pith. sign in

arxiv: 2405.12444 · v3 · submitted 2024-05-21 · 📡 eess.SY · cs.SY

Power-Duration Characterization of Aggregated Thermostatically Controlled Loads via Reach and Hold Sets

Mazen Elsaadany , Hamid R. Ossareh , Mads R. Almassalkhi This is my paper

Pith reviewed 2026-05-24 01:23 UTC · model grok-4.3

classification 📡 eess.SY cs.SY
keywords thermostatically controlled loadsreach-and-hold setsMarkov-chain modelaggregate flexibilitypower-duration characterizationsetpoint controlsecond-order dynamics
0
0 comments X

The pith

A Markov-chain model of TCL dynamics yields a tractable optimization that inner-approximates the reach-and-hold set of an aggregated fleet.

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

The paper shows how to compute an inner approximation of the reach-and-hold set for a population of thermostatically controlled loads by solving an optimization problem built on a Markov-chain representation of their second-order temperature dynamics. This set tells an operator the largest power deviation the fleet can sustain and the longest duration it can hold that deviation after a setpoint change. Because the optimization is tractable, it can be solved ahead of time to produce usable bounds on flexibility. Simulations confirm that the resulting sets correctly bound the power trajectories the fleet can actually follow.

Core claim

The reach-and-hold set of an aggregated TCL fleet is characterized by formulating a tractable optimization problem that computes an inner approximation from a Markov-chain model of second-order TCL dynamics. The model tracks the joint evolution of on/off states and temperature bins, and the optimization finds the largest reachable power interval that can be maintained for a prescribed hold time. Simulation results show that the computed sets accurately describe the fleet's controllable power consumption under setpoint changes, and a robustness study quantifies sensitivity to initial-condition and parameter uncertainty.

What carries the argument

The reach-and-hold set, obtained as the feasible set of an optimization problem posed over the transition matrices of a Markov-chain model that discretizes second-order TCL temperature dynamics.

If this is right

  • Grid operators obtain pre-computed numerical bounds on how much power an AC fleet can deliver or absorb and for how many minutes.
  • The same optimization can be re-solved when the fleet size, outdoor temperature, or heterogeneity parameters change.
  • The inner approximation supplies a conservative but guaranteed-feasible region for real-time dispatch decisions.
  • Uncertainty in initial state or parameter values can be propagated through the same optimization to produce robustified sets.

Where Pith is reading between the lines

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

  • The method could be embedded inside a model-predictive controller that repeatedly solves the optimization over a receding horizon to schedule flexibility.
  • If the Markov-chain bins are made finer, the same optimization structure would produce tighter approximations without changing the overall formulation.
  • The reach-and-hold characterization may extend directly to other populations whose aggregate dynamics admit a similar Markov-chain description, such as electric-vehicle chargers.

Load-bearing premise

The Markov-chain model is assumed to be accurate enough that its inner-approximated reach-and-hold sets remain valid bounds for real aggregate behavior.

What would settle it

Run the same setpoint-change experiment on a detailed nonlinear simulation of the TCL population and check whether every observed power trajectory stays inside the computed inner-approximation set for the claimed hold duration.

Figures

Figures reproduced from arXiv: 2405.12444 by Hamid R. Ossareh, Mads R. Almassalkhi, Mazen Elsaadany.

Figure 1
Figure 1. Figure 1: Outdoor temperature profile with shaded region showing considered DR [PITH_FULL_IMAGE:figures/full_fig_p003_1.png] view at source ↗
Figure 4
Figure 4. Figure 4: Change in aggregate power consumption Pagg during a DR event with corresponding Preach and Thold values. With these notations, we formally define the reach-and-hold set of a TCL fleet as follows: Definition 1 (Reach and Hold Set). Let k0 be the first timestep of the DR event of interest. The reach–and–hold set at timestep k0, denoted by R, is the set of all pairs (Preach,Thold) for which there exists an ad… view at source ↗
Figure 5
Figure 5. Figure 5: Reach-and-hold sets for setpoint changes of [PITH_FULL_IMAGE:figures/full_fig_p007_5.png] view at source ↗
Figure 7
Figure 7. Figure 7: Reach-and-hold sets for a ±1 ◦C setpoint change: (a) with different peak limiting parameters γ, and (b) with different coarse initial-condition estimates. of the TCL fleet. This section therefore evaluates the realized performance of the proposed schedules and then examines the robustness of the resulting reach-and-hold characterization to uncertainty in initial conditions and TCL parameters. A. Open-loop … view at source ↗
Figure 8
Figure 8. Figure 8: True and coarse 2-D initial-condition estimates for the ON and OFF [PITH_FULL_IMAGE:figures/full_fig_p009_8.png] view at source ↗
Figure 9
Figure 9. Figure 9: Realized Preach values from open-loop implementation of α˜ schedule obtained with and without parameter perturbation. The mean of all the parameter perturbation cases is illustrated using the blue square. Table V PERFORMANCE UNDER ETP PARAMETER UNCERTAINTY. Thold εmax(δ) εmax(δ) εmax(δ) εmax(δ) εmax(δ) (hrs) (10-sec) (5-min) (15-min) (60-min) (Thold) 1 22.3% 17.5% 11.7% 4.83% 4.83% 2 17.8% 15.9% 13.7% 6.40… view at source ↗
read the original abstract

Aggregations of thermostatically controlled loads (TCLs), such as air conditioners, offer valuable flexibility to the power grid. The aggregate power consumption of a TCL fleet can be controlled by adjusting thermostat setpoints. An \textit{ex-ante} quantification of the flexibility that results from such setpoint change can inform grid operator decisions. This paper develops a rigorous, yet practical method to quantify flexibility in terms of the `reach-and-hold' set of TCL aggregations, which defines how much power can be shifted (reach) and for how long (hold). To quantify the reach-and-hold set, we employ a Markov-chain-based model of the TCL aggregation that captures second-order TCL dynamics, enabling accurate characterization of reach-and-hold sets. A tractable optimization problem is then formulated to numerically compute an inner approximation of these sets. Simulation results validate that our method accurately characterizes the fleet's flexibility and effectively controls its power consumption. Furthermore, a robustness analysis is carried out to investigate the effects of uncertainty in initial conditions and TCL parameters.

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 paper develops a Markov-chain model of aggregated TCLs that captures second-order dynamics, formulates a tractable optimization problem to compute an inner approximation of the reach-and-hold set (power that can be reached and held for a given duration), and reports simulation validation plus a robustness analysis under initial-condition and parameter uncertainty.

Significance. If the discretization error remains controlled, the method supplies a concrete, numerically tractable characterization of TCL flexibility that grid operators could use for ex-ante scheduling; the explicit simulation validation and robustness checks are concrete strengths that increase practical relevance.

major comments (2)
  1. [§3] §3 (Markov-chain construction): the transition probabilities are obtained by uniform state-space discretization of the second-order TCL dynamics, yet no explicit bound (Hausdorff distance or otherwise) is derived on the distance between the discrete and continuous reach-and-hold sets; because the inner-approximation optimization is solved on the discrete model, this omission directly affects the claimed accuracy of the power-duration pairs.
  2. [§4] §4 (Simulation validation): the text states that simulations “validate that our method accurately characterizes the fleet’s flexibility,” but supplies neither quantitative error metrics (e.g., maximum deviation from a continuous-time reference trajectory), nor the number of Monte-Carlo runs, nor any comparison against the underlying ODE model; without these, the support for the central claim cannot be assessed.
minor comments (2)
  1. [Definition of reach-and-hold set] The notation for the reach-and-hold set (Definition 1 or equivalent) should explicitly state whether the “hold” interval is required to be contiguous or may be interrupted.
  2. [Figures] Figure captions should indicate the discretization granularity (number of temperature and auxiliary-state bins) used to generate each plotted set.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for the constructive comments, which help clarify the presentation of our results. We address each major comment below.

read point-by-point responses

  1. Referee: [§3] §3 (Markov-chain construction): the transition probabilities are obtained by uniform state-space discretization of the second-order TCL dynamics, yet no explicit bound (Hausdorff distance or otherwise) is derived on the distance between the discrete and continuous reach-and-hold sets; because the inner-approximation optimization is solved on the discrete model, this omission directly affects the claimed accuracy of the power-duration pairs.

    Authors: We acknowledge that the manuscript does not derive an explicit bound (e.g., Hausdorff distance) between the discrete and continuous reach-and-hold sets. The Markov chain is obtained by uniform discretization of the continuous second-order dynamics, and the optimization computes an inner approximation relative to this discrete model. The discretization resolution is selected to balance accuracy and tractability, with simulation results used to assess practical fidelity. In revision we will expand §3 to include a discussion of the discretization parameters, their relation to the continuous dynamics, and any available a-priori error estimates from the Markov-chain approximation literature; however, a rigorous Hausdorff bound would require substantial additional analysis that lies outside the present scope. revision: partial

  2. Referee: [§4] §4 (Simulation validation): the text states that simulations “validate that our method accurately characterizes the fleet’s flexibility,” but supplies neither quantitative error metrics (e.g., maximum deviation from a continuous-time reference trajectory), nor the number of Monte-Carlo runs, nor any comparison against the underlying ODE model; without these, the support for the central claim cannot be assessed.

    Authors: The simulations in §4 demonstrate that the Markov-chain-based controller keeps aggregate power inside the computed reach-and-hold sets for the indicated durations. We agree that the current presentation lacks quantitative error metrics, the number of Monte-Carlo realizations, and explicit comparisons to the underlying continuous-time ODE trajectories. In the revised manuscript we will add these elements: maximum deviation statistics between achieved and target power, the number of Monte-Carlo runs performed, and side-by-side trajectory comparisons against the original second-order ODE model to provide stronger quantitative support for the validation claims. revision: yes

Circularity Check

0 steps flagged

No circularity: derivation uses standard Markov-chain discretization and independent optimization formulation

full rationale

The paper presents a Markov-chain model derived from standard state-space discretization techniques for second-order TCL dynamics, then formulates a separate tractable optimization to compute an inner approximation of the reach-and-hold set. No equation reduces a claimed prediction or uniqueness result to a fitted parameter defined by the same data, nor does any load-bearing step rely on a self-citation chain that itself lacks external verification. The abstract and described method treat the Markov model as an input approximation whose accuracy is validated by simulation rather than enforced by construction. This is the common honest case of a self-contained derivation against external benchmarks.

Axiom & Free-Parameter Ledger

0 free parameters · 1 axioms · 0 invented entities

The central claim rests on the domain assumption that a Markov-chain abstraction adequately captures aggregate second-order TCL dynamics and that the resulting optimization yields a practically useful inner approximation; no explicit free parameters or invented entities are named in the abstract.

axioms (1)
  • domain assumption The Markov-chain model captures the essential second-order dynamics of the TCL aggregation.
    Invoked to enable accurate characterization of reach-and-hold sets.

pith-pipeline@v0.9.0 · 5723 in / 1197 out tokens · 27515 ms · 2026-05-24T01:23:06.135897+00:00 · methodology

discussion (0)

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

Lean theorems connected to this paper

Citations machine-checked in the Pith Canon. Every link opens the source theorem in the public Lean library.

What do these tags mean?
matches
The paper's claim is directly supported by a theorem in the formal canon.
supports
The theorem supports part of the paper's argument, but the paper may add assumptions or extra steps.
extends
The paper goes beyond the formal theorem; the theorem is a base layer rather than the whole result.
uses
The paper appears to rely on the theorem as machinery.
contradicts
The paper's claim conflicts with a theorem or certificate in the canon.
unclear
Pith found a possible connection, but the passage is too broad, indirect, or ambiguous to say the theorem truly supports the claim.

Reference graph

Works this paper leans on

26 extracted references · 26 canonical work pages

  1. [1]

    A new definition and research agenda for demand response in the distributed energy resource era

    J. L. Mathieu et al., “A new definition and research agenda for demand response in the distributed energy resource era”,IEEE Transactions on Energy Markets, P olicy and Regulation, pp. 1–15, 2025

    work page 2025

  2. [2]

    2025 California Demand Response Potential Study – Charting California’s Demand Response Future

    P . Alstone et al., “2025 California Demand Response Potential Study – Charting California’s Demand Response Future”, en, LBNL, Tech. Rep. LBNL-2001113, Mar. 2017

    work page 2025

  3. [3]

    Aggregate flexibility of thermostatically controlled loads

    H. Hao et al., “Aggregate flexibility of thermostatically controlled loads”, IEEE Transactions on P ower Systems, vol. 30, no. 1, pp. 189–198, 2014

    work page 2014

  4. [4]

    Aggregate flexibility capacity of tcls with cycling constraints

    A. R. Coffman et al., “Aggregate flexibility capacity of tcls with cycling constraints”,IEEE Transactions on P ower Systems, vol. 38, no. 1, pp. 52–62, 2022

    work page 2022

  5. [5]

    Flexibility estimation and control of thermostatically controlled loads with lock time for regulation service

    P . W ang et al., “Flexibility estimation and control of thermostatically controlled loads with lock time for regulation service”,IEEE Transactions on Smart Grid, vol. 11, no. 4, pp. 3221–3230, 2020

    work page 2020

  6. [6]

    Enhanced sufficient battery model for aggregate flexibility of thermostatically controlled loads considering coupling constraints

    G. W ang et al., “Enhanced sufficient battery model for aggregate flexibility of thermostatically controlled loads considering coupling constraints”,IEEE Transactions on Sustainable Energy, vol. 12, no. 4, pp. 2493–2496, 2021

    work page 2021

  7. [7]

    A geometric approach to aggregate flexibility modeling of thermostatically controlled loads

    L. Zhao et al., “A geometric approach to aggregate flexibility modeling of thermostatically controlled loads”,IEEE Transactions on P ower Systems, vol. 32, no. 6, pp. 4721–4731, 2017

    work page 2017

  8. [8]

    Data-driven demand-side flexibility quantification: Prediction and approximation of flexibility envelopes

    N. Hekmat et al., “Data-driven demand-side flexibility quantification: Prediction and approximation of flexibility envelopes”, in2023 IEEE Belgrade P owerT ech, 2023, pp. 1–6

    work page 2023

  9. [9]

    Experimental implementation of an emission-aware prosumer with online flexibility quantification and provision

    H. Cai and P . Heer, “Experimental implementation of an emission-aware prosumer with online flexibility quantification and provision”,Sustainable Cities and Society, vol. 111, p. 105 531, 2024

    work page 2024

  10. [10]

    Novel quantification method of aggregated energy flexibility based on power-duration curves

    F . Plaum et al., “Novel quantification method of aggregated energy flexibility based on power-duration curves”,IEEE Access, vol. 12, pp. 132 825– 132 837, 2024

    work page 2024

  11. [11]

    Electric load model synthesis by diffusion approximation of a high-order hybrid-state stochastic system

    R. Malhame and C.-Y . Chong, “Electric load model synthesis by diffusion approximation of a high-order hybrid-state stochastic system”,IEEE Transactions on Automatic Control, vol. 30, no. 9, pp. 854–860, 1985

    work page 1985

  12. [12]

    T apping the energy storage potential in electric loads to deliver load following and regulation, with application to wind energy

    D. S. Callaway, “T apping the energy storage potential in electric loads to deliver load following and regulation, with application to wind energy”, Energy Conversion and management, vol. 50, no. 5, pp. 1389–1400, 2009

    work page 2009

  13. [13]

    Modeling and control of aggregate air condi- tioning loads for robust renewable power management

    S. Bashash and H. K. Fathy, “Modeling and control of aggregate air condi- tioning loads for robust renewable power management”,IEEE Transactions on Control Systems T echnology, vol. 21, no. 4, pp. 1318–1327, 2012

    work page 2012

  14. [14]

    Aggregate power control of heterogeneous tcl populations governed by fokker–planck equations

    J. Zheng et al., “ Aggregate power control of heterogeneous tcl populations governed by fokker–planck equations”,IEEE Transactions on Control Systems T echnology, vol. 28, no. 5, pp. 1915–1927, 2020

    work page 1915

  15. [15]

    Randomized customer-sensitivity-aware con- trol of thermostatic loads for better integration of intermittent renewables

    T . Julien and R. P . Malhame, “Randomized customer-sensitivity-aware con- trol of thermostatic loads for better integration of intermittent renewables”, IEEE Transactions on Control Systems T echnology, 2024

    work page 2024

  16. [16]

    Modeling and control of thermostatically controlled loads

    S. N. Backhaus et al., “Modeling and control of thermostatically controlled loads”, Los Alamos National Laboratory (LANL), Jan. 2011

    work page 2011

  17. [17]

    Aggregated modeling and control of air conditioning loads for demand response

    W . Zhang et al., “Aggregated modeling and control of air conditioning loads for demand response”,IEEE Transactions on P ower Systems, vol. 28, no. 4, pp. 4655–4664, 2013

    work page 2013

  18. [18]

    Model predictive control of aggregated heterogeneous second-order thermostatically controlled loads for ancillary services

    M. Liu and Y . Shi, “Model predictive control of aggregated heterogeneous second-order thermostatically controlled loads for ancillary services”,IEEE transactions on power systems, vol. 31, no. 3, pp. 1963–1971, 2015

    work page 1963

  19. [19]

    Building model calibration: V alidation of gridlab- d thermal dynamics modeling

    C. D. Goodman et al., “Building model calibration: V alidation of gridlab- d thermal dynamics modeling”, Pacific Northwest National Laboratory (PNNL), Richland, W A (United States), T ech. Rep., Nov. 2022

    work page 2022

  20. [20]

    Development and validation of a second-order thermal net- work model for residential buildings

    J. W ang et al., “Development and validation of a second-order thermal net- work model for residential buildings”,Applied Energy, vol. 306, p. 118 124, 2022

    work page 2022

  21. [21]

    Dynamic models of house heating based on equivalent thermal parameters

    R. C. Sonderegger, “Dynamic models of house heating based on equivalent thermal parameters.”, Ph.D. dissertation, Princeton University, 1978

    work page 1978

  22. [22]

    Aggregation and control of populations of thermostatically controlled loads by formal abstractions

    S. Esmaeil Zadeh Soudjani and A. Abate, “Aggregation and control of populations of thermostatically controlled loads by formal abstractions”, IEEE Transactions on Control Systems T echnology, vol. 23, no. 3, pp. 975– 990, 2015

    work page 2015

  23. [23]

    State estimation and control of electric loads to manage real-time energy imbalance

    J. L. Mathieu et al., “State estimation and control of electric loads to manage real-time energy imbalance”,IEEE Transactions on P ower Systems, vol. 28, no. 1, pp. 430–440, 2013

    work page 2013

  24. [24]

    Demand response advanced measurement methodology: Analysis of open-source baseline and comparison group methods to enable caiso demand response resource performance evaluation

    J. Glass et al., “Demand response advanced measurement methodology: Analysis of open-source baseline and comparison group methods to enable caiso demand response resource performance evaluation”, California Independent System Operator (CAISO), T ech. Rep., Feb. 2022

    work page 2022

  25. [25]

    69-2680EF—09, Rev

    Resideo T echnologies, Inc.,Th303 programmable thermostat owner’s guide, Document No. 69-2680EF—09, Rev. 05-25, Resideo Technologies, Inc., May 2025

    work page 2025

  26. [26]

    Parameter identifiability and estimation of thermostati- cally controlled loads

    I. M. Granitsas et al., “Parameter identifiability and estimation of thermostati- cally controlled loads”, in2023 IEEE Belgrade P owerT ech, 2023, pp. 01–06

    work page 2023