Redispatch Contracts (RCs) Overview

• Redispatch contracts (RCs) are arrangements that allow TSOs to modify generation or load relative to market baselines for grid security.
• They are used to manage congestion, improve oscillation damping, and enhance transient stability under various network and renewable conditions.
• RCs take multiple forms—from TSO-directed plant-specific dispatch to energy-based contracts—highlighting key operational trade-offs and economic impacts.

Redispatch Contracts (RCs) are arrangements through which a system operator secures the capability to modify generation or consumption relative to a market-based or forecast baseline in order to preserve secure network operation. In the power-systems literature, the term spans several adjacent institutional forms: TSO-directed plant-specific redispatch with compensation, energy-based intraday congestion-management contracts activated after the day-ahead market, and regional redispatch-market access to otherwise curtailed renewable electricity. Across these forms, redispatch is treated as a downstream intervention around an existing market schedule rather than a re-clearing of the market, and its operational value depends on location, direction, timing, uncertainty, and the security criterion being protected (Molodchyk et al., 27 Oct 2025, Titz et al., 2023, Holst et al., 18 Sep 2025, Brandt et al., 25 Jul 2025).

1. Conceptual scope and basic mechanics

In its narrow technical sense, redispatch means changing generator real-power injections while preserving active-power balance. A canonical pairwise implementation is $G_i^+,G_j^-$: increase generator $i$ by $dP_i$, decrease generator $j$ by the same amount, and satisfy

$\sum_{i=1}^m dP_i = 0.$

This formulation treats redispatch as a controlled change of the steady-state operating point, so that line-angle differences, voltages, and ultimately modal or congestion-related system properties change accordingly (Mendoza-Armenta et al., 2015).

A broader operational formulation appears in security-constrained redispatch models. There, a market schedule already exists and satisfies system-wide active-power balance, while the transmission system operator geographically reallocates injections around that schedule to avoid or remove network violations. Conventional generators can be moved up or down, and renewable units can be curtailed if needed. In this framing, redispatch is explicitly not a market re-clearing; it is an ex-post network-feasibility intervention applied to an existing dispatch outcome (Molodchyk et al., 27 Oct 2025).

Institutionally, the literature distinguishes redispatch from neighboring congestion-management instruments. In Germany, redispatch is a TSO-directed adjustment of specific power plants in which the TSO decreases generation upstream of a congestion and increases generation downstream while keeping total generation unchanged; the paper also states that the TSO targets specific plants directly and pays a predetermined compensation price. Countertrading is differentiated as a more market-mediated action in which the TSO procures generation increases and decreases offered in a bidding zone on the intraday market but does not determine which specific plant will be affected (Titz et al., 2023).

The Dutch RC formulation is narrower and more contract-specific. There, an RC is an energy-based congestion-management product activated after the day-ahead market closes. The baseline is the congestion service provider’s submitted prognosis, and redispatch means changing actual consumption or generation relative to that baseline. In the studied case, only downward redispatch is modeled because congestion is caused by EV charging load; the RC therefore provides intraday corrective flexibility to reduce charging relative to prognosis (Holst et al., 18 Sep 2025).

2. Security functions served by RCs

The dominant function of RCs in recent work is congestion management under secure network operation. Deterministic and stochastic redispatch models impose line-flow limits in the intact network and, in the security-constrained case, after any single branch outage. An operating point is therefore unacceptable if it respects only the base case but violates limits under an $(N-1)$ contingency. This makes RCs relevant not merely for economic correction of market schedules, but for preventive security management under credible outages (Molodchyk et al., 27 Oct 2025).

In German system-operation studies, redispatch is presented as a necessary short-term remedy for congestion generated by the interaction of renewable geography, network bottlenecks, and market design. Large quantities of wind generation are concentrated in northern and eastern Germany and offshore, while major demand centers are in the south and west. Because the market-clearing design assumes unlimited transmission within the German-Luxembourg bidding zone, internal north-south bottlenecks are not represented in market clearing, and TSOs must correct the outcome ex post by redispatch or countertrading. The same study links redispatch to the $N\!-\!1$ rule and treats it as part of preventive congestion management rather than as a purely curative action after thermal overloads have materialized (Titz et al., 2023).

A second security function is oscillatory stability. Generator redispatch can improve the damping of poorly damped interarea electromechanical oscillations by changing the steady-state operating point around which the system is linearized. The relevant modes are low-frequency interarea oscillations, typically in the range $0.1$ to $1.0$ Hz, in which groups of generators in one area swing against groups in another. Because redispatch shifts the operating equilibrium, it shifts modal eigenvalues and can improve damping without redesigning controllers such as PSSs; the action is therefore open-loop and operational rather than feedback-based (Mendoza-Armenta et al., 2015, Mendoza-Armenta et al., 2013).

A third security function is transient stability under uncertainty. Preventive redispatch can be optimized before a contingency occurs so that, after the contingency, the system is more likely to remain transiently stable. In this setting, redispatch is judged not only by steady-state feasibility or cost but also by a probabilistic post-fault rotor-angle criterion, represented through the transient stability index

$$TSI_t = \frac{360^\circ-|\Delta \delta_t|_{\max}}{360^\circ+|\Delta \delta_t|_{\max}}, \qquad t\in T.$$

This extends the scope of RCs from congestion relief toward dynamic-security support under uncertain renewable injections (Wang et al., 2024).

3. Institutional forms and contractual architectures

The literature does not use a single institutional template for RCs. Instead, several distinct architectures appear, each tied to a particular system-operator problem.

Context	Contract or product logic	Distinctive features
German redispatch in the study period	TSO-directed plant-specific redispatch	Predetermined compensation price
Dutch RC model	Energy-based intraday contract	Prognosis baseline; activated up to gate closure
German regional redispatch markets	Hourly regional redispatch products	Relief-region specific; upstream of day-ahead

Historically, German redispatch was not organized as a dedicated redispatch services market in the period analyzed by the congestion-prediction study, even though the EU recommended such a market. The same study states that Germany introduced Redispatch 2.0, integrating smaller and renewable units, and that renewable plants had previously been largely outside standard redispatch and instead subject to curtailment as a last resort. This establishes an institutional trajectory from plant-specific operator instruction toward broader participation of assets in redispatch processes (Titz et al., 2023).

The Dutch RC formulation is more explicitly contractual. It is coordinated with Capacity Limitation Contracts (CLCs) in a two-stage stochastic mixed-integer model. A CLC is capacity-based and typically activated before day-ahead market closure; an RC is energy-based, changes consumption or generation relative to prognosis, and can be activated continuously intraday up to gate closure, which is 45 minutes before delivery. In the optimization model, the CLC is the first-stage decision and the RC is the second-stage recourse action. This makes the RC the instrument used after additional information about EV fleet conditions and redispatch market conditions has become available (Holst et al., 18 Sep 2025).

A third institutional form appears in the German regional redispatch markets created under §13k EnWG. These markets make otherwise downward redispatched renewable electricity available as hourly products in regional markets upstream of the day-ahead market. The paper treats this less as a bespoke long-duration bilateral contract than as a new power-purchase option for eligible flexible loads, especially water electrolysers. Redispatch procurement is therefore framed as regional and hourly, with economics that depend sharply on the relief region, annual redispatch availability, and price level (Brandt et al., 25 Jul 2025).

A plausible synthesis is that RCs now cover at least three contractual layers: plant-specific activation rights, recourse-based congestion contracts around a prognosis baseline, and market-access products for locally absorbable redispatch energy. The literature is consistent in one respect: all three remain downstream of market clearing and are activated because the market outcome alone does not guarantee physical security.

4. Technical foundations for contract activation and valuation

The technical value of an RC is not uniform across assets or directions. In modal-stability applications, the value of a redispatch pair is derived from first-order eigenvalue sensitivity. Using the quadratic eigenvalue formulation

$i$0

the redispatch-induced eigenvalue change for a simple mode is

$i$1

The paper then rewrites the sensitivity directly in redispatch coordinates as $i$2, which permits ordered generator-pair ranking. Operationally, the most effective redispatches are those that change line-angle differences across lines with substantial power flow and large mode-shape differences; in the New England 10-machine study, this identified small clusters of highly effective generator pairs and repeatedly singled out $i$3 as critical, though with opposite redispatch direction for different modes (Mendoza-Armenta et al., 2015).

The earlier derivation makes the same point in a more foundational form. For constant voltage magnitudes, the redispatch effect simplifies to

$i$4

This provides a physically interpretable marginal-effectiveness signal: RC value is line-specific, direction-specific, and mode-specific rather than symmetric across buses or generators. A plausible implication is that any RC intended for oscillation damping should be qualified not only by MW volume but also by the sign and location of the admissible redispatch (Mendoza-Armenta et al., 2013).

For congestion-management RCs under uncertainty, the technical object being optimized is often not a single redispatch vector but a policy. In the stochastic $i$5-secure redispatch paper, renewable uncertainty is represented by polynomial chaos expansion (PCE), and the redispatch solution is recovered as random variables encoded by deterministic coefficient vectors. The paper explicitly states that the optimized object is not a single dispatch vector but a recourse policy; from an RC perspective, redispatch capability can therefore be interpreted as a contracted policy mapping forecast errors into upward, downward, and curtailment responses. The same paper enforces chance-constrained generator and line limits and iteratively adds post-outage constraints until no $i$6-critical outage remains (Molodchyk et al., 27 Oct 2025).

In transient-stability-constrained preventive redispatch, the technical value of RC-like flexibility depends on probabilistic improvement in post-contingency stability. The uncertainty-aware objective is

$i$7

subject to probabilistic voltage and generator constraints, uncertain renewable injections, AC power-flow equations, and the TSI-based transient-security requirement. The proposed GD2RL framework then learns the full distribution of post-control $i$8, not only its expectation. This suggests that RC activation criteria can be defined in terms of risk reduction rather than only deterministic feasibility restoration (Wang et al., 2024).

Empirical explainability studies contribute an additional targeting layer. In Germany, a gradient-boosted tree model with SHAP analysis identifies northern wind generation as the dominant driver of redispatch, hydropower in southern Germany as a major mitigator, imports from Denmark as congestion-aggravating, and imports from France as congestion-relieving. The reduced model’s six features are concentrated on the north or south side of the bottleneck, reinforcing the interpretation that redispatch demand is structurally locational. A plausible implication is that recurrent RC procurement should be geographically targeted toward the asset classes and regions repeatedly associated with positive or negative redispatch (Titz et al., 2023).

5. Compensation, pricing, and settlement

Compensation structures differ sharply across RC settings. In the German redispatch study, the TSO directs specific power plants and pays a predetermined compensation price. The paper does not give a full compensation formula, but it directly links redispatch to plant-specific payment rather than to a general market-clearing redispatch service (Titz et al., 2023).

The Dutch RC model is more explicit. The objective minimizes deterministic CLC costs plus expected RC activation costs, and the RC unit cost borne by the system operator is

$i$9

The model contains no separate capacity-reservation payment for RCs; remuneration is activation-based and driven by redispatched energy and market spread. RC volume is bounded below by a minimum redispatch bid size and above by available counter-volume in the redispatch market, so price risk and market-liquidity risk are endogenous determinants of RC use. The principal empirical finding is that combining CLCs and RCs is generally the most cost-effective approach, but the optimal mix depends on fleet size and RC activation timing; for large fleet sizes such as $dP_i$0, the optimal policy limits redispatch because liquidity risk in the immature redispatch market becomes material (Holst et al., 18 Sep 2025).

In the German regional redispatch markets for green hydrogen, pricing is studied parametrically through redispatch electricity prices between $dP_i$1 and $dP_i$2 ct/kWh. The paper treats redispatch access as a non-firm, locational procurement option rather than a classic long-term bilateral supply contract. Its core result is that exclusive redispatch procurement is generally not cost-competitive for relevant electrolyser sizes, while redispatch plus PPAs can materially reduce hydrogen production cost when redispatch prices are low, especially around $dP_i$3 ct/kWh. At higher price levels, the benefit can vanish. This makes redispatch price regulation and locational access central to the commercial value of RC-like participation (Brandt et al., 25 Jul 2025).

Settlement among network operators introduces a different layer of RC-related compensation. A Shapley value-based redispatch cost allocation framework assigns ex-post congestion-management costs to congested lines and can then aggregate them to system operators. Its central result is that the underlying physical model matters decisively: DC OPF and AC OPF can yield different congested lines, different total redispatch costs, and different Shapley allocations. For practical congestion management with Shapley-based cost sharing, the paper concludes that AC OPF solutions are required for fair cost allocation. This is directly relevant wherever RC settlement depends on a modeled attribution of redispatch causation rather than only on realized activation energy (Bauer et al., 2024).

6. Limitations, controversies, and open directions

A recurrent limitation is that much of the RC-relevant literature is operational rather than institutional. The modal-sensitivity papers, the transient-stability redispatch paper, and the stochastic $dP_i$4-secure redispatch paper do not provide contract duration, bidding formats, pay-as-bid versus marginal pricing rules, settlement clauses, procurement auctions, or incentive-compatibility analysis. They provide the technical quantification layer needed for RC qualification, activation, or valuation, but not a complete market design (Mendoza-Armenta et al., 2015, Mendoza-Armenta et al., 2013, Wang et al., 2024, Molodchyk et al., 27 Oct 2025).

Modeling assumptions are also restrictive. The stochastic $dP_i$5-secure redispatch study is DC only; reactive power, voltage limits, losses, and nonlinear AC feasibility are ignored. Uncertainty is limited to renewable injections, demand is deterministic, outages are limited to branches, and the chance constraints use a Chebyshev-based reformulation that is distribution-agnostic but conservative. The transient-stability study relies on an HMPNN surrogate simulator trained offline on a 39-bus system and does not demonstrate scalability to real large-scale systems. These assumptions do not invalidate the methods, but they delimit the direct portability of their RC implications (Molodchyk et al., 27 Oct 2025, Wang et al., 2024).

Institutional evidence remains fragmented. In Germany, the explainable-AI study is explicit that it is not a legal or contractual study; it models aggregate redispatch volume for Germany as a whole, uses a target dataset whose quality is described as “far from optimal,” and cannot observe congested lines directly. The hydrogen study, conversely, focuses on the economics of regional redispatch markets rather than on bespoke RC clauses, and it emphasizes the non-firmness of redispatch electricity, the strong dependence on relief-region location, and the possibility that high redispatch prices can deter participation (Titz et al., 2023, Brandt et al., 25 Jul 2025).

Several controversies follow directly from these limitations. One concerns market design: the German evidence suggests that heavy redispatch volumes are partly a symptom of the single German-Luxembourg bidding zone and the copper-plate assumption, so structural reform such as bidding-zone reconfiguration could reduce reliance on out-of-market redispatch. Another concerns fairness: if ex-post redispatch costs are allocated using simplified network models, cost-sharing rules that satisfy formal axioms may still be operationally disputed because congestion attribution is model-sensitive. A third concerns baseline design: prognosis-based RCs create forecast-risk and gaming concerns, including the baseline dependence highlighted in the Dutch EV study (Titz et al., 2023, Bauer et al., 2024, Holst et al., 18 Sep 2025).

The current literature therefore supports a relatively precise, but still incomplete, encyclopedia definition of RCs. They are not a single standardized product. Rather, they are a family of arrangements by which system operators secure locationally and directionally specific redispatch capability around an existing market schedule, sometimes as plant-specific instructions, sometimes as stochastic recourse policies, and sometimes as regional market-access products. Their technical rationale is well developed: they can relieve congestion, preserve $dP_i$6 security, damp interarea modes, and improve transient stability. Their institutional and economic design remains less settled, with open questions on procurement, pricing, settlement, legal form, and the integration of AC-feasible deliverability, temporal ramping, and market incentives across operating timescales.