{"id":"86414eee-913a-4a52-887e-4b65ee4614cd","arxiv_id":"2505.11027","paper_version":2,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":5.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":3,"one_line_summary":"A two-player game between V2G revenue and battery degradation, solved as a convex potential-function optimization, lets EV users choose how many high-price time slots go to V2G and how many go to battery preservation.","lead":"This paper proposes a game-theoretic charging schedule that lets electric vehicle owners split their parking time between earning money from vehicle-to-grid (V2G) services and protecting battery health. It gives users a single knob, the number of time slots assigned to the V2G player, to dial the trade-off, and reports that this approach is more robust to battery-model uncertainty than standard multi-objective optimization.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Section II.B precomputes battery temperature from a fixed heat term ρ(P_max/V_bat) and never reconciles ρ with the game's hyperparameter w, so the degradation coefficients in (3) ignore the actual optimized power schedule.","rationale":"The reader's weakest_assumption correctly identifies the offline temperature computation as the least secure condition for the central trade-off claim. The game-theoretic construction and the exact potential function (Proposition 1, Appendix B) are mathematically sound for fixed T_b, so the structural framework is not in question. The main risk is quantitative: if T_b is not coupled to the actual optimized power schedule, the degradation costs in (10) and hence the user-facing trade-off curves in Figs. 4–9 are miscalibrated. The paper explicitly promises a role for ρ in Section III.B that never materializes, which strengthens this concern beyond a mere modeling idealization. The proposed fixed-point test would directly measure the magnitude of the bias. Since this concern is already reflected in the reader's CONDITIONAL verdict, no verdict adjustment is needed.","tokens_in":16684,"tokens_out":15316,"duration_ms":157921,"concrete_test":"Re-run the case study of Section IV with a fixed-point iteration: for each w ∈ {0, 6, 12, 18, 24, 30, 36, 42, 48}, solve (12); use the resulting P_bat to compute Q_t = (P_bat,t/(s V_bat))^2 R_int; re-solve (4) to update T_b; update B1,t and B2,t; re-solve (12); iterate until T_b converges. Compare the resulting trade-off curve with Fig. 4. If any point on the curve shifts by more than 10% in either cost axis, the offline decoupling is quantitatively significant and the paper's conclusions about w should be revised.","verdict_should_be":"UNCHANGED","load_bearing_attack":"In Section II.B, the battery temperature T_b is obtained by solving (4) with the heat generation approximated as Î^2 R_int with Î := ρ(P_max/V_bat). The text promises the role of ρ is elaborated in Section III.B, but Section III.B defines w and m and never mentions ρ again. In the subsequent optimization (12), the coefficients B1,t(T_b) and B2,t(T_b) in the degradation model (3) are fixed for all w, even though w determines how many high-price intervals are assigned to the V2G player and thus the actual C-rates. The real heat term is Q_t = I_t^2 R_int with I_t = P_bat,t/(s V_bat), so under high w the average |P_bat,t| is larger, raising T_b and increasing B1,t and B2,t (which enter quadratically/exponentially in Q_cyc_loss,t). The offline decoupling therefore underestimates degradation for high w and overestimates for low w, systematically biasing the trade-off curves in Figs. 4–7 and the long-term projection in Fig. 9. Since the central promise is an accurate user-centric balance, this is a load-bearing modeling gap.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper proposes a user-centric smart charging framework for vehicle-to-grid (V2G) services, formulated as a two-player generalized Nash game. One player maximizes V2G revenue, the other minimizes cyclic battery degradation, and the user selects a hyperparameter w that allocates time intervals between the two players. The authors show that the game is an exact potential game (Proposition 1), so a generalized Nash equilibrium can be obtained by solving the convex program (12). They compare this approach with a multi-objective formulation and an MPC baseline using sensitivity and regret metrics, and they present trade-off studies with respect to ambient temperature, tariff volatility, charger rating, battery capacity, and a one-year projection. The central claims are that the game-theoretic formulation is more robust to degradation-parameter uncertainty and that the parameter w gives EV users a practical, interpretable knob for balancing financial benefit against battery health.","tokens_in":16917,"tokens_out":14489,"duration_ms":140939,"significance":"If the technical issues are resolved, this is a useful contribution to user-centric V2G scheduling. The game-theoretic modeling of the trade-off is novel, the exact-potential characterization in Proposition 1 and Appendix B is sound, and reducing the equilibrium computation to the convex problem (12) is a genuine computational advantage. The paper also provides a broad numerical study that addresses practically relevant factors such as ambient temperature, charger rating, and battery capacity, which is valuable for guiding user decisions. However, the numerical results rest on several modeling approximations and parameter conventions that are not yet fully validated, and the headline robustness claim is partly built into the formulation. The paper would be suitable for publication after these issues are addressed and the numerical implementation is verified.","major_comments":[{"comment":"The battery temperature used in the degradation coefficients is computed offline from a representative current Î = ρ(Pmax/Vbat), but the promised elaboration of ρ in Section III.B never appears; Section III.B defines w and m and does not mention ρ, and Section III.C later reuses ρ as a multi-objective weight in (13). Because B1,t and B2,t in (3) are fixed from this offline Tb profile, the optimized schedule Pbat,t does not feed back into the degradation cost. A high-w solution assigns more high-price intervals to the V2G player and therefore increases the average |Pbat,t| and the heat term Q = I²Rint; fixing Tb from a single ρ will under-estimate degradation for large w and over-estimate it for small w, systematically biasing the trade-off curves in Figs. 4-7 and the one-year projection in Fig. 9. Please couple Tb to the decision variables, or provide a concrete test showing that the degradation results are insensitive to the schedule used for the offline computation, and state explicitly how ρ relates to w.","section":"Section II.B, Eq. (4), and Section III.B"},{"comment":"The headline robustness result is partly built into the game formulation. In the perturbed potential P~(u,ζ), the uncertainty ζ is assumed to affect only the degradation term Qcyc, which appears in P only on the intervals assigned to the BD player; the V2G player's objective terms in (11) have no ζ dependence. Consequently, the V2G player's decision variables are ζ-invariant by construction, so the sensitivity Sgt and regret Rgt in (15)-(16) are mechanically bounded. This does not invalidate the comparison with the multi-objective formulation (13), but the claim that the game is 'significantly more robust' should be qualified as a property of the hard time-slot partition rather than an empirical discovery. To make the comparison informative, the authors should also perturb quantities that affect both approaches symmetrically, such as the price coefficients αt or all degradation coefficients over the entire horizon.","section":"Section IV.B, Eqs. (14)-(16), Fig. 3"},{"comment":"There is a serious parameter/unit inconsistency in the degradation model. With Tb in kelvin as stated, the Table I constants give B1(298 K) = 8.61e-6·298² - 5.13e-3·298 + 0.763 ≈ -0.0011, so the cyclic capacity loss in (3) would be negative at typical operating temperatures, and the quadratic term in P would be concave rather than convex, contradicting the convexity argument used for (12). This is only physically meaningful if the temperature in (2b)-(2c) is in Celsius as in the original reference [16]; please correct the unit statement and re-verify all numerical results. Relatedly, Eq. (3) and Appendix A define Vbat as the terminal voltage of a single cell, while Section IV.A states that the terminal voltage of the battery pack is 350 V; using the pack voltage in (3) changes the computed C-rate by a factor of approximately nseries and would alter the degradation magnitudes in Figs. 4-9. Please confirm which convention was actually used in the code.","section":"Section II.A, Table I, and Section IV.A"},{"comment":"The smooth approximation exp(α|x|) ≈ 1 + α²x²/h with h = 0.0465 is used without an accuracy assessment. The paper does not report the approximation error over the C-rates and temperatures actually encountered in the case studies, e.g., Pbat up to 22 kW with the pack parameters in Section IV.A. Since all degradation costs and resulting trade-off curves are computed with (3), a quantitative validation of the approximation error over the operating envelope should be provided.","section":"Section II.A, Eq. (3), Appendix A"}],"minor_comments":[{"comment":"The symbol ρ denotes a user participation level in the temperature model in Section II.B and a convex-combination weight in (13); using the same symbol for two different quantities is confusing even if the promised elaboration were provided.","section":"Notation (Sections II.B and III.C)"},{"comment":"In the second half of the proof of Proposition 1, 'given x2,y2∈Ω1(u−2)' should read 'x2,y2∈Ω2(u−2)'.","section":"Appendix B"},{"comment":"The perturbation support set is written inconsistently as '[a,a]' and '[a,a]'; likely ̲a and ̄a are intended.","section":"Section III.C"},{"comment":"The one-year projection does not state whether the coefficients B1,t/B2,t and the battery capacity are updated as degradation accumulates; if they are kept constant, that approximation should be stated explicitly and ideally justified.","section":"Section IV.E"},{"comment":"The text should clarify how the discrete sweep over w for the game-theoretic method is aligned with the continuous sweep over ρ for the multi-objective method so that the sensitivity and regret box plots are directly comparable.","section":"Section IV.B, Fig. 3"}],"recommendation":"major_revision","confidential_remarks":"The theoretical game formulation and the exact-potential proof are sound; my main concern is the numerical implementation and parameter consistency. Please ask the authors to verify the unit conventions in Table I and Section IV.A, to reconcile the offline thermal profile with the optimized schedule, and ideally to provide the code or data so that the revised numerical results can be checked. The robustness comparison should also be reframed or extended."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"The paper's real contribution is the horizon-splitting game: the user picks w, the high-price intervals go to a V2G-revenue player, the rest go to a battery-degradation player, and each player optimizes its own objective over disjoint time slots. That is a genuinely new and practical framing, and the resulting GNE is computed by a single convex program with a correct exact-potential argument. The price-based interval assignment rule is sensible, the solve is fast, and the comparative studies (temperature, charger power, battery capacity, plus a one-year projection) are useful. They also report the MPC comparison honestly, including that their method has a slightly higher expected cost.\n\nThe main soft spot is the battery temperature in Section II.B. T_b is precomputed offline from a representative current rho(P_max/V_bat) and then fixed in the degradation model, even though the optimized schedule, which changes with w, determines the actual currents and hence the actual heat. The text says the role of rho is elaborated in Section III.B, but Section III.B never mentions rho again; it defines w and m instead. So the degradation coefficients are fixed while the C-rates vary with w, which biases the trade-off curves: aggressive V2G profiles are modeled at too low a temperature, conservative ones at too high. A couple of fixed-point iterations that couple temperature to the schedule, or a sensitivity study showing the decoupling is benign, would fix this. As published it weakens the quantitative claims, though probably not the qualitative direction of the insights.\n\nThe second soft spot is that the headline robustness advantage is partly structural. In the potential function, perturbing the degradation parameter zeta only moves the BD player's intervals; the V2G part is immune by construction. The multi-objective baseline puts the degradation term across the whole horizon, so it is affected everywhere. The comparison shows the interval split buys robustness, which is the point of the design, but 'significantly more robust' oversells what is essentially a property of the construction. Third, no code or data, and external tariff/thermal inputs are not reproduced. Minor for this venue, but it limits verification.\n\nThe math is correct, the authors are upfront about their modeling assumptions, and the central framework holds up. It is incremental but useful work that deserves a serious referee. The revision ask should be the temperature coupling or justification, plus tempered robustness language.","headline":"A clean, incremental V2G trade-off paper whose horizon-splitting game is a genuinely usable idea; the offline battery temperature shortcut is the one real soft spot.","tokens_in":17442,"tokens_out":4545,"would_cite":true,"duration_ms":43494,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"deepseek-v4-flash","headline":"A two-player game lets EV owners dial in their V2G-versus-battery-wear trade-off.","keywords":["vehicle-to-grid","V2G","battery degradation","generalized Nash equilibrium","exact potential game","smart charging","horizon splitting","convex optimization"],"falsifier":"Simulate the optimized schedule from (12) with the full coupled cabin–battery temperature equations, recompute the cyclic capacity loss with the resulting $T_b$, and compare it with the loss using the paper's precomputed temperature profile; if the difference is large enough to change the ordering of the trade-off curves in the paper's case studies, the central trade-off claim fails.","tokens_in":16398,"feed_emoji":"🔋","tokens_out":9163,"duration_ms":85149,"temperature":0.7,"pith_summary":"This paper puts the vehicle-to-grid versus battery-wear decision in the hands of the EV owner through one integer dial. It models the conflicting goals of earning from V2G discharge and preserving battery health as two self-interested players in a game, then proves the game has an exact potential function, so solving one convex program yields a generalized Nash equilibrium. The user's dial w decides how many of the highest-price time intervals go to the revenue player and how many go to the degradation player, while constraints keep the final charge at the user's target. In the paper's numerical comparisons, this game-theoretic schedule is less sensitive to perturbations of the battery degradation parameters than a weighted multi-objective formulation and an MPC baseline, and it reaches the desired final energy in every tested setting.","feed_headline":"One dial sets an EV's V2G profit versus battery-wear balance","feed_subtitle":"Picking one number w splits intervals by price; one convex program balances revenue against battery aging.","key_machinery":"The load-bearing object is the exact potential function $P(u) := \\sum_{j\\in T^w_m} \\alpha_j P_{\\mathrm{bat},j}\\Delta t + \\gamma \\sum_{j\\in \\tilde T^w_m} Q^{\\mathrm{cyc}}_{\\mathrm{loss},j}(P_{\\mathrm{bat},j})$ from Proposition 1. It makes the two players' selfish improvements point in the same direction as a single common objective, so the game's generalized Nash equilibria coincide with the minimizers of one convex program, (12). Supporting this, the paper replaces the non-smooth exponential cyclic-aging term of the empirical model with the smooth, power-quadratic approximation in (3), which keeps $P$ convex and cheap to solve while binding the degradation cost to the actual charging power.","core_discovery":"The paper's central claim is that a user-centric V2G charging schedule can be computed as the equilibrium of a two-player game, not as a scalarized optimization over both objectives. Player V2G minimizes charging cost on a chosen set of time slots; player BD minimizes cyclic battery-capacity loss on the remaining slots. Proposition 1 shows the game is an exact generalized potential game, with potential function P(u) given in (11), so any global solution of the convex program (12) is a generalized Nash equilibrium. The horizon-splitting rule, assigning the w highest-price intervals to the V2G player and the rest to the degradation player, gives the user a single integer control over the trade-off, and constraint (8) ensures the final energy target is always met within tolerance.","pith_inferences":["If the offline-temperature assumption is nearly exact, the same exact-potential construction should extend to any convex, power-dependent battery wear model, because convexity is the only property the proof of Proposition 1 uses beyond compactness.","A direct test of the robustness claim would be to feed the optimized schedule back into the full coupled temperature dynamics and recompute degradation; until that loop is closed, the numerical robustness results are conditional on the precomputed temperature profile.","The horizon-splitting mechanism is a template for any two-sided scheduling problem with a monetary and a wear objective, and the price-sorting rule could be adapted to other signals such as carbon intensity or grid congestion."],"forward_implications":["An EV owner can choose any degree of V2G participation by setting a single integer $w$, and the resulting schedule still meets the user's desired final energy within tolerance.","Because the equilibrium is computed by a convex program, each 48-interval, 12-hour schedule solves in under 50 ms on a laptop-class processor, which the paper argues makes real-time embedded implementation feasible.","The reported sensitivity and regret comparisons imply that the game-theoretic schedule inherits robustness to uncertainty in the degradation parameters $B_1,B_2$ without needing an additional robustification layer.","The trade-off studies give concrete guidance: colder ambient temperatures and smaller battery capacities make aggressive V2G participation costlier in battery health, while high tariff volatility and higher-rated chargers make participation more financially attractive."],"supporting_citations":[{"why":"It supplies the semi-empirical calendar and cyclic aging equations that the paper smooths into the convex degradation cost.","marker":"[16]"},{"why":"It supplies the coupled cabin–battery temperature dynamics used to precompute the battery temperature profile.","marker":"[11]"},{"why":"It provides the definition and existence conditions of generalized Nash equilibria on which the game formulation rests.","marker":"[25]"},{"why":"It provides the definition of exact generalized potential games used in Proposition 1.","marker":"[26]"},{"why":"It serves as the MPC-based benchmark for the total-cost and regret comparisons.","marker":"[14]"},{"why":"It supplies the V2G tariff profile and dynamic pricing signal used in the numerical studies.","marker":"[27]"},{"why":"It provides the lumped-thermal-network parameters used to solve the battery temperature equations.","marker":"[22]"},{"why":"It provides the experimental capacity-fade data cited in Remark 1 as evidence that the degradation model is reasonably accurate.","marker":"[12]"}],"fun_headline_variants":["V2G game balances EV profit and battery wear","One number tunes EV's V2G profit vs battery health","Two-player game optimizes EV charging for profit and health","Game-theoretic V2G scheduling balances revenue and aging","Single control sets EV's V2G profit vs battery degradation"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The scheme assumes the battery's temperature can be fixed in advance from a representative charging current rather than recomputed from the actual optimized schedule; if the real temperature under that schedule is materially different, the degradation costs and the trade-off curves shift.","fun_headline_variants_meta":{"raw":{"variants":["V2G game balances EV profit and battery wear","One number tunes EV's V2G profit vs battery health","Two-player game optimizes EV charging for profit and health","Game-theoretic V2G scheduling balances revenue and aging","Single control sets EV's V2G profit vs battery degradation"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000165,"raw_usage":{"total_tokens":1137,"prompt_tokens":721,"completion_tokens":416,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":337,"completion_tokens_details":{"reasoning_tokens":334}},"tokens_in":337,"tokens_out":416,"duration_ms":4564,"temperature":1.0,"reasoning_tokens":334,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-15T20:59:22.017735+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Simulate the optimized schedule from (12) with the full coupled cabin–battery temperature equations, recompute the cyclic capacity loss with the resulting $T_b$, and compare it with the loss using the paper's precomputed temperature profile; if the difference is large enough to change the ordering of the trade-off curves in the paper's case studies, the central trade-off claim fails.","supporting_citations":[{"cited_title":"A coordinated model predictive control-based approach for vehicle-to-grid scheduling considering range anxiety and battery degradation,","cited_arxiv_id":null,"evidence_quote":"It serves as the MPC-based benchmark for the total-cost and regret comparisons."},{"cited_title":"Pricing for electric vehicle charging stations based on the responsiveness of demand,","cited_arxiv_id":null,"evidence_quote":"It supplies the V2G tariff profile and dynamic pricing signal used in the numerical studies."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"It supplies the semi-empirical calendar and cyclic aging equations that the paper smooths into the convex degradation cost."},{"cited_title":"Generalized Nash equilibrium problems,","cited_arxiv_id":null,"evidence_quote":"It provides the definition and existence conditions of generalized Nash equilibria on which the game formulation rests."},{"cited_title":"Decomposition algo- rithms for generalized potential games,","cited_arxiv_id":null,"evidence_quote":"It provides the definition of exact generalized potential games used in Proposition 1."},{"cited_title":"Thru-life impacts of driver aggression, climate, cabin thermal management, and battery thermal management on battery electric vehicle utility,","cited_arxiv_id":null,"evidence_quote":"It provides the lumped-thermal-network parameters used to solve the battery temperature equations."},{"cited_title":"Empirical capacity measurements of electric vehicles subject to battery degrada- tion from V2G services,","cited_arxiv_id":null,"evidence_quote":"It provides the experimental capacity-fade data cited in Remark 1 as evidence that the degradation model is reasonably accurate."}],"review_version":1}