Understood! I will conduct a literature review on workplace-based EV charging in the German aFRR market, focusing on:

• Simulation approaches vs. field tests, highlighting dominant methods (optimization, heuristic, AI-based) and real-world case studies.
• Aggregator models and baseline measurement techniques, particularly under the "free bids" framework.
• Profitability analyses of smart charging in the aFRR market, comparing "free bids" and pre-contracted capacity.
• Bidding strategies used for market optimization and handling uncertainty.
• Regulatory developments, particularly § 26a StromNZV and the separation of capacity/energy markets.
• Identifying research gaps and promising methodologies. I will summarize key studies in a literature matrix and provide a multi-page synthesis without going into deep methodological details. I'll focus on academic papers from the past 10 years and strictly European sources.I'll notify you once the research is complete!

Workplace EV Charging in the German aFRR Market: Literature Review

Literature Matrix of Key Studies

Simulation Approaches vs. Field Tests

Dominant Simulation Methods: The literature is dominated by simulation-based analyses and optimization studies. Researchers typically use mathematical optimization (e.g. mixed-integer or linear programming) to schedule EV charging and discharging for ancillary serviceswww.scipedia.comfile-6ectyd6zdjzljfmkzzcgiz. For example, Doumen & Paterakis (2019) developed a model combining EV charging data with market prices to calculate potential profits in Dutch reserve marketswww.scipedia.com. Many works assume an aggregator who optimizes a fleet’s charging against aFRR market signals – often using deterministic or stochastic programming. Some recent studies introduce sophisticated methods (e.g. stochastic optimization for uncertaintiesorbilu.uni.lu, robust optimizationorbilu.uni.lu, or heuristics) to improve upon simple deterministic schedules. AI-based scheduling has been less common in this niche; most authors rely on rule-based or optimization algorithms rather than machine learning. However, a few studies explore predictive algorithms for EV availability or price forecasting as part of the scheduling process (e.g. scenario generation for uncertain reserve activation in Pavić et al. 2023orbilu.uni.lu).

Insights from Simulations: Simulated scenarios generally show that controlled smart charging can yield cost savings or revenues compared to unmanaged charging. For instance, Rücker et al. (2020) found unidirectional smart charging (timing the charge when renewable power or low prices are available) cut annual charging costs by ~17%, and V2G could reach ~26% cost savings for a home userwww.mdpi.com. When adding aFRR participation, they observed a positive net return for providing reserve energy over 9 months – but it was very modest, not even covering the extra control infrastructure costswww.mdpi.com. This highlights a common simulation finding: theoretical revenues from aFRR can be positive but small with a single EV or small fleet under recent price conditions. Larger fleets or aggregators are expected to scale these profits.

Limited Field Test Data: Real-world field evidence is scarce. _“Most research approximates profitability via simulations; only one field test (a car-sharing fleet) has been reported”_file-6ectyd6zdjzljfmkzzcgiz. The primary field pilot in Europe has been the V2X Suisse project (2022–2024) with 50 car-sharing EVs providing frequency regulation. Its findings confirm many simulation assumptions, but also surface practical challenges. User acceptance in this trial was very high – over 6,600 different customers used the V2G-enabled cars with no noticeable negative feedbackwww.mobility.ch. This real-world validation is encouraging, suggesting that if managed properly, drivers don’t mind or even notice their workplace or shared EV is supporting the grid. The V2X Suisse pilot also demonstrated technically reliable operation: vehicles responded to control signals within ~1–2 seconds as required for reserveswww.mobility.chwww.mobility.ch. However, economic results in the field lag behind simulations. The Swiss trial reported that current revenue from frequency reserve markets does not yet cover the high investment and operating costs of V2G systems in a car-sharing modelwww.mobility.chwww.mobility.ch. In other words, today’s aFRR/FCR prices would not yield a profit once you factor in expensive bidirectional chargers, upgraded EV batteries, communication infrastructure, etc. This aligns with early simulations that showed slim profit margins. It underscores the need for either higher reserve prices, lower technology costs, or additional value streams to make workplace or fleet aFRR integration profitable in practice.

In summary, simulation studies dominate and indicate promising but modest economic benefits from netzdienliches (grid-serving) EV charging. Field tests are just emerging – they validate that the concept works and is acceptable to users, but also warn that real-world costs and operational complexities can erode the theoretical gains. This gap between simulated potential and proven real-world benefit is a central theme in recent literature.

Aggregator Models and Baseline Issues

Handling EV Availability & Uncertainty: Aggregators pooling workplace EV chargers face significant uncertainties in scheduling: vehicle arrival and departure times, state-of-charge (SoC) upon arrival, and actual reserve activation requests. Research increasingly highlights advanced models to manage this uncertainty. “Electric vehicle aggregators face uncertainty both on the reserve activation and the vehicle availability... which can hurt profitability and user comfort”orbilu.uni.lu. Pavić et al. (2023) showed that naive deterministic models often overestimate how much aFRR an EV fleet can reliably provide, because they neglect variability in user behavior and grid requestsorbilu.uni.lu. They proposed stochastic scheduling (multiple scenarios for arrivals/activations) and robust optimization (worst-case aware bids) to hedge against these unknownsorbilu.uni.lu. The improved methods substantially increased the aggregator’s successful delivery of aFRR and profits compared to a deterministic approachorbilu.uni.lu. In practice, this means aggregators should integrate probabilistic forecasts – e.g. the likelihood a given charging station will have a car plugged in and how deeply it can discharge – when deciding how much capacity to offer.

Other studies reinforce the importance of modeling EV user behavior. A Belgian case study found the profitability of providing aFRR at semi-public chargers is “heavily dependent on the plug-in ratio... determined by EV users’ behavior.”www.mdpi.com. If many employees plug in for long durations, an aggregator has ample flexibility to sell; if EVs are sporadically connected, the available reserve capacity shrinks and revenue drops. Aggregators in literature handle this by either offering conservative bids (assuming only the minimum expected EVs) or by incentivizing behavior (e.g. reward employees for plugging in as soon as they arrive and not unplugging until needed). Some works also introduce penalty costs for unplanned EV departures – effectively, if a car that was promised for reserve leaves early, the aggregator must buy out of its commitment – to model the risk of availability uncertainty. By iterating scenarios or using robust bounds, algorithms can find a middle ground between overcommitting (and risking penalties) and underutilizing the resource.

Baseline Measurement Techniques: A critical challenge in demand-side aFRR (such as flexible EV charging) is how to define and measure the baseline, i.e. the counterfactual power consumption if no reserve activation occurred. Baseline accuracy is crucial for both fair compensation and for not double-counting energy. In Germany, the regulatory framework (§26a StromNZV) explicitly requires a baseline-based settlement: when an aggregator delivers positive reserve (up-regulation) by reducing charging or feeding energy from EV batteries, the EV owner’s supplier is “virtually treated as if they had delivered the energy that would have been consumed without the activation”raue.comraue.com. In practice, the end-user (or aggregator on their behalf) must determine a baseline load profile for the EVsraue.com. This is often the hardest piece of aggregation: you must predict what each EV would have drawn from the grid in the next minutes if it were just charging normally, then measure the difference when you adjust charging for aFRR.

Various baseline methodologies appear in the literature and industry: e.g. “Last observed period” (using the last X minutes’ average as the forecast), or “High X-of-Y” (taking a representative high consumption from recent similar days to avoid underestimating baseline). Some European TSOs allow declarative baselines, where the aggregator submits their calculated baseline ahead of time, subject to verification. For example, Belgium’s TSO (Elia) provides default methods like last quarter-hour or high-of-day, but also lets aggregators propose their own baseline if they prove its accuracy in a test periodwww.elia.bewww.elia.be. The high granularity of aFRR (4-second control signals in the PICASSO platform) makes baselining especially challengingwww.elia.be. Elia found that forecasting an EV’s baseline one minute in advance for every 4-second interval is error-prone, and is moving toward letting aggregators update the baseline in real time with proper auditingwww.elia.bewww.elia.be.

In academic studies, baseline issues have only recently begun to get attention, often in the context of free bids. One study (Dreher et al. 2024) discusses using new smart meter systems (iMSys) to measure EV charging load precisely for baseline calculation, indicating that metering and data handling must evolve to support accurate baselines. Aggregators in practice might use historical charging profiles of each workplace EV user as a baseline model – but if users behave unpredictably (e.g. an employee decides to leave early and unplugs), the baseline and actual diverge. An inaccurate baseline can lead to paying for “phantom” flexibility or not getting paid for real response, undermining the business case. Thus, baseline calculation is a non-trivial component of any EV reserves aggregation model. The consensus is that baseline methods need to be highly adaptive (near-real-time updates) and transparentwww.elia.bewww.elia.be, to handle the fast timescales and prevent gaming of the baseline (e.g. intentionally inflating expected consumption).

Under Germany’s “free bids” framework, baseline accuracy is even more crucial. Since free bids allow an aggregator to inject or drop load opportunistically without a long-term capacity contract, the TSO must rely on the baseline to quantify the energy actually delivered. The German guidelines require that the baseline be established independently of the activation signal (to avoid self-fulfilling prophecies) and that the supplier’s balance is corrected using this baselineraue.comraue.com. In summary, research acknowledges that better baseline methodologies and technologies (e.g. high-resolution sub-metering at chargers, machine-learning load forecasts, or standard baseline formulas tested for EV profiles) are needed to unlock reliable aFRR services from workplace charging.

Profitability of Smart Charging for aFRR

Cost-Benefit Analyses: Recent European studies have attempted to quantify the economic upside of integrating aFRR services into workplace charging. Most conclude that there is a potential revenue stream, but its size varies widely with context. Simulation-based cost-benefit analyses often compare scenarios: (a) uncoordinated charging, (b) smart charging for cheapest energy (peak shaving or PV self-use), and (c) smart charging with ancillary service participation (e.g. aFRR). For example, Rücker et al. (2020) compared these strategies and found that grid-responsive charging (timed to support the grid and respond to aFRR signals) did reduce net costs for the EV owner/employer versus dumb charging. The combination of PV optimization and occasional aFRR response yielded the greatest savings (up to 26% reduction in annual charging cost in their case)www.mdpi.com. However, the absolute monetary gain was relatively small – one EV providing aFRR for 9 months earned a positive return but on the order of only tens of Euros, not enough to justify the hardware investmentwww.mdpi.com.

“Free Bids” vs. Pre-contracted Capacity: A key profitability question is whether an aggregator should rely on the new “free bids” model (energy-only bidding) or pursue traditional capacity contracts for reserve provision. In a pre-contracted capacity model, the aggregator offers a fixed kW capacity for aFRR in advance (e.g. in a monthly or weekly auction) and is paid a capacity fee regardless of whether energy is activated, plus a regulated energy payment when called. This guarantees income but requires the EV fleet to be on standby continuously, and penalties apply if the capacity isn’t delivered on request. In contrast, the “free bids” approach (now possible in Germany’s aFRR since the market redesign) lets an aggregator bid energy on the fly without a prior capacity commitment – essentially, you only sell aFRR energy from EVs when it’s advantageous or when cars happen to be available. The trade-off reflects revenue stability vs. flexibility:

• Pre-contracted capacity provides a stable revenue floor (capacity payments), which can make the business case more predictable. Studies note that this approach requires very conservative operational planning – the EVs must always hold enough charge or be ready to curtail load to deliver the reserved capacity, even if the probability of activation is low. The upside is limited by the capacity price (which in recent years for aFRR has been declining in competitive markets). If the EVs are rarely called upon, the aggregator still earns the capacity fee, essentially monetizing the cars’ idle flexibility. Several simulation studies incorporate capacity revenues: e.g. an EV fleet bidding into FCR (which is purely a capacity market) shows decent payback if enough vehicles are aggregated because the steady payments accumulatewww.mdpi.com.

• Free bids (energy-only) can in theory earn more if there are frequent high-priced activations. Under this model, EVs charge or discharge only when the TSO dispatches aFRR from them. Germany’s integration into the European platform has led to periods of very high balancing energy prices (especially during grid stress or renewable forecast errors). An aggregator using free bids could take advantage of price spikes without the obligation to be available at all times. However, this comes with volatility and risk. If the year has few aFRR activations or prices are low, the aggregator earns little. Also, competing against traditional providers means free bids will often set a downward pressure on balancing energy priceswww.next-kraftwerke.com– effectively, the opportunity is only lucrative when the system is short on other reserves. Next Kraftwerke notes that allowing free bids increases competition and tends to reduce the clearing price for aFRR energywww.next-kraftwerke.com. Thus, the average revenue per MWh of aFRR may drop as more players (like EV aggregators) join via free bids. Literature directly comparing the two approaches is sparse (because “free bids” are very new), but we can glean some insights. Pavić et al. (2023) implicitly address this by modeling the “balancing energy procurement uncertainty”orbilu.uni.lu– in a free-bid world, the amount of energy an EV aggregator ends up selling is uncertain, whereas in a capacity contract world, the capacity payment is fixed but the activation volume might be low. Their stochastic results suggest that if an aggregator optimizes for expected balancing energy, it can improve profits, indicating that taking some calculated risk on energy activation (akin to free bids) can out-perform an overly safe capacity-only strategyorbilu.uni.lu. On the other hand, the Swiss field test (V2X Suisse) effectively operated more like a capacity model (providing FCR capacity continuously) and found that current revenues did not cover costswww.mobility.ch. This suggests that until technology costs drop, chasing only capacity payments might not be profitable either.

In summary, “free bids” offer higher upside but with higher variance. An aggregator focusing on workplace charging might adopt a hybrid strategy: secure some base capacity contracts for steady income and use a portion of EV flexibility for opportunistic free bids when many cars are connected or when prices peak. Importantly, free bids lower the barrier to entry (small workplace fleets can participate without prequalification for large capacity blocks), which could boost overall profitability for more players if managed well.Revenue Potential vs. Risk: The revenue potential from aFRR at workplaces is highly scenario-dependent. A well-utilized workplace charging lot (many EVs plugged in all day) in a country with frequent aFRR activation and decent prices can generate meaningful income. Goncearuc et al. (2023) estimated a “significant positive incremental profitability” for adding aFRR to a hospital’s 12-charge-point networkwww.mdpi.com– essentially turning a cost-center (providing employee charging) into a modest revenue stream for the site operator. They stress this is contingent on user behavior: in their best-case, EVs were plugged in long enough each day to frequently respond to grid signals. If, for example, employees only plug in for 2 hours, the available flexibility window shrinks and so do earnings. Risk comes in the form of uncertainty of both availability and market conditions. Researchers incorporate risk by simulating worst-case scenarios (e.g. an aggregator commits to a capacity but half the cars leave unexpectedly – incurring imbalance costs). Generally, the findings encourage pursuing aFRR integration only if user charging needs are guaranteed to be met first (no one wants an empty battery at 5 PM because the car discharged for grid services) and if the aggregator has a sound strategy for handling variability (like a buffer of extra EVs or limiting commitments to less than expected available capacity).

Overall, the literature’s cost-benefit analyses show promise of additional revenue or savings on charging costs – typically on the order of 10-30% improvement over unoptimized chargingwww.mdpi.com. Yet, they also caution that profitability is currently margin-thin. High equipment costs (for V2G chargers, control systems, certification) and the need to share revenues (between employer, aggregator, possibly employees) can erode the business case. As one field study bluntly concluded, “currently high costs and the regulatory context do not yet allow a profitable V2G business model in a car-sharing fleet”www.mobility.ch. Workplaces may have a slight edge: unlike car-sharing, an employer might be willing to invest for indirect benefits (resilience, green image, providing cheap charging as a perk) in addition to pure profit. Going forward, profitability should improve as technology costs decline and as regulatory frameworks (like free bids) enable more competitive market participation.

Bidding Strategies in the aFRR Market

Optimal Bidding Methods: Bidding strategies refer to how an EV aggregator decides the quantity and price of aFRR it offers to the market. In reviewed literature, this is typically formulated as an optimization problem where the objective is to maximize profit (or minimize cost) while respecting EV constraints (battery limits, user needs). Most studies use mathematical optimization to derive bidding strategies. Common approaches include:

• Deterministic optimization: assume expected values for uncertain parameters and optimize accordingly (e.g. offer the average available capacity). This is simpler but can lead to suboptimal or infeasible results if reality deviates.
• Scenario-based stochastic optimization: generate multiple scenarios of EV availability and aFRR activation and optimize expected profit or a risk-weighted metric. Pavić et al. used this to capture balancing energy uncertainty, effectively co-optimizing energy and capacity bids across scenariosorbilu.uni.lu.
• Robust optimization: optimize for the worst-case within some uncertainty bounds. In Pavić’s work, a robust model ensured the aggregator could meet its commitment even in low availability or high activation cases, trading off some profit for higher reliabilityorbilu.uni.lu.
• Heuristics and rules: Some works propose simpler heuristic bidding rules, like always bid a fixed fraction of the total plugged-in EV capacity, or bid more aggressively during certain hours (e.g. midday when many workplace EVs are present and aFRR needs are historically higher). These are often used when computational simplicity is needed, but they might not capture the full economic optimum. Market Bidding Optimization: The problem can be complex because an EV aggregator effectively has to decide on two things: how much capacity to offer (if participating in capacity auction) and how to price the energy (or whether to bid at all if using free bids). Seo et al. (2024) specifically address bidding in both FCR and aFRR markets. Their strategy likely involves allocating battery resource to FCR (which might be more lucrative per kW but ties it up continuously) versus aFRR (which might use the battery only when activated but requires flexibility). While we don’t have their detailed results here, such studies often use multi-stage optimization: first decide capacity bids for FCR/aFRR day-ahead, then plan intra-day energy bids or reserve activations. The most-used method is to assume perfect competition and bid at marginal cost – for an EV, the “cost” of providing aFRR up or down is mainly the value of energy and battery wear. For instance, if discharging an EV for aFRR, the bid price might account for the expected cost to recharge that energy later plus a degradation cost on the battery. In practice, academic models sometimes simplify and bid zero or very low prices to ensure they get selected, then focus on the scheduling problem of actually delivering the energy.Handling Uncertainty in Reserve Activation: A major focus of bidding strategy research is uncertainty: an aggregator doesn’t know in advance how often or when aFRR will be activated, or exactly which EVs will be available. Strategies to handle this include:
• Probabilistic bidding: Only offer the amount of reserve that you are, say, 95% confident you can deliver given the uncertain number of EVs. This reduces risk of failure. Pavić’s stochastic model is an example – it effectively bids considering probability distributions of activationorbilu.uni.lu.
• Adaptive bids: Update bids closer to real-time as uncertainty resolves. For example, in some markets you can adjust offered capacity intra-day. An aggregator might bid conservatively day-ahead, then increase their offer in the intra-day session if by that morning they see an unusually high number of EVs plugged in. This adaptive approach wasn’t possible in older weekly auctions, but with Germany moving to daily auctions and free bids, it’s feasible. Some literature points out that shorter bidding intervals favor aggregators of EVs, as they can align bids with actual parking lot status.
• Incorporating forecast error costs: Bidding strategies often include penalty terms or expected imbalance costs if the aggregator cannot deliver. By doing so, the optimization naturally scales back bids in the face of high uncertainty. For example, a model might estimate the cost of a shortfall (needing to buy energy from the TSO at a potentially high price if an EV disappears)orbilu.uni.lu, and will only bid an amount where the expected profit outweighs that risk.
• Pricing strategy under uncertainty: If uncertain, an aggregator might bid a higher price for activation to reduce the chance of being called unless it’s truly needed. This way, if they are not absolutely sure they can deliver easily, they only get activated in extreme cases (and likely get paid a high price which compensates the trouble). This is a rational strategy in free bid energy markets – some participants bid very high prices to ensure they are last in the merit order and only activated as a last resort. There’s an ethical/regulatory boundary here (very high bids can look like price gouging). The German regulator, after some extreme bids (up to €77,777/MWh in 2017)www.next-kraftwerke.comwww.next-kraftwerke.com, had to reform the market. So aggregators need to balance aggressive bidding with regulatory acceptable behavior. In literature, most optimization-based bidding strategies show that intelligently accounting for uncertainties can significantly improve outcomes. Pavić et al. demonstrated that robust or stochastic bids lead to far fewer costly delivery failures and higher average profit than simpler approachesorbilu.uni.lu. Another study by Seo et al. likely found an optimal split between FCR and aFRR bidding – e.g., maybe dedicating some EVs to FCR capacity and others to aFRR energy yields the best overall revenue for a fleet.

We also see proposals for hierarchical bidding: the aggregator might first decide a desired aggregate energy profile for the whole workplace (to minimize costs or maximize revenue), and then a lower-level scheduler allocates specific charging rates to each EV to meet that profile. This separation of market bidding and local dispatch is common for virtual power plant models. Mödder et al. (2023) in their fleet cost optimization likely did something along these lines – optimize the fleet’s grid exchange, then control each EV accordingly.In summary, the literature suggests sophisticated bidding strategies are needed to unlock EV flexibility value. Simply bidding all available EV battery power into aFRR at all times is not optimal. The best results come from optimization methods that consider uncertainties, user requirements, and multi-market opportunities, producing a bidding strategy that balances risk and reward. Techniques like scenario-based stochastic programming and robust optimization are increasingly used to handle the uncertain nature of EV flexibility and aFRR activationorbilu.uni.luorbilu.uni.lu. As computational tools improve, we even see discussion of AI (e.g. reinforcement learning could learn bidding policies, though concrete examples in EU aFRR aren’t prevalent in current literature).

Regulatory Developments and Their Impact

The regulatory landscape in Germany and Europe for ancillary services has evolved significantly in the past few years, directly affecting netzdienliches Laden (grid-serving charging) with EVs:

• Separation of Capacity and Energy Markets: Germany historically used a “mixed price” system for aFRR (SRL) where capacity and energy bids were combinedwww.next-kraftwerke.com. Since mid-2022, in line with EU Electricity Balancing Guidelines (Reg. 2017/2195), Germany reformed aFRR procurement to separate capacity reservation from energy activation. This means providers bid capacity in advance auctions, and aFRR energy is settled in a real-time balancing energy market. A key outcome is the introduction of “free bids” – energy-only bids from participants who did not necessarily hold a capacity contract. These free bids are allowed to enter the activation merit order alongside contracted resourceswww.next-kraftwerke.com. For EV aggregators at workplaces, this is a game-changer: even small aggregators who might not win day-ahead capacity can still attempt to sell energy during activation periods. The presence of free bids increases competition and has been observed to put downward pressure on balancing energy priceswww.next-kraftwerke.com, benefitting the system with lower costs but meaning aggregators must be efficient to make money in this competitive environment. Overall, the separation incentivizes participants to be very accurate in bidding (since capacity payments no longer cushion poor energy performance) and rewards those who can respond flexibly at short notice (which EVs can, if coordinated properly).

• Baseline Requirements (§26a StromNZV): Section 26a of the German Stromnetzzugangsverordnung (Electricity Grid Access Ordinance) was a legal enabler for independent aggregation of demand resources. It established that a consumer (Letztverbraucher) can offer reserves via a third-party aggregator, and outlines how the supplier’s balance is corrected using a baselineraue.comraue.com. In practice, this requires the aggregator to calculate and provide a baseline consumption profile for the EVs. The ordinance and subsequent BNetzA rulings (e.g. BK6-17-046) set expectations that baseline methods should be accurate and typically subject to approval or verification. The recent rollout of intelligent metering systems (iMSys) in Germany is partly to facilitate such detailed measurements. Impact: For netzdienliches Laden, §26a formalized the process but also imposed responsibility: without a reliable baseline, an EV fleet cannot participate. This has spurred research into baseline methodologies (as discussed) and pilot programs to test baseline accuracy. Aggregators must also navigate billing and taxes – §26a clarifies that for upward reserve (reducing consumption) the supplier is made whole as if the energy was delivered, and for downward reserve (increasing consumption) the extra energy is treated normally with grid fees etc.raue.com. This ensures end-users aren’t penalized with imbalance costs, but it means the aggregator has to manage the baseline settlement and any deviations. The regulatory emphasis on baseline accuracy (and independence from manipulation) is highwww.elia.bewww.elia.be, adding a layer of compliance work for any aggregator.

• PICASSO Platform and EU Integration: Germany’s aFRR is now part of the EU-wide PICASSO platform for balancing. This brings 4-second control cycles and cross-border competition. Regulations now mandate standard products (minimum bid sizes have reduced – in aFRR often 1 MW or less now, enabling smaller aggregators to participate). Also, the gate closure times are closer to real-time, allowing more dynamic bidding. For workplace charging, this means an aggregator could decide aFRR offers just 15-30 minutes before the delivery period, using up-to-date info on how many cars are plugged in. The regulatory push for 15-minute or shorter markets aligns well with EV flexibility, which is often short-term.

• §14a EnWG (in Germany) and others (notably §26a StromNZV): These national regulations encourage flexible loads (including EVs) to support the grid, e.g. via load management or reserves, and provide frameworks for interruptible loads. Recent amendments (like the draft versions of §14a) aim to simplify how EV charging can be controlled by grid operators or market parties. This regulatory development might introduce tariffs or incentives for “grid-friendly” charging. While not directly about aFRR, it could complement aFRR revenue by lowering electricity costs for flexible EV charging. There is active discussion in Germany on how to coordinate distribution grid management (like peak shaving of EV load under §14a) with participation in markets like aFRR – ensuring an EV is not “double controlled” or that baseline for aFRR isn’t disrupted by a distribution grid operator throttling the same charger.

• Decoupling of energy and capacity pricing & price caps: After some extreme events, regulators introduced price caps for balancing energy (to prevent the €77,000/MWh incident from repeatingwww.next-kraftwerke.com). Caps might limit the maximum aFRR price (e.g. a cap of €15,000/MWh was discussed). This protects consumers but also caps the windfall an aggregator might get in a rare scarcity event. Thus, an EV aggregator’s “jackpot” scenario is somewhat curtailed, reinforcing the need for a solid everyday business case, not just betting on rare extremes.

• Accounting for battery degradation: While not a regulation yet, some markets consider requiring energy-limited resources (like batteries or EVs) to account for their state of charge in bids. For instance, if an EV is low on charge, it might not be allowed to offer upward regulation. German prequalification rules require demonstrating sustained delivery for aFRR (typically 15 minutes full activation). This effectively forces aggregators to manage state-of-charge of EVs. Regulations might evolve to be more EV-friendly (e.g. shorter sustained activation requirements when a fleet of EVs can swap which one provides power). Research by Dreher et al. (2024) hints at using the new smart meter framework to integrate many small EV resources to collectively meet such requirements. Impact on Netzdienliches Laden: On the positive side, recent regulations have opened the door wider for EVs to play in aFRR. Lower bid size minimums, free bids, and clear baseline rules mean even a modest workplace EV fleet can technically qualify and participate, something that was practically impossible 5–10 years ago. This is spurring pilot projects and commercial interest in “EV aggregators” across Germany and Europe. On the challenging side, the regulations impose strict measurement and performance standards. Workplace charging operators must invest in telemetry, metering, and reliable control systems to meet these standards, incurring upfront costs. Also, the split of markets (capacity vs energy) means an aggregator has to navigate two revenue streams and more complex bidding, rather than just securing a simple contract – effectively more market risk to manage.Finally, §26a StromNZV ensures that the act of netzdienliches Laden is legally recognized: an employer can let a third-party aggregate its charging stations for aFRR without violating their supplier contract, as long as baseline and balancing responsibilities are sorted. This legal clarity was necessary and is a positive development, but it also means any failure (like baseline errors leading to imbalances) is squarely the aggregator’s responsibility. Therefore, only players with the technical capability to handle this will jump in.In conclusion, regulatory developments in Germany (and the EU) have largely been enabling – making it feasible for EVs at scale to contribute to aFRR – but they also add new requirements (baseline accuracy, real-time data, prequalification tests) that shape the implementation of netzdienliches Laden.

Research Gaps and Recommendations

Despite the progress in both modeling and pilot demonstrations, several research gaps remain in the realm of workplace-based EV charging for aFRR:

• Limited Field Validation: There is a notable lack of real-world data on aggregated EV performance in aFRR beyond small pilot studies. The literature is still hungry for more field test results at workplaces – e.g. office campuses or factory parking lots – to validate simulation findings. So far, the only comprehensive field trial (V2X Suisse) was in a car-sharing context, not a typical employer-employee scenario. Gap: How do private EV drivers at a workplace respond in terms of plug-in behavior when their charging is controlled for aFRR? Does it differ from the car-share case? Recommendation: Conduct targeted pilot projects at corporate campuses or public-sector office sites with significant EV adoption. These should measure not just technical performance and cost, but also track user engagement (do employees opt in? do they abide by requested plug-in schedules? how do they feel about the process?). Such studies will provide empirical evidence on user acceptance and the real cost savings achievable, strengthening or adjusting the assumptions made in simulations.

• User Behavior and Engagement: Relatedly, user behavior remains one of the least certain aspects. Many studies assume EV users will input their departure times accurately and always plug in on arrival – but reality may differ. The Swiss trial fortunately showed very high acceptancewww.mobility.ch, but that was with professional fleet management. Gap: In a workplace, how to maintain high participation? For instance, if employees frequently forget to plug in, the aggregator’s available capacity might consistently fall short. Recommendation: Research into incentive mechanisms and human-factors is needed. This could involve experiments with reward systems (e.g. discounted charging or gift cards for those who keep their EV plugged in during certain hours), or app-based nudges reminding users to plug in. Studying these “soft” aspects in a real workplace setting would complement the technical literature.

• Aggregator Algorithms & Tools: On the algorithmic side, while advanced models exist, there is room to improve practical aggregator tools. Many stochastic or robust optimization models are complex and may be difficult to run in real-time or integrate with energy markets. Gap: Bridging the gap between academic models and deployable software. For example, implementing a stochastic bidding strategy in a live trading system is non-trivial. Recommendation: Develop and open-source prototype aggregator algorithms that can take live data (current EV plug-in status, day-ahead prices, etc.) and output market bids and charging plans. These should be tested in simulation and ideally in field pilots. Including AI techniques (like machine learning predictors for EV arrival or reinforcement learning for bidding under uncertainty) could be explored to see if they outperform traditional methods in practice.

• Baseline Accuracy and Simplification: Baseline determination remains an Achilles’ heel. Only one or two studies explicitly deal with baseline methods for EV flexibility, often assuming an ideal baseline. Gap: A lack of research on baseline refinement for EV charging loads. The baseline for an EV could perhaps leverage machine learning (training on past charging sessions to predict what charging profile would have been) – this is an area not deeply explored. Also, the impact of baseline errors on revenues has not been quantified well in literature. Recommendation: Future research should test different baseline methodologies on historical EV charging data to quantify accuracy and potential biases. For example, take a dataset of uncontrolled workplace charging, simulate various baseline calculation methods (last 15-min usage, average of last 3 similar days, etc.), and then simulate aFRR activations to see how much error each method introduces in estimating delivered reserve. Additionally, collaboration with regulators to refine baseline rules (perhaps simplifying requirements for smaller resources) could be informed by such research.

• “Free Bids” Strategy & Market Interaction: Free bids are new and thus under-researched. Gap: Uncertainty about how an influx of EV aggregators using free bids might affect the balancing market (prices, reliability) and their own risk exposure. Recommendation: Studies could simulate scenarios where many EV aggregators participate with free bids to see if, for instance, balancing energy prices become very volatile or if aggregators cannibalize each other’s profits. Also, analyzing risk vs reward: e.g. comparing an aggregator always doing free bids vs always contracting capacity over a year of realistic operations – which yields better profitability and under what conditions? This would guide business models for aggregators (whether to prioritize long-term contracts or spot market opportunities).

• Battery Degradation and Life-Cycle Costs: Only a few papers touch on battery wear cost due to V2G. Rücker et al. (2020) mentions battery aging in passingwww.mdpi.com. Gap: Precise evaluation of how frequent aFRR cycling (with its rapid up-and-down commands) impacts EV battery life, and the economic cost of that. If providing aFRR significantly hastens battery degradation, any revenue could be offset by the need for earlier battery replacement – a crucial factor for long-term sustainability. Recommendation: Integrate battery degradation models into the cost-benefit analysis. Some ongoing research in energy storage could be applied to EVs, calculating an approximate cost in euros per MW of aFRR provided in terms of battery wear. Including this in profitability assessments would give a more realistic picture. It may turn out that aFRR, which typically has many short cycles, could be more wearing than, say, participating in slower mFRR or just energy arbitrage. If so, that might shift recommendations toward using workplace EVs more for slower services or only using a fraction of the battery for aFRR to limit wear.

• Scale and Grid Integration: Many studies consider a single site or a single aggregator in isolation. Gap: How multiple aggregators or larger scale deployment would interact with grid constraints. For example, if many workplaces in one city all start providing aFRR by charging or discharging EVs, could this create local grid issues (like a distribution transformer overload at noon when all suddenly charge for downward aFRR)? This multi-actor, grid-level effect is not deeply studied yet. Recommendation: Perform power grid impact studies that incorporate large numbers of flexible chargers responding to reserve signals. Identify if any regulatory guardrails or coordination (with DSOs) is needed to ensure that what is good for the TSO (frequency control) is also okay for the local grid. This touches on the integration of §14a EnWG with aFRR: research could propose control schemes that satisfy both local peak management and global aFRR needs.

• Holistic Business Models: Finally, there is a gap in translating these technical findings into business models for aggregators and employers. Who invests in the equipment and who reaps the savings? If the employer installed the chargers, will they get a cut of aFRR revenue? How to split incentives so that EV drivers, the aggregator, and the site owner all benefit enough to participate? This is more of a techno-economic and perhaps social science question. Recommendation: Case studies or demonstration projects should document the contractual arrangements and participant feedback, which can then be analyzed to recommend models (for example, an aggregator might offer an employer lower electricity rates for workplace charging in exchange for using the EV flexibility, or employees might get a monthly credit if their car participates in aFRR while parked). Ensuring alignment of incentives will be key to scaling up adoption, and literature could contribute by evaluating different schemes. In conclusion, further research should combine simulation and real-world experimentation. Simulations are invaluable to explore control algorithms and market scenarios at scale, while real-world prototypes and pilot implementations are essential to capture human factors, unforeseen technical hurdles, and true economic outcomes. A multi-pronged approach is recommended:

• Use simulation to test new bidding algorithms, baseline methods, and to perform sensitivity analyses (for instance, what if EV arrival times are wrong 20% of the time?).

• Simultaneously, run small-scale field trials in diverse settings (private company, public agency, car fleet, etc.) to gather data on user behavior and system performance.

• Iteratively refine the models with field data (improve, for example, the stochastic parameters using real plug-in patterns observed).

• Address the interdisciplinary aspects by involving regulatory studies and perhaps behavioral research (to ensure solutions are not just technically optimal but also user-friendly and compliant). By addressing these gaps, the next wave of research can provide a much clearer roadmap for turning the theoretical potential of workplace EV aFRR integration into practical, scalable, and profitable programs. The overarching goal is to validate that “netzdienliches Laden” can deliver on its promise: balancing the grid while saving costs for drivers and employers – and to identify and fix any weak links (be it baseline accuracy, user participation, or economic viability) that currently limit that promise.Sources:

• Doumen, S. & Paterakis, N.G. (2019). Economic Viability of Smart Charging EVs in the Dutch Ancillary Service Markets. Proceedings of SEST 2019www.scipedia.com.

• Rücker, F. et al. (2020). Evaluation of the Effects of Smart Charging Strategies and FRR Market Participation of an EV. Energies, 13(12), 3112www.mdpi.comwww.mdpi.com.

• Pavić, I. et al. (2023). Electric Vehicle Aggregator as an Automatic Reserves Provider Under Uncertain Balancing Energy Procurement. IEEE Trans. Power Systems, 38(1), 396–410orbilu.uni.luorbilu.uni.luorbilu.uni.lu.

• Goncearuc, A. et al. (2023). Incremental Profitability Evaluation of V2G-Enabled aFRR Services for Semi-Public Charging Infrastructure: Case Study in Belgium. World Electr. Veh. Journal, 14(12), 339www.mdpi.comwww.mdpi.com.

• Mödder, L. et al. (2023). Cost Optimization of a Vehicle Fleet Using V2G. In Antriebe und Energiesysteme von morgen 2022, Bd. 2. (Springer).

• Seo, M. et al. (2024). Strategies for EV bidding in the German FCR and aFRR market. Electric Power Systems Research, 228, 110040.

• Piffaretti, M. et al. (2024). V2X Suisse – Abschlussbericht (Swiss Federal Office of Energy)www.mobility.chwww.mobility.chwww.mobility.ch.

• Raue LLP (2017). Aggregatoren-Festlegung beschlossen – Summary of BNetzA BK6-17-046 and §26a StromNZVraue.comraue.com.

• Next Kraftwerke (2020). Lessons Learnt from Germany’s Mixed Price Systemwww.next-kraftwerke.com.

• Consentec GmbH (2022). Beschreibung von Konzepten des Systemausgleichs und der Regelreservemärkte in Deutschland – TSO report on balancing market designwww.elia.bewww.elia.be.