iterates remain feasible with respect to the power flow equations at every step without additional restoration steps.
  • Second-order information can be incorporated while preserving the feasible-path property.
  • Graphics processing units render the reduced Hessian computation practical for both static and real-time optimal power flow instances.
  • The same framework applies to both static and time-varying optimal power flow problems.

Where Pith is reading between the lines

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

  • The approach could be adapted to other equality-constrained problems where the equalities define a manifold that can be eliminated exactly.
  • Maintaining feasibility throughout may reduce the risk of intermediate solutions that would be physically unrealizable in an operating grid.
  • The penalty formulation may allow trading a small amount of constraint violation tolerance for faster per-iteration cost in highly dynamic environments.

Load-bearing premise

The power flow equations can be used to induce a reduced space where operational constraints are handled solely by Augmented Lagrangian penalties while still guaranteeing convergence to a feasible point that satisfies the original problem.

What would settle it

A numerical test case in which an iterate produced by the algorithm violates the power flow equations or the method fails to reach a feasible point within a real-time time budget.

Figures

Figures reproduced from arXiv: 2110.02590 by Daniel Adrian Maldonado, Fran\c{c}ois Pacaud, Michel Schanen, Mihai Anitescu, Sungho Shin.

Figure 1
Figure 1. Figure 1: Solving the static OPF for case1354pegase: the first plot displays the evolution of the primal and the dual infeasibility, and the second plot displays the evolution of the relative infeasibility for the different kind of operational constraints. making it difficult to track the optimal solution. For all time t, the real-time algorithm should update the tracking control w in a time-span ∆t, with ∆t 1mn. 0 … view at source ↗
Figure 2
Figure 2. Figure 2: Evolution of the active load at bus 1 for case1354pegase. At time t = 0, the initial primal-dual solution (w0, y0) is computed using the static OPF algorithm. For all time t, we use our GPU implementation to update the tracking point. This operation involves two expensive operations: (i) computing the reduced Hessian for the new load conditions (p d t , q d t ) (ii) solving the bounded QP problem (23). We … view at source ↗
Figure 3
Figure 3. Figure 3: Tracking the AC-OPF solution of case1354pegase with the RT￾OPF algorithm. The first and the second plots display respectively (i) the operating costs computed respectively with the real-time OPF and with Ipopt. (ii) the convergence of the real-time OPF. The third plot displays for each generator the absolute difference between Ipopt solution and real-time OPF’s setpoint. VI. CONCLUSION We have devised a fe… view at source ↗
read the original abstract

We propose a novel feasible-path algorithm to solve the optimal power flow (OPF) problem for real-time use cases. The method augments the seminal work of Dommel and Tinney with second-order derivatives to work directly in the reduced space induced by the power flow equations. In the reduced space, the optimization problem includes only inequality constraints corresponding to the operational constraints. While the reduced formulation directly enforces the physical constraints, the operational constraints are softly enforced through Augmented Lagrangian penalty terms. In contrast to interior-point algorithms (state-of-the art for solving OPF), our algorithm maintains feasibility at each iteration, which makes it suitable for real-time application. By exploiting accelerator hardware (Graphic Processing Units) to compute the reduced Hessian, we show that the second-order method is numerically tractable and is effective to solve both static and real-time OPF problems.

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 / 2 minor

Summary. The manuscript proposes a feasible-path algorithm for the optimal power flow (OPF) problem aimed at real-time applications. It augments the Dommel-Tinney reduced-space approach with second-order derivatives, directly enforcing power-flow equalities while handling operational inequality constraints exclusively via Augmented Lagrangian penalty terms. The central claims are that the method maintains feasibility with respect to the original problem at every iteration (unlike interior-point methods) and that GPU acceleration renders the reduced-Hessian computation tractable, with numerical evidence of effectiveness on both static and real-time OPF instances.

Significance. If the feasibility-maintenance property and convergence to a point satisfying the original constrained OPF are established, the approach would constitute a useful alternative to prevailing interior-point solvers for real-time OPF, where strict feasibility during iterations is often required. The explicit use of GPU hardware to compute the reduced Hessian is a concrete strength that directly tackles the computational barrier of second-order methods in this domain.

major comments (2)
  1. [Abstract] Abstract (third paragraph) and the description of the reduced formulation: the claim that the algorithm 'maintains feasibility at each iteration' and is therefore suitable for real-time use rests on the premise that Augmented Lagrangian penalties alone suffice to recover exact satisfaction of the operational inequalities at convergence. Standard AL theory requires the penalty parameter to tend to infinity (or an explicit recovery step) to guarantee feasibility for the original problem; no such mechanism, update rule, or convergence theorem addressing this point in the reduced space is indicated.
  2. [Reduced-space formulation] The reduced-space formulation section: the statement that 'the reduced formulation directly enforces the physical constraints' while operational constraints are 'softly enforced' does not address whether the AL subproblems remain well-posed when the power-flow equalities are eliminated exactly, nor does it provide a proof that limit points of the sequence satisfy the original KKT conditions without additional assumptions on the penalty growth.
minor comments (2)
  1. The abstract refers to 'exploiting accelerator hardware (Graphic Processing Units) to compute the reduced Hessian' but does not specify the linear-algebra kernels or data-layout choices used; a brief description of the GPU implementation would improve reproducibility.
  2. Notation for the reduced variables (e.g., the mapping from full to reduced space) is introduced late; defining it explicitly in the problem-statement section would aid readability.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for the thoughtful and detailed review. The comments correctly identify that the manuscript would benefit from greater clarity on the precise sense in which feasibility is maintained and on the convergence properties of the Augmented Lagrangian method in reduced space. We address each major comment below and will revise the manuscript accordingly.

read point-by-point responses

  1. Referee: [Abstract] Abstract (third paragraph) and the description of the reduced formulation: the claim that the algorithm 'maintains feasibility at each iteration' and is therefore suitable for real-time use rests on the premise that Augmented Lagrangian penalties alone suffice to recover exact satisfaction of the operational inequalities at convergence. Standard AL theory requires the penalty parameter to tend to infinity (or an explicit recovery step) to guarantee feasibility for the original problem; no such mechanism, update rule, or convergence theorem addressing this point in the reduced space is indicated.

    Authors: We agree that the abstract and formulation sections would be improved by distinguishing the two types of constraints. The statement that the algorithm 'maintains feasibility at each iteration' refers to exact satisfaction of the power-flow equality constraints, which are eliminated exactly by the reduced-space formulation. The operational inequality constraints are treated by the Augmented Lagrangian and are therefore satisfied only approximately for finite penalty values. In the implementation we increase the penalty parameter whenever the maximum violation exceeds a prescribed tolerance; this is a standard practical update rule. We will revise the abstract to make the distinction explicit and add a short paragraph in the formulation section that recalls the standard Augmented Lagrangian convergence result (limit points satisfy the KKT conditions of the original problem when the penalty tends to infinity and a constraint qualification holds) and notes that the same result applies after reduction because the power-flow equations are satisfied exactly at every point. revision: yes

  2. Referee: [Reduced-space formulation] The reduced-space formulation section: the statement that 'the reduced formulation directly enforces the physical constraints' while operational constraints are 'softly enforced' does not address whether the AL subproblems remain well-posed when the power-flow equalities are eliminated exactly, nor does it provide a proof that limit points of the sequence satisfy the original KKT conditions without additional assumptions on the penalty growth.

    Authors: When the power-flow equations are eliminated exactly, the reduced problem is an equality-constrained-free optimization problem whose only constraints are the operational inequalities. Applying the Augmented Lagrangian to those inequalities yields an unconstrained (or bound-constrained) minimization problem in the space of independent variables. Under the standard assumption that the power-flow Jacobian remains nonsingular (which is verified at each iteration), the reduced objective and its derivatives are well-defined and smooth, so each subproblem is well-posed. Convergence of the overall sequence to a KKT point of the original problem follows from the classical Augmented Lagrangian theory once the penalty parameter is driven to infinity; the reduction does not alter the argument because the equalities are satisfied identically. We will insert a concise remark citing the relevant AL convergence theorem and stating that it carries over directly to the reduced formulation. revision: yes

Circularity Check

0 steps flagged

No significant circularity in derivation chain

full rationale

The paper presents an algorithmic contribution: a feasible-path second-order method in the reduced space induced by power-flow equalities, with operational inequalities handled via Augmented Lagrangian penalties. The central claims (feasibility at each iteration, numerical tractability via GPU Hessian, suitability for real-time OPF) follow from the explicit construction of the reduced formulation and the choice of AL penalties; they do not reduce to any fitted parameter renamed as prediction, self-definitional loop, or load-bearing self-citation. The reference to Dommel and Tinney is external prior work. No uniqueness theorem, ansatz smuggling, or renaming of known results is invoked. The derivation is therefore self-contained as a method description rather than a circular inference.

Axiom & Free-Parameter Ledger

0 free parameters · 1 axioms · 0 invented entities

Review performed on abstract only; no explicit free parameters, axioms, or invented entities can be extracted beyond the implicit domain assumptions of nonlinear optimization and power-flow modeling.

axioms (1)
  • domain assumption Standard assumptions of nonlinear optimization and the power-flow equations hold and permit reduction of the variable space.
    Implicit when the authors state that the reduced formulation directly enforces the physical constraints.

pith-pipeline@v0.9.0 · 5686 in / 1303 out tokens · 33624 ms · 2026-05-24T12:52:18.182071+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

27 extracted references · 27 canonical work pages · 1 internal anchor

  1. [1]

    Critical review of recent advances and further developments needed in AC optimal power flow,

    F. Capitanescu, “Critical review of recent advances and further developments needed in AC optimal power flow,” Electric Power Systems Research, vol. 136, pp. 57–68, Jul. 2016. [Online]. Available: https://doi.org/10.1016/j.epsr.2016.02.008

    work page doi:10.1016/j.epsr.2016.02.008 2016

  2. [2]

    Real-time optimal power flow,

    Y . Tang, K. Dvijotham, and S. Low, “Real-time optimal power flow,” IEEE Transactions on Smart Grid , vol. 8, no. 6, pp. 2963–2973, 2017

    work page 2017

  3. [3]

    An online gradient algorithm for optimal power flow on radial networks,

    L. Gan and S. H. Low, “An online gradient algorithm for optimal power flow on radial networks,” IEEE Journal on Selected Areas in Communications, vol. 34, no. 3, pp. 625–638, 2016

    work page 2016

  4. [4]

    Online optimization in closed loop on the power flow manifold,

    A. Hauswirth, A. Zanardi, S. Bolognani, F. D ¨orfler, and G. Hug, “Online optimization in closed loop on the power flow manifold,” in 2017 IEEE Manchester PowerTech. IEEE, 2017, pp. 1–6

    work page 2017

  5. [5]

    Optimal power flow pursuit,

    E. Dall’Anese and A. Simonetto, “Optimal power flow pursuit,” IEEE Transactions on Smart Grid , vol. 9, no. 2, pp. 942–952, 2016

    work page 2016

  6. [6]

    Optimal Power Flow Solutions,

    H. Dommel and W. Tinney, “Optimal Power Flow Solutions,” IEEE Transactions on Power Apparatus and Systems , vol. PAS-87, no. 10, pp. 1866–1876, Oct. 1968

    work page 1968

  7. [7]

    Real-time nonlinear optimization as a generalized equation,

    V . M. Zavala and M. Anitescu, “Real-time nonlinear optimization as a generalized equation,” SIAM Journal on Control and Optimization , vol. 48, no. 8, pp. 5444–5467, 2010

    work page 2010

  8. [8]

    An augmented lagrangian filter method for real-time embedded optimization,

    N.-Y . Chiang, R. Huang, and V . M. Zavala, “An augmented lagrangian filter method for real-time embedded optimization,” IEEE Transactions on Automatic Control , vol. 62, no. 12, pp. 6110–6121, 2017

    work page 2017

  9. [9]

    Optimal power flow: a bibliographic survey I: Formulations and deterministic methods,

    S. Frank, I. Steponavice, and S. Rebennack, “Optimal power flow: a bibliographic survey I: Formulations and deterministic methods,” Energy Systems , vol. 3, no. 3, pp. 221–258, Sep. 2012. [Online]. Available: http://link.springer.com/10.1007/s12667-012-0056-y

    work page doi:10.1007/s12667-012-0056-y 2012

  10. [10]

    Reduced-space interior point methods in power grid problems,

    J. Kardos, D. Kourounis, and O. Schenk, “Reduced-space interior point methods in power grid problems,” arXiv preprint arXiv:2001.10815 , 2020

    work page arXiv 2001

  11. [11]

    History of optimal power flow and formulations,

    M. B. Cain, R. P. O’neill, A. Castillo et al., “History of optimal power flow and formulations,” Federal Energy Regulatory Commission, vol. 1, pp. 1–36, 2012

    work page 2012

  12. [12]

    Power flow solution by newton’s method,

    W. F. Tinney and C. E. Hart, “Power flow solution by newton’s method,” IEEE Transactions on Power Apparatus and systems , no. 11, pp. 1449– 1460, 1967

    work page 1967

  13. [13]

    The second order adjoint analysis: theory and applications,

    Z. Wang, I. M. Navon, F.-X. Le Dimet, and X. Zou, “The second order adjoint analysis: theory and applications,” Meteorology and atmospheric physics, vol. 50, no. 1, pp. 3–20, 1992

    work page 1992

  14. [14]

    Augmented lagrangians and applications of the proximal point algorithm in convex programming,

    R. T. Rockafellar, “Augmented lagrangians and applications of the proximal point algorithm in convex programming,” Mathematics of operations research, vol. 1, no. 2, pp. 97–116, 1976

    work page 1976

  15. [15]

    Multiplier methods: a survey,

    D. P. Bertsekas, “Multiplier methods: a survey,” Automatica, vol. 12, no. 2, pp. 133–145, 1976

    work page 1976

  16. [16]

    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

  17. [17]

    E. G. Birgin and J. M. Mart ´ınez, Practical augmented Lagrangian methods for constrained optimization . SIAM, 2014

    work page 2014

  18. [18]

    Nocedal and S

    J. Nocedal and S. J. Wright, Numerical optimization , 2nd ed., ser. Springer series in operations research. New York: Springer, 2006, oCLC: ocm68629100

    work page 2006

  19. [19]

    R. A. Horn and C. R. Johnson, Matrix analysis. Cambridge university press, 2012

    work page 2012

  20. [20]

    A. R. Conn, G. Gould, and P. L. Toint, LANCELOT: a Fortran package for large-scale nonlinear optimization (Release A) . Springer Science & Business Media, 2013, vol. 17

    work page 2013

  21. [21]

    A Julia implementa- tion of Algorithm NCL for constrained optimization,

    D. Ma, D. Orban, and M. A. Saunders, “A Julia implementa- tion of Algorithm NCL for constrained optimization,” arXiv preprint arXiv:2101.02164, 2021

    work page arXiv 2021

  22. [22]

    Improving ultimate convergence of an augmented lagrangian method,

    E. G. Birgin and J. M. Mart ´ınez, “Improving ultimate convergence of an augmented lagrangian method,” Optimization Methods and Software, vol. 23, no. 2, pp. 177–195, 2008

    work page 2008

  23. [23]

    Computing a closest bifurcation instability in multidimen- sional parameter space,

    I. Dobson, “Computing a closest bifurcation instability in multidimen- sional parameter space,” Journal of nonlinear science , vol. 3, no. 1, pp. 307–327, 1993

    work page 1993

  24. [24]

    Graph-Based Modeling and Decomposition of Energy Infrastructures

    S. Shin, C. Coffrin, K. Sundar, and V . M. Zavala, “Graph-based modeling and decomposition of energy infrastructures,” arXiv preprint arXiv:2010.02404, 2020

    work page internal anchor Pith review Pith/arXiv arXiv 2010

  25. [25]

    The power grid library for benchmarking ac optimal power flow algorithms,

    S. Babaeinejadsarookolaee, A. Birchfield, R. D. Christie, C. Coffrin, C. DeMarco, R. Diao, M. Ferris, S. Fliscounakis, S. Greene, R. Huang et al., “The power grid library for benchmarking ac optimal power flow algorithms,” arXiv preprint arXiv:1908.02788 , 2019

    work page arXiv 1908

  26. [26]

    Powermodels. jl: An open-source framework for exploring power flow formulations,

    C. Coffrin, R. Bent, K. Sundar, Y . Ng, and M. Lubin, “Powermodels. jl: An open-source framework for exploring power flow formulations,” in 2018 Power Systems Computation Conference (PSCC). IEEE, 2018, pp. 1–8

    work page 2018

  27. [27]

    A real-time optimization with warm-start of multiperiod ac optimal power flows,

    Y . Kim and M. Anitescu, “A real-time optimization with warm-start of multiperiod ac optimal power flows,” Electric Power Systems Research, vol. 189, p. 106721, 2020. 8 The submitted manuscript has been created by UChicago Argonne, LLC, Oper- ator of Argonne National Laboratory (“Argonne”). Argonne, a U.S. Department of Energy Office of Science laboratory, ...

    work page 2020