Stacking DC and the BM - exploring the data

On 27 January 2021, the rules for the Dynamic Containment (DC) service changed to allow for 'stacking' (i.e. providing Dynamic Containment and participating in the Balancing Mechanism (BM) at the same time). This change is great news for the energy storage sector and opens the doors to more complex optimisation in the future, but how much value is really being captured?

Now that assets have been stacking for a few months, we decided to explore the data to see what's really been happening.

In this piece we explore:

What is DC-BM stacking?
Which optimisers are stacking and how are they doing it?
What's the commercial case for stacking?

Spoiler: DC-BM stacking is mostly being used as a state of charge (SoC) management tool (for now).

What is DC-BM stacking?

DC-BM stacking is the participation in both DC and the BM at the same time. For clarity, here's a recap of those services:

Dynamic Contaminant (DC) - Assets provide fast (sub-second), post-fault frequency response, to maintain grid frequency.
Balancing Mechanism (BM) - Every settlement period (30 mins), assets send their bid/offer prices to National Grid Electricity System Operator (NG ESO). If NG ESO needs the asset to ramp up (export) or ramp down (import), it can choose to accept the bid or offer and the asset must respond.

Stacking sites can secure a DC contract to deliver frequency response at the same time as receiving bid-offer acceptances (BOAs) in the BM. Frequency response is then delivered around the BOA (see Figure 1).

Importantly, providers cannot offer the ‘same MW’ to deliver both services simultaneously. Instead, DC-BM stacking lifts the restriction on receiving BOAs whilst providing DC.

Graph%20showing%20indicative%20delivery%20of%20BOA%20and%20frequency%20response — Figure 1 - Indicative power delivery of frequency response around an accepted bid-offer acceptance.

How much DC-BM stacking is going on?

First, let's see how much of the DC market is BM registered.

Figure 2 shows the eligible capacity in the DC market by market participant. As of 1 July 2021, there is 878 MW of capacity eligible for DC participation, 551 MW of which is BM registered (63%).

Note: all sites currently participating in DC are battery energy storage systems (BESS).

Graph%20showing%20eligible%20DC%20capacity — Figure 2 - Eligible DC capacity split by market participant and BM registration status.

What kinds of volumes are being accepted by DC providers in the BM?

For BM registered assets, we have full visibility of physical positions taken in the market through physical notifications (PNs) and BOAs. This allows us to understand both the prices and volumes associated with stacking.

Figure 3 shows the total BOA volume for (BM-registered) sites with DC contracts. Since January 2021, 99% of BOA volume from assets with DC contracts has been through bids (i.e. assets buying power) rather than offers.

Graph%20showing%20DC-BM%20stacking%20volume — Figure 3 - Total BOA volume from assets under DC contract. Source - BMRS, NG ESO Data Portal.

Why is it almost all bids?

At the moment, DC is exclusively an asymmetric service - participants only provide low-frequency response, by exporting/discharging power into the grid. Since providers must always be able to deliver their full contracted capacity in DC, this prevents them from offering to deliver power through the BM. The data shows that DC providers are not taking arbitrage positions - they have no means to sell the energy without compromising their DC delivery.

It would appear from the data that BM-DC stacking is currently used as tool to manage SoC (i.e. to recharge following the energy losses of providing DC).

How much value are optimisers securing through DC-BM stacking?

If DC-BM stacking is mainly used for managing SoC, how much additional value can it provide vs. simply taking the imbalance price?

Since the introduction of stacking, there are two ways to manage SoC in DC:

Baselining - Posting PNs to import power, exposing the asset to either the imbalance (system/cashout) price or intraday/day-ahead prices (if traded).
DC-BM stacking - Posting bids (and receiving a BOA from NG ESO) to import power in the BM, taking the price of the accepted bid/offer.

In both cases, the goal is the same: maintain SoC at sufficient levels to provide DC, whilst minimising the costs.

What's best for managing SoC: Baselining or DC-BM stacking?

To answer this question, we have analysed the costs of:

Baselining - Using system price for each settlement period.
DC-BM stacking - Using actual accepted bids/offers by DC assets (i.e. real operational data).

Figure 4 compares the two strategies for managing SoC.

The yellow lines represent the cost of managing SoC using baselining (and taking the system price).
The blue lines show the costs of managing SoC in the BM (stacked with DC).

For the avoidance of doubt, the BOAs in this analysis are accepted BOAs for real assets with DC contracts (i.e prices that were actually accepted). Similarly, system price averages are calculated using PNs for real assets under DC contact.

Graph%20showing%20the%20charging%20price%20of%20stacking%20vs.%20baselining — Figure 4 - Daily volume-weighted average charging prices for acceptances of assets with DC contracts. System price average is calculated as the volume-weighted average of system prices during periods where DC contracted assets posted PNs. Similarly, BOA price average is calculated as the volume-weighted average of BOA prices for periods where DC contract assets received BOAs.

When comparing average charging prices, assets charging in the BM are securing prices that are on average £58/MWh cheaper than those exposed to system price. Also, average accepted BM charging prices (i.e. the BOAs accepted for assets with DC contracts) have historically been negative 48% of the time. This means BESS assets can be paid to recharge.

Despite the benefits of using the BM to charge, the energy throughput (and losses) for providing DC are relatively low, so the energy required for SoC management in DC is also relatively low (see here for more info).

How much does the charging price actually impact the cost of service delivery?

Figure 5 shows the average daily revenues and costs associated with the DC market, including the SoC management costs associated with baselining and stacking.

Graph%20showing%20stacking%20vs.%20baselining%20as%20a%20revenue%20comparison — Figure 5 - Average costs and revenues of BM-registered sites under DC contract. Stacking costs reflect the average spend of assets pursuing baselining and DC-BM stacking SoC management strategies.

Comparing the daily cost of stacking and baselining, participants could save an average of ~£14/MW/day by pursuing a DC-BM stacking strategy for SoC management. This represents an uplift in DC revenues (net of SoC management costs) of 3%.

These cost savings (summarised in Table 1) are relatively small when compared to the £408/MW/day value in DC, but this is skewed by the high prices in DC (£17/MW/h). This will not last forever. When BESS buildout catches up to procurement targets (see here for Modo’s future BESS buildout report), DC prices will fall. When this happens, reducing SoC management costs will be more important for DC participants.

Table%20showing%20the%20average%20charging%20price%20and%20SoC%20management%20cost%20of%20stacking%20vs.%20baselining — Table 1 - Stacking vs. baselining summary. Includes average charging prices (Figure 4) and average SoC management costs (Figure 5).

How are things going to change?

Over the next 18 months, there are numerous changes coming to frequency response (for more info see here), most notably the introduction of symmetric DC. In symmetric DC participants will be required to both inject power (for low-frequency events) and absorb power (for high-frequency events). While the exact details are still subject to change, DC-BM stacking will likely be impacted in three ways:

Reduced requirement for SoC management - With symmetric frequency response, high and low-frequency responses will (on average) cancel each other out. At this point, any SoC losses will largely be driven by the round trip efficiency of BESS. This will significantly reduce the cost associate with SoC management, irrespective of strategy.
Stacking high-frequency DC with offers - When high-frequency DC (DC-HF) launches, providers can choose to only deliver high-frequency response (absorbing power from the grid). In this scenario, providers will be able to stack DC-HF with offers, but not bids.
Stacking BM with the symmetric service - For market participants providing symmetric DC (both HF and LF) the future of BM-stacking becomes more uncertain. Since BM activity cannot compromise DC delivery, under the current guidance, participants would not be able to participate in the BM at all. This is something to keep an eye on over the coming months (but we’ll keep you updated as the service develops).

Roundup

As of 1 June, there is 878 MW of eligible DC capacity, 63% of which is BM-registered and capable of DC stacking.
BM arbitrage is not possible (for now) alongside DC delivery since the sale of power prevents assets from meeting their DC obligations.
DC-BM stacking is a tool for SoC management and opens up a new market in which to import power.
Charging using DC-BM stacking is preferable to baselining, with lower average prices and more instances of negative prices. Together, this can save providers ~£14/MW/day on SoC management costs.
Cost savings are low compared to the revenues earned from DC participation due to the low throughput of DC. However, when DC prices fall, the cost of SoC management will become more important to net revenues in DC. This will create a stronger commercial incentive for BM-DC stacking.
The future of DC-BM stacking is somewhat uncertain with changes to the procurement methodology over the next 18 months.