The Cost of Stranded Capacity: What Poor BMS Design Costs BESS Asset Owners
BESS financial models assume throughput follows a clean degradation curve. Real fleets don’t behave that way. Stranded capacity is the gap between the two.
Every year, some capacity goes stranded. That’s capacity the pack technically still has, but the BMS can’t safely deliver it. Stranded capacity never shows up in a standard ROI model.
Our guide to calculating BESS ROI covers the standard model: capex, opex, revenue streams, payback period. This piece covers a value-at-risk category that model doesn’t capture.
Quick Answer
Stranded capacity is battery capacity a BESS has but can’t deliver, due to SOC drift, cell imbalance, or overly conservative cutoffs. It’s a hidden cause of revenue below projections. Fixing the underlying BMS design can recover real throughput — up to 10% in mid-to-late project life, in one internal analysis.
Key Takeaways
- Stranded capacity is real capacity the battery holds but the BMS can’t safely deliver. It’s a design and control issue, not a warranty defect or a safety problem by itself.
- One internal DCIR-adaptive cutoff analysis found up to 10% of effective throughput can be recovered in mid-to-late project life, just by replacing a fixed cutoff voltage with a DCIR-adaptive one.
- Cell imbalance alone can strand real energy. A pack stops charging or discharging when its weakest cell hits a limit, well before the rest of the pack is full or empty.
- SOC estimation drift compounds silently. Left uncorrected, it produces wrong dispatch decisions and wrong revenue forecasts, not just an inaccurate percentage on a screen.
- None of these issues show up in a standard BESS ROI model’s assumptions. Asset owners and EPCs who ask about them at the RFP or commissioning stage catch value a generic warranty review misses.
Why Stranded Capacity Doesn’t Show Up in a Standard ROI Model
This is a different problem from capacity stranded by deliberate oversizing. See our BESS oversizing guide for that scenario, where idle nameplate capacity is a sizing choice, not a BMS malfunction. The stranded capacity covered here comes from BMS design and control issues on capacity the system was never meant to leave unused.
A standard BESS ROI model tracks one degradation number: State of Health. It assumes capacity fades on a smooth, predictable curve, and revenue scales down with it.
That model misses a second category entirely. Stranded capacity isn’t capacity the battery has lost. It’s capacity the battery still has, that the BMS can’t reach.
Our own DCIR-adaptive cutoff design analysis found this gap directly. A fixed low-voltage cutoff is sized conservatively for a fresh pack.
It stops discharge earlier and earlier as internal resistance rises with age.
A liquid-cooled BESS cycle-life comparison echoes the same finding. A DCIR-adaptive cutoff can recover up to 10% of effective throughput in mid-to-late project life.
That’s real revenue a fixed-cutoff design leaves on the table, every single cycle.
What Stranded Capacity Actually Looks Like
The mechanism is well documented in industry practice. Trade coverage of battery imbalance in BESS describes stranded energy this way: capacity that stays inaccessible because a small number of cells reach their limits first.
Charging can stop even though most of the pack still has room. Discharge can end while real energy still sits in the stronger cells.
Neither event trips an alarm. Neither shows up as a fault. The pack just quietly delivers less than it should, cycle after cycle.
Performance guarantees compound the problem. Industry analysis of BESS performance guarantees notes that a generic throughput guarantee often ignores which degradation mechanism actually applies.
Frequency regulation stresses a pack differently than energy shifting does.
A guarantee calibrated for the wrong use case can look satisfied on paper while stranded capacity quietly erodes real revenue underneath it.
Four Root Causes That Strand Throughput
Stranded capacity has a small number of well-understood causes. Each one has a specific design fix, not just a monitoring dashboard.
SOC Estimation Drift
A BMS that only counts coulombs accumulates error every cycle, with no way to self-correct. Dispatch decisions built on a drifted SOC number are wrong, even when the battery itself is healthy.
This kind of drift is easy to miss because it doesn’t trigger any alarm. The dispatch software keeps making decisions confidently, just on the wrong number.
Over months, the gap between estimated and true state of charge can grow large enough to strand real capacity at both ends of the cycle — charge or discharge stopping based on a number that no longer matches reality.
See our EKF SOC estimation design guide for how a Kalman-filter-based estimator corrects for this instead of just counting and hoping.
In-Service Cell Imbalance
Cells drift apart in charge level over months of real-world cycling, even when they started out closely matched. The weakest cell then governs the whole string’s usable window.
Imbalance rarely announces itself either. Two cells built to the same spec can still diverge under real thermal and manufacturing variation, cycle after cycle.
Left uncorrected, that gap widens on its own. More of the pack’s true capacity quietly becomes capacity the BMS won’t touch.
See our in-service cell imbalance guide for how to detect and correct drift before it strands real capacity.
Premature Low-SOC Cutoffs
A fixed cutoff voltage, set conservatively for a brand-new pack, gets more conservative every year as internal resistance climbs. It ends discharge earlier than the chemistry actually requires.
The effect compounds quietly with age. A cutoff that was appropriately conservative in year one becomes needlessly conservative by year five, since resistance keeps climbing while the cutoff voltage stays fixed.
Nobody adjusts it, because nothing about the system looks broken. It just delivers a little less every year than it safely could.
See our DCIR-adaptive cutoff design guide for the design behind the 10% recovery figure cited above.
Fragmented BMS and EMS State
SOC, SOH, imbalance data, and cutoff logic often live in separate modules that don’t share state cleanly. Each module compensates conservatively for what it doesn’t know, and those safety margins stack up.
Each layer, acting alone, makes a reasonable decision. Stacked together, those decisions compound into capacity nobody intended to strand.
See our integrated BMS control architecture guide for how a shared-state design removes the guesswork each layer would otherwise carry on its own.
Root Causes, Symptoms, and Fixes at a Glance
The table below lines up all four causes of stranded capacity covered here, alongside how each one shows up and the design fix that addresses it.
| Root Cause | How It Shows Up | Design Fix |
| SOC estimation drift | Wrong dispatch decisions; inaccurate revenue forecasts | Kalman-filter-based SOC estimation with periodic recalibration |
| In-service cell imbalance | String stops early on charge or discharge | Active balancing tuned to LFP’s flat voltage curve |
| Premature low-SOC cutoffs | Discharge ends before chemistry limits are reached | DCIR-adaptive cutoff voltage, not a fixed one |
| Fragmented BMS/EMS state | Conservative margins stack across separate modules | Integrated, shared-state control architecture |
Translating Stranded Capacity Into Revenue Terms
A simple illustration makes the scale concrete. Consider a 10 MWh system earning a blended $80 per MWh-cycle across arbitrage and grid services, cycling roughly 300 times a year.
That works out to about $240,000 in annual gross revenue at full throughput.
If stranded capacity quietly removes 5% of usable throughput, roughly $12,000 in potential annual revenue never gets captured. Not because the battery lacks the energy — because the BMS won’t release it.
At the higher end of the up-to-10% figure cited earlier, that figure roughly doubles to about $24,000 a year.
These are illustrative figures, not a forecast for any specific project. Actual revenue per cycle and stranded-capacity percentage vary by market, duty cycle, and system design.
But the exercise makes the point: stranded capacity deserves the same scrutiny during procurement as round-trip efficiency or a cycle-life warranty.
What This Means for Asset Owners and EPCs
Most procurement processes don’t ask about any of this directly. A typical RFP asks for round-trip efficiency, cycle-life warranty, and a capacity fade curve — all useful numbers, but none of them capture stranded capacity.
A vendor can meet every number on that list and still ship a BMS that strands a meaningful share of usable throughput.
None of these four causes show up in a standard capex/opex ROI model. They also rarely show up in a standard commissioning checklist.
A pack can pass acceptance testing and still strand capacity over its operating life.
The practical fix isn’t a new financial model. It’s a short set of design questions asked before contracts are signed, not after a project underperforms.
- Does the BMS use a fixed or DCIR-adaptive cutoff voltage, and how was that threshold validated?
- Does the SOC estimator include a correction mechanism, or does it rely on Coulomb counting alone?
- How does the BMS report cell imbalance, and what balancing current does it actually deliver?
- Do the BMS and EMS share state directly, or does each layer apply its own separate safety margin?
None of these questions require a vendor to reveal proprietary algorithm details. They just require a vendor who can explain, in plain terms, how the BMS handles SOC estimation, imbalance, and cutoff voltage as the pack ages — not just what it does on day one.
Asking these questions at the RFP or commissioning stage costs nothing. Discovering the answers three years into operation, in a shortfall against a revenue forecast, costs real money.
Sunlith Energy provides technical consultancy for BESS specification, BMS design review, and lifecycle modeling. Contact us to discuss where your project’s design may be leaving throughput stranded.
Frequently Asked Questions
Is stranded capacity the same as normal battery degradation?
No. Degradation is a real, permanent loss of capacity over time, tracked by State of Health.
Stranded capacity is different. It’s capacity the battery still physically has, that the BMS simply can’t reach.
Can stranded capacity void a BESS warranty?
Not directly. Warranties typically cover capacity retention against a defined degradation curve, not the BMS’s ability to access all available capacity. But a system that strands capacity may also look like it’s underperforming its warranty, which is worth raising with the integrator.
How much revenue does stranded capacity actually cost?
It depends heavily on the system, duty cycle, and root cause involved.
One internal analysis found up to 10% of effective throughput recoverable from a single fix: replacing a fixed cutoff voltage with a DCIR-adaptive one, in mid-to-late project life.
Can monitoring software alone fix stranded capacity?
Monitoring can reveal that capacity is being stranded, but it can’t fix the underlying cause on its own. SOC drift, cell imbalance, and conservative cutoffs are BMS design issues, not dashboard issues — they need to be corrected at the control-algorithm level.
Further Reading
For more on the design issues behind stranded capacity, and where this piece differs from adjacent site coverage:
BESS Oversizing: Pros, Cons & the Right-Sizing Strategy
Designing an LFP BESS Against SOC Drift, Cell Imbalance, and Premature Cutoffs
EKF SOC Estimation Design for LFP BESS
In-Service Cell Imbalance in LFP BESS
DCIR-Adaptive Cutoff Design for LFP BESS
Integrated BMS Control Architecture
The Economics of BESS: A Practical Guide to Calculating ROI
Sources
- Adding Efficiency and Better Performance to Battery Energy Storage Systems. Electrical Contractor Magazine, 2026
- Why BESS performance guarantees are more complex than they seem. Energy-Storage.News, 2026
Mixed-Age String Design: Managing a BESS String After Augmentation Adds New Cells
Mixed-age string design is the problem every DC-shuffled augmentation eventually runs into. New modules and years-old modules end up sharing hardware.
The BMS then has to make that mix actually work. Our BESS augmentation guide and our AC block addition vs. DC shuffling comparison cover whether to augment and whether to add capacity on the AC or DC side.
This article picks up after that decision. Once new and old modules are physically on site together, mixed-age string design becomes the question that matters.
It’s narrower and more technical: how do you run a string that mixes cells at very different points in their degradation curve? And how do you do it without dragging new capacity down to old capacity’s level?
What is Mixed-Age String Design in BESS?
Mixed-age string design means managing a BESS string where old and new modules share hardware after augmentation. A series string’s weakest module caps its capacity, so mixing ages often pulls new capacity down. Segregating old and new modules into separate strings usually preserves more capacity than mixing them.
Key Takeaways for BESS Mixed-Age String Design
- Weakest-Link Capacity Cap: A series string’s total usable capacity is limited by its most degraded module, preventing newly added modules from delivering full rated power.
- Impedance & Degradation Mismatch: Resistance differences between old and new cells create uneven electrical and thermal stress, accelerating divergent degradation under load.
- Module Segregation Strategy: Grouping modules into separate strings by State of Health (SOH) preserves significantly more capacity than mixing old and new cells within a single string.
- Joint Constraint Sorting: Multi-variable assembly algorithms (matching capacity, impedance, and self-discharge) reduce cell mismatch by 76–87% compared to single-metric sorting.
- String-Level BMS Control: BMS and EMS logic must manage cutoffs and dispatch priorities at the string level rather than the whole-pack level to protect new capacity.
Why Mixed-Age String Design Is a Real Engineering Problem
Augmentation sounds simple on paper. Add new modules, restore lost capacity, move on.
The DC-shuffling path makes mixed-age string design harder than it sounds. New modules added this way often end up wired into the same series string as modules that have already lost real capacity.
Years of cycling and calendar aging cause that loss. However, it doesn’t happen evenly across a fleet.
In a series-connected string, the weakest module caps total usable capacity. The string can’t deliver more than its most degraded member allows before hitting a cutoff. Peer-reviewed research on active balancing for serially connected battery packs confirms this constraint directly: without balancing, the lowest-capacity cell in a series string limits the capacity the whole string can deliver.
A brand-new module wired into an old string doesn’t contribute its full rated capacity to the pack. It gets pulled down to whatever the weakest module in that same string can still deliver.
Mixed-age string design is the discipline of avoiding that outcome. It works through physical string assembly, BMS cutoff logic, or both.
The Weakest-Link Problem: How New Capacity Gets Dragged Down
The mechanism is straightforward once you see it. Cells or modules in series all carry the same current. Voltage, not current, is what varies between them.
As the string discharges, the most degraded module reaches its low-voltage cutoff first. It has the least remaining capacity. The BMS has to stop the whole string there.
That happens regardless of how much charge the newer modules still have left. The same happens in reverse on charge.
The weakest module hits its upper voltage limit first. The whole string stops charging there too, even if the new modules could still accept more.
Impedance mismatch compounds the problem. A 2014 study on parallel-connected lithium-ion cells found that a 20% internal-resistance difference between two cells led to roughly a 40% reduction in cycle life.
While parallel sets suffer from uneven current sharing, this mechanism directly impacts series strings as well: higher-impedance modules in series generate greater localized heat under load, creating thermal hotspots that accelerate degradation across the entire string.
Separate experimental work on parallel-connected cells under thermal gradients found a related pattern: a 30°C thermal gradient across a pack roughly doubled the degradation rate and produced 50% more capacity loss after 1,000 cycles.
Both results are from parallel-connected cells specifically, not a series-string augmentation case. But the underlying mechanism generalizes.
Impedance mismatch between old and new modules creates uneven electrical and thermal stress. That stress tends to widen the gap between them, not close it.
BMS and EMS Strategies for Mixed-Age String Design
Matching Modules Before They Go Into a String
The first lever is physical: which modules actually get grouped into which string.
A 2026 optimization study on assembling heterogeneous battery packs from repurposed cell inventories tackled this directly. The study targeted second-life EV packs, not BESS augmentation, but used the same string-assembly logic.
It used a mixed-integer linear program to jointly minimize capacity spread, internal resistance spread, and self-discharge spread across a string, rather than sorting by one metric alone.
That joint approach cut mismatch by 76-87% compared to simple single-metric sorting — grouping modules by capacity alone, or resistance alone, and hoping the rest lines up.
The practical takeaway for mixed-age string design: sorting new and old modules by one number, like nameplate capacity, leaves real mismatch on the table. Impedance and self-discharge rate matter just as much.
Segregating Strings by Health Instead of Mixing Them
The second lever is architectural: don’t mix ages within one string at all.
Research on heterogeneous reconfigurable battery systems notes that prior SoH-aware reconfiguration studies show grouping cells of similar health within the same series string improves delivered pack capacity.
The logic holds because a string’s usable capacity is still constrained by its weakest member either way. Grouping by health just keeps that weakest member from dragging down modules that don’t belong with it.
Applied to mixed-age string design, this points toward keeping new modules in their own dedicated strings. Therefore, they should remain separate from the legacy fleet, wherever the site’s busbar and rack layout allow it.
Each string can then run its own SOH-appropriate cutoffs. The new string isn’t held back by the old one. The old string isn’t pushed harder trying to keep pace with modules it can no longer match.
Mixed-Age String Design: Setting Cutoffs and Dispatch at the String Level
Whichever physical layout a site ends up with, the BMS and EMS configuration has to follow it.
A single pack-wide cutoff voltage, sized for the newest modules, risks over-discharging or over-charging the old ones.
A cutoff sized for the oldest modules protects them. But it wastes real capacity sitting unused in the new modules every single cycle.
Per-string or per-block SOH tracking, feeding into a shared control layer, avoids that trade-off. It’s the same shared-state pattern covered in our integrated BMS control architecture guide — mixed-age string design is one more reason that architecture earns its complexity.
On the EMS side, operators can also set dispatch priority deliberately. Favoring the newer string during high-stress events — fast frequency response, high-C-rate calls — protects its cycle life.
Reserving the older string for gentler duty does the same in reverse, instead of averaging stress evenly across mismatched hardware.
Mixed-Age String Design: Segregated Strings vs. Mixed Strings
Neither approach is free. Segregation asks more of the site’s electrical layout; mixing asks more of the BMS.
| Factor | Segregated Strings (old and new separate) | Mixed Strings (old and new combined) |
| Delivered new-module capacity | Close to full rated capacity | Reduced — capped by the weakest module in the shared string |
| BMS/EMS complexity | Lower per string; needs per-string SOH tracking at the control layer | Higher — cutoffs must reconcile two very different degradation states in real time |
| Physical/electrical requirement | Needs spare busbar/breaker capacity for a genuinely separate string | Fits within existing string wiring, easier where space is tight |
| Best fit | Sites with layout headroom, or where preserving new capacity matters most | Sites with hard busbar/space constraints and modest capacity mismatch |
This decision sits downstream of the AC-vs-DC augmentation choice covered in our AC block addition vs. DC shuffling guide. AC block addition sidesteps mixed-age string design entirely, since the new block runs on its own inverters.
It never shares a string with the old fleet. DC shuffling is where this problem actually shows up.
Frequently Asked Questions
Does mixed-age string design apply to AC block addition too?
Not directly. AC block addition installs a new, self-contained power block with its own inverters. The new modules never share a DC string with the old fleet.
Mixed-age string design is specifically a DC-shuffling problem, since that path reuses the existing DC bus and wiring.
How much capacity is actually lost by mixing old and new modules in one string?
It depends on how degraded the old modules are relative to the new ones. A series string’s weakest module caps its usable capacity, so the loss scales with that gap.
A small gap between oldest and newest modules costs little. A large gap can waste a meaningful share of the new capacity added.
Can a BMS correct for mixed-age imbalance with active balancing alone?
Active balancing helps with cell-level SOC drift within a string. It doesn’t remove the underlying capacity or impedance gap between old and new modules.
Balancing hardware and string-level design address different parts of the same problem — see our active balancing topologies guide for how the hardware itself works.
Is segregating strings always the better choice?
Not always. It preserves more new capacity, but it needs spare busbar and breaker headroom the existing site layout may not have.
Where space is tight, a mixed string with carefully matched modules and string-appropriate cutoffs can still be the more practical choice.
Further Reading
BESS Augmentation: The Complete Guide to Restoring Capacity Lost to Degradation
BESS Augmentation: AC Block Addition vs. DC Shuffling
In-Service Cell Imbalance in LFP BESS
Integrated BMS Control Architecture
Active Balancing Hardware Topologies Compared
Sources
- Optimal Assembly of Repurposed Lithium-Ion Battery Packs under Cell Heterogeneity and Screening Uncertainty. arXiv:2607.12951 (2026)
- Target-Mean State-of-Charge Control for Maximum Utilization of Heterogeneous Reconfigurable Battery Systems Under Constant-Bus Constraints. Batteries 2026, 12, 221. MDPI (peer-reviewed, open access)
- Degradation in parallel-connected lithium-ion battery packs under thermal gradients. Communications Engineering (Nature), 2024
- Gogoana, R.; Pinson, M.B.; Bazant, M.Z.; Sarma, S.E. Internal resistance matching for parallel-connected lithium-ion cells and impacts on battery pack cycle life. Journal of Power Sources 252 (2014), 8-13
- A novel active cell balancing topology for serially connected Li-ion cells in the battery pack for electric vehicle applications. PMC (peer-reviewed, open access)
Active Balancing Topologies Compared: Which Circuit Fits Your BESS?
Every BMS spec sheet lists “active balancing” as a feature, but few explain which circuit is doing the work. Active balancing topologies vary widely in cost, speed, and design.
Active balancing topologies compared side by side reveal real trade-offs. Cost, speed, and how the pack gets built all differ by circuit.
Switched-capacitor, switched-inductor, transformer-based, and DC-DC converter circuits all move charge between cells. They are not interchangeable.
Each one changes the balancing current you get and the board space it needs. Each one also scales differently as your string grows.
| Quick Answer Active balancing topologies compared: switched-capacitor and switched-inductor circuits are cheapest but only balance adjacent cells. Transformer-based (flyback) circuits balance any cell directly but cost more and add magnetic complexity. Bidirectional DC-DC converter circuits offer the best flexibility and efficiency at the highest component count. Most BESS packs use switched-inductor or converter-based designs. |
Key Takeaways
- Active balancing topologies fall into four hardware families: capacitor-based, inductor-based, transformer-based, and DC-DC converter-based. Each is defined by the element that temporarily stores energy during transfer.
- Switched-capacitor and switched-inductor circuits are the cheapest and simplest. Both are generally limited to adjacent-cell balancing, which slows equalization across a long string.
- Transformer-based (flyback) circuits can move charge between any two cells directly, improving equalization speed. They need bulky isolation transformers and usually aren’t bidirectional.
- A 2025 peer-reviewed prototype of a switched-inductor BMS balanced 22 series cells. It measured 84% energy transfer efficiency and a 908 mA balancing current, cutting a 1.18V pack imbalance to 0.47V in under 2.5 hours.
- For LFP BESS strings, the practical choice is usually switched-inductor circuits for cost-sensitive designs, or bidirectional DC-DC converter circuits where balancing speed matters more than component cost.
Why Active Balancing Topologies Matter for BESS Design
Passive balancing burns off excess charge as heat through a resistor. It’s cheap and simple, but it wastes energy and only ever removes charge — it can’t move it anywhere.
Active balancing takes a different approach. It transfers energy from higher-charge cells to lower-charge ones instead of dissipating it. Ideally, almost none of the pack’s total energy is lost in the process.
That difference matters more in a BESS than in a phone or laptop pack. A BESS string has hundreds of cells cycling daily for a decade or more. Manufacturing tolerances, thermal gradients across a rack, and uneven aging all pull cells apart in state of charge over time.
Active balancing topologies compared on paper all claim to solve this. In practice, the circuit topology decides more than the “active” label alone.
It sets how fast a pack re-balances, what it costs per cell, and whether the design scales to a 200+ cell string.
Switched-Capacitor Balancing: How It Works and Where It Fits
A switched-capacitor circuit places a capacitor between two adjacent cells, along with a set of switches. The switches alternately connect the capacitor across the higher-voltage cell, then the lower-voltage cell.
Each switching cycle moves a small packet of charge between the two.
This is the simplest active topology to build. It needs no inductor or transformer, uses relatively few components, and keeps voltage stress on the switches and capacitor low.
Among active balancing topologies, this is the simplest to build. The tradeoff is scope: a basic clocked switched-capacitor circuit only transfers energy between neighboring cells.
To move charge from one end of a long string to the other, it has to hop cell-by-cell. That’s slow, and it compounds switching losses at every hop.
Research on this topology also notes it works best when the voltage gap between cells is meaningful.
LFP’s voltage curve stays unusually flat across most of the state-of-charge range. A small SOC gap barely shows up as a voltage difference there, which limits how well a capacitor-based circuit can detect real imbalance.
Switched-Inductor Balancing: How It Works and Where It Fits
A switched-inductor circuit works similarly to a switched-capacitor one. It stores energy in an inductor’s magnetic field instead of a capacitor’s electric field.
Two switches alternate. One path pulls current from the higher-voltage cell into the inductor; the other pushes that stored energy into the lower-voltage cell.
A 2025 peer-reviewed study built and tested a switched-inductor BMS designed to balance 22 series-connected cells per submodule. The full-scale prototype measured an 84% energy transfer efficiency between adjacent cells and a 908 mA average balancing current.
On a real pack, the same prototype cut an initial 1.18V voltage difference down to 0.47V in about 2 hours and 30 minutes. The equivalent SOC gap fell from 91.2% to 49.4% over that window.
Switched-inductor is one of the more common active balancing topologies in EV and BESS designs. It shares the same adjacent-cell limitation as switched-capacitor designs, but inductors tolerate higher currents and voltage differentials better.
That’s part of why this topology shows up often in EV and stationary BMS designs needing faster balancing than a capacitor-only circuit delivers.
One caution applies here: a generic multi-chemistry BMS platform often runs conservative balancing current and thresholds that don’t suit LFP’s flat voltage curve well.
The circuit topology only pays off if the balancing algorithm triggers it at the right SOC threshold.
Transformer-Based Balancing: Flyback and Multi-Winding Topologies
Transformer-based circuits are the active balancing topologies best suited to non-adjacent cell balancing. They use a transformer, rather than a single capacitor or inductor, as the energy-storage element.
In the common flyback arrangement, energy from a higher-voltage cell is stored in the transformer core. It’s then released to a specific lower-voltage cell chosen by the control circuit.
The real advantage here is reach. The transformer can route energy to any winding, so these circuits can balance non-adjacent cells directly — cell 1 to cell 20, for example — without hopping through every cell in between.
That comes at a real cost. Isolation transformers are physically bulky compared to a capacitor or inductor, which matters in a dense BESS rack.
Multi-winding designs, where one primary winding serves many secondary cell taps, also demand tight parameter matching across every winding. That gets harder as cell count grows.
Flyback-based circuits also typically aren’t bidirectional the way DC-DC converter designs are. Energy generally flows one direction per switching cycle — source cell into the transformer, out to the target cell.
Despite the added cost, transformer-based balancing remains attractive where balancing speed across a long string matters more than per-cell hardware cost. Grid-scale BESS racks with hundreds of series cells are a plausible fit.
Bidirectional DC-DC Converter Balancing: The Newer Approach
Bidirectional DC-DC converters are the newest of the four active balancing topologies covered here. They use a full converter — often a buck-boost or bidirectional design — as the balancing circuit itself.
These circuits generally offer the best mix of efficiency, control precision, and bidirectional flexibility. Energy can flow either direction between any two points in the pack, under closed-loop control.
Recent research on bidirectional DC-DC converter balancing, including designs that route excess pack energy to an auxiliary battery, points to this family as the direction most new active-balancing research is heading.
The cost is component count and control complexity. A full converter needs more semiconductors, more sophisticated switching control, and more careful thermal design than a passive switched circuit.
For most BESS integrators, that cost is justified only when balancing speed or precision genuinely limits pack performance — not as a default upgrade.
Active Balancing Topologies Compared: Efficiency, Cost, and Scalability
Put side by side, the four hardware families trade off in predictable ways. No single topology wins on every axis.
| Topology | Balancing scope | Relative cost | Relative complexity | Typical fit |
| Switched-capacitor | Adjacent cells only | Lowest | Low | Small strings, cost-sensitive designs |
| Switched-inductor | Adjacent cells (higher current than capacitor) | Low–moderate | Moderate | EV packs, LFP BESS submodules |
| Transformer-based (flyback) | Any cell, non-adjacent | High | High (magnetic design, matching) | Long strings needing fast cross-pack balancing |
| DC-DC converter (bidirectional) | Any cell, bidirectional | Highest | Highest (control + semiconductors) | Precision-critical or research-grade designs |
Cost and complexity climb together for a reason. Reaching non-adjacent cells, or making energy flow bidirectionally, both need more active control over the switching network — not just a bigger version of the same simple circuit.
Choosing Among Active Balancing Topologies for BESS
Weighing active balancing topologies for most utility-scale and C&I LFP BESS designs, switched-inductor circuits remain the practical default. They deliver meaningfully higher balancing current than a capacitor-based design, at a cost most BMS suppliers can integrate at scale.
Transformer-based or DC-DC converter circuits earn their added cost in two situations: very long strings where adjacent-cell hopping slows equalization, or applications where balancing current itself needs to be large enough to matter for pack-level performance.
Whichever topology a supplier uses, the hardware only matters if the control logic triggers it correctly. See our guide on BMS algorithms for how balancing decisions actually get made.
And see our BMS for LiFePO4 batteries guide for the balancing-current and threshold questions worth asking a supplier directly.
Frequently Asked Questions
Is active balancing always better than passive balancing for BESS?
Not always. Active balancing avoids wasting energy as heat and corrects larger imbalances faster, which matters for high-cycle BESS applications.
But it adds cost and complexity that isn’t justified for small residential systems with high cell quality and low cycle frequency. Passive balancing is often the more practical choice there.
Which active balancing topologies are most common in BESS packs today?
Switched-inductor circuits are common in current EV and BESS BMS designs. They offer meaningfully higher balancing current than switched-capacitor circuits at a moderate cost increase.
Transformer-based and DC-DC converter circuits appear more often in research prototypes and higher-end designs, where balancing speed matters more than component cost.
Why don’t switched-capacitor circuits work well with LFP cells?
LFP’s voltage curve stays nearly flat across most of the state-of-charge range. A capacitor-based circuit senses and acts on voltage differences.
So a real SOC gap between LFP cells can show up as only a tiny voltage difference. That limits how much genuine imbalance the circuit can detect and correct.
Can a BMS combine more than one balancing topology?
Yes. Some designs pair two active balancing topologies — a fast adjacent-cell method like switched-inductor with a slower non-adjacent method for periodic full-pack equalization. This hybrid trades added control complexity for better overall balancing coverage.
How fast should active balancing correct a real imbalance?
There’s no universal target — it depends on balancing current and pack size.
As a reference point, a peer-reviewed 22-cell switched-inductor prototype reduced a 1.18V imbalance to 0.47V in about 2.5 hours, at a measured 908 mA balancing current. Slower or faster designs are both normal, depending on the topology and current rating chosen.
Further Reading
Battery Management System (BMS) Explained
BMS for LiFePO4 Batteries: Requirements, Parameters, and What to Check Before You Buy
BMS Algorithms Explained: SOH Estimation, SoP, SoE, Cell Balancing, and Safety Diagnostics for BESS
Flow-Based Market Coupling & Grid Congestion: How TSO Bottlenecks Shape BESS Dispatch in Central Europe
Central Europe’s power market no longer clears prices on simple, static border limits. Since June 2022, thirteen countries in the Core region have used flow-based market coupling to set day-ahead prices. This method models the physical grid, not just contract borders.
It has made transmission bottlenecks visible in a new way. Redispatch is the fix TSOs use when the market can’t physically deliver.
That fix has become one of the fastest-growing costs in European electricity bills. BESS developers need to understand how flow-based market coupling exposes congestion, and where storage can relieve it.
That understanding is now central to the investment case in Germany, Austria, and their Core-region neighbors.
| Quick Answer Flow-based market coupling is the day-ahead pricing method used across Central Europe’s Core region since June 2022. It replaces flat cross-border limits with a physics-based model of grid constraints. When those constraints near their limits, TSOs must redispatch power. Localized BESS can relieve that congestion directly. |
Key Takeaways
- Flow-based market coupling models real grid limits, called critical network elements, instead of flat cross-border caps. That is why congestion inside bidding zones, not just at borders, now drives market outcomes.
- Germany’s total grid congestion management costs have climbed sharply. They rose from about €187 million a decade ago to roughly €2.95 billion in 2024, and about €3.07 billion in 2025, per Bundesnetzagentur/SMARD data.
- Wind-heavy northern Germany and load-heavy southern Germany sit on opposite ends of a constrained corridor. The 2018 Germany-Austria bidding zone split was a direct response to the loop flows this mismatch caused across Poland, Czechia, and Slovakia.
- TSOs are now deploying transmission-connected battery storage as “storage-as-transmission-assets.” These projects are moving from pilot to rollout on the German grid, targeting the corridors with the highest redispatch volumes.
- Academic modeling shows preventive congestion management alone rarely clears as a standalone BESS business case. Revenue stacking with balancing and wholesale markets is what makes the economics work.
What Flow-Based Market Coupling Actually Does
Before 2015, most European day-ahead markets cleared against net transfer capacity, or NTC. This method set one fixed MW limit per border. It ignored what was happening elsewhere on the grid.
NTC is simple to run, but it gets the physics wrong. Electricity does not follow contract paths. It flows along the path of least resistance across a meshed AC network.
Flow-based market coupling replaces that fiction with a real physical model. It centers on critical network elements, or CNECs — the specific lines and transformers whose loading actually limits trade.
For each CNEC, TSOs calculate power transfer distribution factors against every bidding zone. Together, these form a “flow-based domain.”
The market-clearing algorithm treats this domain as its real constraint set. That replaces a string of separate bilateral limits.
The Core Capacity Calculation Region covers Germany, France, the Benelux countries, Austria, Poland, Czechia, Slovakia, Hungary, Slovenia, Croatia, and Romania.
This region went live with day-ahead flow-based market coupling on 8 June 2022. It built on a Central Western Europe pilot that had run since May 2015.
For BESS operators, the practical effect is simple. Market prices in the Core region now reflect grid physics far more closely than they did a decade ago.
But the model still can’t fully replace physical intervention. When a network element overloads after the market has cleared, someone still has to act. That gap is where redispatch, and storage, comes in.
Why TSOs Still Hit Bottlenecks Under Flow-Based Market Coupling
Germany’s grid shows the structural problem clearly. Wind generation is concentrated in the north. Turbines run offshore in the Baltic and North Seas, and onshore across the northern plains.
The largest industrial and urban load centers sit in the south and west. Conventional generation there, including nuclear, has been retired fastest.
Moving that northern power south needs corridor capacity. In many hours, that capacity simply doesn’t exist yet.
Until 2018, Germany, Austria, and Luxembourg formed one bidding zone. The market cleared huge volumes of scheduled trade between German wind and Austrian demand.
It never priced the physical constraint of actually delivering that power. Electricity that couldn’t take the direct path flowed instead through neighboring grids.
It moved through Poland, Czechia, Slovakia, and Hungary as unplanned “loop flow.” Those TSOs then had to intervene to protect their own networks.
After years of complaints to the EU’s Agency for the Cooperation of Energy Regulators, the joint zone split on 1 October 2018.
That split introduced an explicit bidding zone border and congestion management at the Germany-Austria interconnection.
Post-split research on cross-border flows found real gains. The split reduced unplanned flows across the Germany-Austria, Czech, and Slovak borders.
But Poland saw unplanned flows increase over the same period. Central Europe’s grid is interconnected enough that a fix at one border can shift pressure elsewhere.
The underlying mismatch between renewable generation and demand geography hasn’t gone away. Flow-based market coupling just makes its cost visible, instead of absorbing it silently.
The Cost of Congestion: Redispatch Spending Under Flow-Based Market Coupling
Redispatch happens after the market clears. Day-ahead and intraday trading produce a dispatch pattern that fits the flow-based domain.
But real-time conditions change: forecasts update, outages happen, grid topology shifts. A specific network element can still push toward its physical limit.
TSOs then pay generators to cut output in the oversupplied area. They pay others to raise output where the power is needed. In effect, they re-solve part of the dispatch problem after the market has already closed.
Germany’s Bundesnetzagentur tracks this spending under “Netzengpassmanagement,” or grid congestion management, and publishes the full-year figures through its SMARD platform. The total includes conventional redispatch, renewable curtailment compensation, and reserve power plant activation.
| Year | Total congestion management cost | Notes |
| ≈2014 | ≈ €187 million (est.) | Baseline implied by the roughly 15-fold increase to 2024 |
| 2022 | ≈ €4.2 billion | Energy-crisis year; sharp rise in gas-fired redispatch costs (figure not independently re-verified against a Bundesnetzagentur primary release — confirm before publish) |
| 2023 | ≈ €3.3 billion | Bundesnetzagentur/SMARD; near-record renewable curtailment |
| 2024 | €2.95 billion | Bundesnetzagentur/SMARD; 9.4 TWh curtailed (3.5% of renewable output); €554m curtailment compensation |
| 2025 | €3.07 billion | Bundesnetzagentur/SMARD; conventional redispatch €1.18bn; curtailment compensation fell 22% to €433m |
Two patterns stand out. First, the mix of costs is shifting. In 2025, conventional redispatch and reserve power plant activation — not curtailment compensation — made up the largest cost blocks.
Second, these costs track gas prices closely. TSOs often ramp gas-fired plants in southern Germany to fill the gap left by curtailed northern wind. When gas is expensive, redispatch gets expensive too.
Both patterns point the same way: relieving congestion with storage, rather than fossil generation, cuts both cost and emissions.
From Curative to Preventive: How BESS Relieves Bottlenecks Under Flow-Based Market Coupling
Conventional redispatch is a curative measure. It responds to congestion that has already appeared, close to real time.
Transmission-connected battery storage adds a second option: preventive congestion management. Here, a battery sits directly on a constrained corridor.
It charges and discharges to keep power flows within limits. That happens before an intervention becomes necessary.
German TSOs have started deploying utility-scale battery storage for exactly this purpose. The concept is often called the “Grid Booster,” or storage-as-transmission-asset (SATA).
Functionally, these systems act like a virtual transmission line. A new overhead line can take years to permit and build.
A battery at a key substation works faster. It absorbs excess generation on the constrained side of a corridor.
It then injects that power back on the other side, raising the effective use of the existing transmission asset.
One 250 MW project on the southern German grid went live in 2025. A second 250 MW system, on a different corridor, has been approved for service by 2028.
Beyond congestion relief, these installations are typically specified to also provide synthetic inertia, dynamic voltage control, and contingency reserves. That stacks several transmission-support functions onto one asset.
Academic modeling is more cautious about the limits. One widely cited techno-economic analysis of grid-operator-owned BESS for the German transmission system reached a clear conclusion.
Using a battery exclusively for preventive congestion management does not, on its own, clear economically out to 2030. The capital and operating costs don’t pay back against congestion-relief value alone.
The same research found something else. Combining dynamic line rating with distributed static series compensators was, in some scenarios, more cost-effective than batteries used alone for relieving wind-driven congestion.
For developers, the practical lesson is clear. Congestion relief works best as one revenue stream among several, not the sole reason for a project.
Where Storage Delivers Most Value: Locational Economics of Congestion Relief
Flow-based market coupling prices the zone, not the node. Germany’s day-ahead market still clears at a single national price, even though the underlying grid is congested internally.
That single-price design means a battery’s locational value for congestion relief never shows up in the wholesale price signal. It only shows up in what a TSO is willing to pay.
That payment comes through a SATA contract, an innovation tender, or a flexibility procurement mechanism, tied to a specific corridor.
Siting decisions hinge on proximity to the critical network elements that recur most often in TSO congestion forecasts and Network Development Plan corridors.
That typically means the north-south transmission backbone through Lower Saxony, Thuringia, and Baden-Württemberg. It also includes interconnection points along the Germany-Austria and Germany-Poland borders.
ENTSO-E’s own market design guidance for utility-scale storage frames this as a broader question. Transmission congestion is one of several distinct system needs storage can address.
The others include renewable curtailment risk, balancing reserve requirements, and voltage stability.
But each need requires its own market signal to attract the right asset to the right location. Where that locational signal is weak or absent, storage tends to cluster wherever wholesale arbitrage is largest. That is not necessarily where the grid needs it most.
Regulatory Tailwinds: The EU Electricity Market Design Reform
The EU’s Electricity Market Design reform was published in the Official Journal on 26 June 2024. It gives Member States a formal way to close that locational gap.
TSOs and DSOs must now submit a harmonized methodology for flexibility needs assessments to ACER.
Member States whose renewable flexibility investment falls short of that assessed need can then run Non-Fossil Flexibility Support Schemes, or NFFSS. These are support payments for new storage and demand-response capacity.
They can explicitly include locational criteria. The scheme must still keep the asset exposed to normal price and market risk.
This reform matters directly for flow-based market coupling regions. A national single-price zone like Germany’s doesn’t reward a battery’s corridor-specific value on its own — a dedicated support scheme can.
For BESS developers, this creates a legitimate, EU-sanctioned route for a congestion-driven siting premium. Developers no longer need to rely solely on a bilateral SATA contract with a single TSO.
A separate 2024 study for the European Commission’s Joint Research Centre covered redispatch and congestion management. It set the analytical groundwork both the flexibility needs assessments and NFFSS design now draw on.
Grid Reinforcement vs. Redispatch vs. Battery Storage-as-Transmission-Asset
In flow-based market coupling regions, these three tools work together rather than compete. Each solves a different part of the congestion problem.
| Factor | Grid reinforcement | Conventional redispatch | BESS (storage-as-transmission-asset) |
| Typical lead time | 10–15 years (permitting, construction) | Immediate, recurring operational cost | 2–4 years (siting, procurement, grid connection) |
| Cost driver | Steel, land, permitting, right-of-way | Spread between curtailed and ramped-up generation, gas-price sensitive | CAPEX plus O&M, offset by revenue stacking |
| Reversibility | Permanent, fixed capacity | Fully reversible, but recurring every constrained hour | Redeployable; can relocate as corridor needs shift |
| Primary role | Long-term structural capacity increase | Short-term correction after market clearing | Preventive and curative; bridges the gap until reinforcement is built |
| Consumer cost exposure | Financed via grid fees over decades | Passed through annually via grid fees | Can reduce both reinforcement need and annual redispatch spend |
Designing a Congestion-Relief BESS for Flow-Based Market Coupling Regions
Projects aimed at TSO congestion relief carry a different technical profile than a standalone frequency-response or wholesale-arbitrage asset. They need to perform reliably across several duty cycles at once.
- High cycling tolerance: Preventive and curative dispatch often means frequent partial cycles, not one clean cycle per day. That favors LFP chemistry’s cycle-life margin over higher energy-density alternatives.
- Fast, bidirectional response: Curative interventions close to real time need sub-second to few-second response. That is a similar profile to FCR products, even though the revenue mechanism differs.
- Siting discipline: Value concentrates at specific substations and corridor pinch points identified in TSO Network Development Plans. Proximity to the recurring critical network element matters more than proximity to load or generation alone.
- Revenue stack design: Congestion relief alone rarely clears a standalone business case. Contracts should let the asset also join balancing markets like aFRR and mFRR, or wholesale arbitrage, when congestion relief isn’t required.
These are largely the same asset traits that already govern the balancing-reserve cluster on this site — see the existing coverage of aFRR and mFRR, the MARI platform, and FCR response requirements. A well-specified BESS is often designed to serve congestion relief and frequency products from the same hardware. Dispatch priority is then set by contract, not by design.
Frequently Asked Questions
What’s the difference between flow-based market coupling and net transfer capacity?
Net transfer capacity (NTC) sets one fixed MW limit per border, independent of conditions elsewhere on the grid. Flow-based market coupling instead models the loading of specific critical network elements across the whole region.
It lets the market clear against that physical constraint set. This generally allows more cross-zonal trade, while representing congestion more accurately.
Why did Germany’s redispatch costs rise so sharply?
The core driver is a geographic mismatch. Most new wind capacity sits in the north, while demand and retiring conventional generation are concentrated in the south.
Transmission build-out hasn’t kept pace. Rising gas prices compound the effect, since TSOs often ramp southern gas plants to replace curtailed northern wind during congestion events.
Can a BESS actually replace the need for grid reinforcement?
Not entirely. Storage-as-transmission-assets can defer or reduce the scale of reinforcement and cut annual redispatch spend on a specific corridor. But transmission planning studies generally treat storage as a complement to reinforcement, not a permanent substitute for new transmission capacity.
Is preventive congestion management alone a viable BESS business case?
Independent modeling of the German transmission system found that using a BESS exclusively for preventive congestion management does not clear economically on its own out to 2030.
Projects that also join balancing markets or wholesale arbitrage alongside congestion relief have a materially stronger case.
How does the EU’s 2024 Electricity Market Design reform affect storage investment for congestion relief?
It requires TSOs and DSOs to formally assess flexibility needs and report them to national regulators. It also lets Member States run Non-Fossil Flexibility Support Schemes with explicit locational criteria, when private investment falls short of that assessed need.
That gives regulators a more direct policy channel for siting storage where congestion is worst.
Further Reading
Frequency Restoration Reserve (FRR): aFRR and mFRR Explained
Frequency Containment Reserve (FCR) Explained
The MARI Platform: Cross-Border mFRR Balancing
Fast Frequency Response (FFR): How BESS Stabilizes Grid Frequency
The MARI Platform: How mFRR Balancing Energy Trades Across Europe
The MARI platform is the piece of European market infrastructure that turns national mFRR reserves into one shared, cross-border pool. For a BESS asset manager, understanding how it clears bids is not optional background reading. It is the difference between a bid that earns revenue and one that never activates.
This guide goes beyond the platform basics covered in our Frequency Restoration Reserve overview, focusing on how the mechanics translate into bidding strategy for BESS asset managers.
| Quick Answer The MARI platform matches mFRR balancing energy bids across TSOs every 15 minutes. It builds a shared merit order list and clears at a marginal price per direction, but only if enough cross-border capacity exists. For asset managers, that capacity constraint often decides whether a bid clears. |
What Is the MARI Platform?
MARI stands for Manually Activated Reserves Initiative. It launched in October 2022, connecting five TSOs at first, under Article 20 of the EU’s Electricity Balancing Guideline.
The MARI platform exists to solve one problem. National mFRR markets used to clear in isolation. A shortage in one country and a surplus next door could not offset each other. MARI lets that offsetting happen automatically.
From National Markets to a Common Merit Order List
Every participating TSO submits its standard mFRR bids into a shared Common Merit Order List, or CMOL. The MARI platform rebuilds this list every 15 minutes, a period it calls a Market Time Unit.
An Activation Optimization Function then selects bids from the CMOL. Its first goal is maximizing total economic surplus. Only after that, it also tries to minimize how many cross-border exchanges the platform needs to meet demand.
How Bids Clear on the MARI Platform
Clearing on the MARI platform is not simply “lowest price wins.” Two separate conditions both have to hold.
First, the bid has to fall inside the accepted portion of the merit order. This applies for its direction and Market Time Unit specifically. Second, enough Cross Zonal Capacity has to exist on the transmission path between the bidder’s zone and the demand.
That second condition is easy to overlook. A BESS asset can offer the cheapest bid on the entire MARI platform. It can still miss clearing, simply because the interconnector between its zone and the TSO with the shortfall sits congested.
Direct vs. Scheduled Activation on the MARI Platform
The MARI platform runs two activation modes side by side. Direct activation is continuous. A pool of bids sits ready. The platform matches a TSO’s balancing need against that pool instantly, whenever it arises.
Scheduled activation, in contrast, works more like the older national mFRR process. It just runs coordinated across the shared platform now, instead of staying confined to one country. Both modes draw from the same underlying CMOL.
Marginal Pricing and Indivisible Bids
Every bid that clears on the MARI platform in a given direction earns the same marginal price. The most expensive accepted bid sets that price, not what each bidder originally offered. This pay-as-cleared structure rewards accurate cost estimation over aggressive underbidding.
Bids can also carry an indivisible tag. An indivisible bid clears entirely or not at all. For a BESS operator, a smaller, divisible bid size can sometimes clear more reliably than one large indivisible block, even at the same price.
MARI Platform Bidding Strategies for Asset Managers
Understanding the mechanics is the starting point. Turning that understanding into a bidding strategy is where the real economics live.
Balancing Markets Are Getting More Complex to Bid Into, Not Less
European balancing markets, including mFRR, are moving toward gate closures set closer to real-time delivery. Bid and price resolutions are shrinking too. That combination opens the market to more participants. It also makes correctly positioning a bid harder than it used to be.
An asset manager bidding on the MARI platform faces this trend directly. A bidding process built around slow, infrequent adjustments falls behind. Reviewing and adjusting bid levels closer to each Market Time Unit keeps pace with how the market has evolved, more than setting a static schedule days ahead ever could.
Treat Bidding and Dispatch as One Decision, Not Two
A common mistake: design a bid in isolation, then react to whatever activates. Research on value-stacking in adjacent European reserve markets, covering FCR alongside imbalance participation, points to a better approach instead. Shape the bid itself around the asset’s later dispatch flexibility, not after the fact. The same logic carries over to mFRR bidding on the MARI platform.
This matters most when a BESS stacks mFRR revenue from the MARI platform with other services. Think FCR and FRR, the products this platform sits alongside. A bid that looks optimal for mFRR in isolation can quietly eat into the state-of-charge headroom another service needs later the same day.
Consider the Cost of Capacity Withholding
Larger BESS portfolios face a different question entirely. Should the operator bid a smaller volume than what’s physically available? Doing so can influence the clearing price instead of simply accepting it. Broader research on storage bidder market power has derived formal bounds on this behavior, validated in simulations calibrated to the ISO New England market. It shows when capacity withholding still pays off, and when it turns self-defeating.
For most independent asset managers, this is more caution than strategy. A BESS large enough to move the clearing price also draws more regulatory scrutiny. Still, the underlying principle explains something real: very large storage fleets sometimes bid less aggressively than their full capacity would suggest, even outside the specific market that research examined.
Do Not Ignore Congestion Risk in Bid Placement
Cross Zonal Capacity gates every clearing decision on the MARI platform. Because of this, a BESS asset manager should treat interconnector congestion forecasts as seriously as price forecasts. A bid priced to win on the merit order alone can still miss clearing when the path to demand runs constrained.
Diversifying bid exposure across delivery windows helps too. Concentrating volume only in the most profitable hours raises the odds that one congestion event wipes out a day’s expected mFRR revenue.
MARI Platform vs. PICASSO
| Platform | Reserve | Activation | Pricing |
| MARI | mFRR | Manual, TSO instruction | Marginal, pay-as-cleared |
| PICASSO | aFRR | Automatic, centralized | Marginal, pay-as-cleared |
The MARI platform and PICASSO share the same underlying logic: a shared merit order and cross-border clearing. Each applies that logic to a different reserve product, with different activation timing.
FAQ
What does MARI stand for?
MARI stands for Manually Activated Reserves Initiative. It is the European platform that clears mFRR balancing energy bids across participating TSOs.
How often does the MARI platform clear bids?
The MARI platform rebuilds its Common Merit Order List and clears bids every 15 minutes. The platform defines this period as a Market Time Unit.
Why would a competitively priced bid fail to clear on the MARI platform?
Price alone is not enough. If not enough Cross Zonal Capacity exists between the bidder’s zone and the TSO with the balancing need, the bid cannot clear, no matter how competitive its price is.
Can a BESS asset manager bid into both MARI and PICASSO?
Yes. MARI handles mFRR and PICASSO handles aFRR separately. A BESS can participate in both, but sizing and state-of-charge planning need to account for both duty cycles at once.
Further Reading
Frequency Restoration Reserve (FRR)
Frequency Restoration Reserve (FRR): aFRR, mFRR, and How BESS Fit In
Grid frequency does not stay balanced on its own once primary reserve acts. Frequency Restoration Reserve (FRR) is the next layer that finishes the job, restoring frequency rather than just holding a line.
FRR is the collective term for two related services, aFRR and mFRR. Together, they bring frequency all the way back to its target value. They do more than just stop the drift.
| Quick Answer Frequency Restoration Reserve (FRR) restores grid frequency after primary reserve contains a deviation. It covers two products: aFRR activates automatically within about 5 minutes. mFRR activates manually within about 12.5 minutes. Both pay a capacity fee for standing ready, plus an energy fee when dispatched. |
What Is Frequency Restoration Reserve (FRR)?
FRR sits second in the grid balancing stack, right after primary reserve. Primary reserve only halts a frequency drift. Frequency Restoration Reserve pushes frequency the rest of the way back to 50 Hz.
FRR splits into two distinct products. Automatic FRR, or aFRR, is also called secondary reserve. Manual FRR, or mFRR, is also called tertiary reserve.
Each product differs mainly in what triggers it, not in its overall goal. Both aim to restore frequency and free up primary reserve for the next event.
FRR vs. FCR: Where It Sits in the Reserve Stack
Frequency Containment Reserve (FCR) acts first, within 30 seconds. It only stops a deviation from growing worse — see our full Frequency Containment Reserve (FCR) guide for how that first layer works. FRR follows, and actually restores the frequency.
Think of FCR as an emergency brake. Frequency Restoration Reserve is the driver correcting course afterward, over the following minutes. Neither product works well without the other.
aFRR: Automatic Frequency Restoration Reserve
A centralized TSO controller dispatches aFRR. Local frequency measurement does not trigger it. This is a key difference from FCR, which reacts locally at each individual asset.
Because dispatch is centralized, the TSO can rank available assets by cost. Operators call this ranking a merit order. It also lets the TSO account for grid congestion when choosing which assets to call on.
Under the harmonized European framework, aFRR assets must reach full output within about 5 minutes of an activation signal. That five-minute window is the deadline for full power, not the start of the response — the asset begins adjusting output within seconds of each signal, then ramps to full power over the five minutes that follow.
Signals typically arrive every few seconds. Instead of a single discrete instruction, setpoints are continuously adjusted in real time.
The PICASSO Platform
European TSOs coordinate aFRR through PICASSO, a shared cross-border platform. It lets balancing energy flow across country borders. This spares each grid from staying siloed inside a single control area.
More than two dozen TSOs now participate in the platform. As a result, a BESS in one country can, in effect, help balance a neighboring grid during a shortfall.
mFRR: Manual Frequency Restoration Reserve
mFRR, in contrast, requires an explicit instruction from the TSO. A human operator or an automated dispatch system tells the asset when to activate. The asset does not react entirely on its own.
Under the same harmonized framework, mFRR assets must reach full output within about 12.5 minutes. After activation, a bid must sustain that output for a minimum of 5 minutes.
mFRR steps in when aFRR alone cannot cover an imbalance. A large, sudden loss of generation, for example, can exhaust the available aFRR capacity fast.
The MARI Platform
Just as PICASSO handles aFRR, the MARI platform coordinates mFRR trading across European TSOs. It works toward the same cross-border balancing goal.
National mFRR auctions still exist alongside this shared platform in many markets today. Full integration is still an ongoing process across the continent.
Read the full MARI platform guide for how the Common Merit Order List clears bids across borders, plus bidding strategies for BESS asset managers.
Market Structure for Frequency Restoration Reserve
TSOs do not sell FRR as a single product. Both aFRR and mFRR split into separate capacity and energy markets, each with its own rules.
The capacity market pays for standing ready, whether or not the TSO calls on the asset. This market typically settles pay-as-bid. So each accepted bid earns exactly what it offered, not a shared clearing price.
The energy market, on the other hand, pays only for power actually delivered during an activation. This market typically settles pay-as-cleared. Every accepted bid then earns the same clearing price, regardless of what it originally bid.
Auction timing also differs by product. aFRR capacity auctions often close in the morning before delivery. mFRR auctions typically close slightly later the same day.
Why BESS Fit Well Into aFRR
A BESS can shift output in either direction almost instantly. This suits the merit-order, centrally dispatched nature of aFRR well.
One European TSO, for example, runs a two-phase aFRR structure: assets first reserve capacity ahead of each operational period. The TSO then dispatches them just-in-time within it, with a response time around 30 seconds after each call.
Operators can submit bids in that structure up to 25 minutes before each delivery window. They split into separate upward and downward regulation. A BESS’s ability to bid cleanly on both sides, from the same hardware, gives it a real edge. Most thermal assets cannot match that flexibility.
Because of this, BESS operators participating in Frequency Restoration Reserve markets often stack aFRR with other revenue streams. This spreads risk across several products instead of relying on just one.
Sizing a BESS for FRR Duty
An FRR asset needs enough usable capacity to sustain a full activation for the required duration. A simple instantaneous power rating is not enough on its own.
This matters more for mFRR, since its minimum delivery duration runs longer than a quick FCR burst. That means energy content, not just power rating, drives the sizing math.
State-of-charge management still applies here too, though less aggressively than under constant FCR cycling. Because activations are less frequent, the BESS has more time to recover between events.
But designers still need to plan for that recovery window, not simply assume it. A poorly sized recovery buffer can leave a BESS unable to respond to the next Frequency Restoration Reserve call.
Frequency Restoration Reserve in Practice: FCR vs. aFRR vs. mFRR
| Reserve | Trigger | Full Activation Time | Minimum Duration |
| FCR | Automatic, local | 30 seconds | 15–30 minutes |
| aFRR | Automatic, centralized | About 5 minutes | Varies by market |
| mFRR | Manual, TSO instruction | About 12.5 minutes | 5 minutes |
So the three products form a relay. FCR buys the first 30 seconds. Frequency Restoration Reserve then takes over in two stages. aFRR covers the next several minutes, and mFRR closes out anything that remains.
FAQ
What is the difference between aFRR and mFRR?
aFRR activates automatically through a centralized TSO signal. mFRR requires a manual instruction from the operator. aFRR also responds faster, typically within 5 minutes versus about 12.5 minutes for mFRR.
Is FRR the same as FCR?
No. FCR only contains an initial frequency deviation, within 30 seconds. Frequency Restoration Reserve, covering aFRR and mFRR, then restores frequency back to its nominal value over the following minutes.
Can the same BESS provide both FCR and FRR?
Some BESS do stack services. But sizing and controls must account for both duty cycles at once, since FCR and Frequency Restoration Reserve stress a system differently. A BESS designed only for FCR’s rapid, shallow cycling does not automatically suit FRR’s longer, sustained activations.
How is FRR paid?
Both aFRR and mFRR pay through separate capacity and energy markets. Capacity payments typically settle pay-as-bid. Energy payments typically settle pay-as-cleared.