Ramp Rates for BESS Units and Voltage Change Limits – What is Credible?

Background

A key question that has been causing some disagreement between DNOs and BESS operators is how to consider BESS units in terms of ramp rates, and the impact to the Step Voltage Change (SVC) and Rapid Voltage Change (RVC) for P28 compliance.? This is not a simple / clear answer and needs some detailed thought. Many DNOs impose limits on ramp rates and operating modes, which can be problematic for developers, but it is equally important to consider that large power swings on the host DNO network could cause problems for the DNO or other customers.

BESS units are highly flexible, and trade in a number of different platforms, such as Frequency Response, as well as Balancing Mechanism and Intraday trading. The general premise of all BESS units, is to buy electricity when its cheap and sell it when its expensive in relation to market signals.

DNO Voltage Change Limits (P28)

The DNOs have to follow the limits defined within ENA P28, and therefore BESS units must comply with these values. The standard contains two types of limit 1) Rapid Voltage Change (RVC), associated with fast events and Step Voltage Change (SVC) associated with slower events.

BESS units providing frequency response fall somewhere between these two categories; they can provide rapid response in relation to system disturbances when in a frequency response mode, or they operate at a slower response to routine changes in frequency, or a flat response if trading in different (non-frequency markets). For a BESS, the RVCs are generally less problematic, as they can be managed by correct configuration and tuning of the various loops within the Power Park Controller (PPC).

The issue with BESS approval and compliance and approval from the DNO tends to therefore relate to staying within the +/-3% limit imposed by the P28 standard. This limit is imposed for slightly muddy historical reasons, but it is currently assumed to be valid and the DNOs are unlikely to agree to move from it. The are certain cases, which allows up to +/-6%, but these are for the DNO, are generally reserved for considering multiple sites and risk of coincidence events.

BESS Providing Frequency Response Services

BESS units often provided frequency responses (FR) services, such as Dynamic Moderation, Dynamic Reserve (DR) and Dynamic Containment. The general idea of all these services is that the BESS tracks the system frequency and alters its export / import behavior to match. The profile of these services can be seen in the below diagram. DR / DM are designed to track within the nominal frequency range to help maintain the system frequency, whilst DC is design to respond to major disturbances. Whilst not explicitly required, many BESS units trading in FR services will tend to operate at unity power factor to limit the MVAr flow into the DNO network, and therefore limit voltage disturbances.

From the below graph it is easy to see that if the frequency moves around a lot in the 49.8Hz to 50.2Hz range then the BESS units providing these services will work more. While if the frequency remains tightly controlled the BESS will operate less.

The concern about how a BESS behaves during frequency response, has been one of the key discussion areas over the last few years. It is therefore useful to take a step back and consider how NESO manages the system frequency first and what limits are imposed.

It is important to note that the system frequency in the UK mainland is not fixed, it ranges between a number of defined boundaries imposed by NESO. The actual frequency changes throughout the day as the balance between load and generation shifts and generation is added and removed from the system. The question is therefore:

What level of system frequency disturbance can occur on the system, and do they occur often enough to drive a BESS from full export to full import or vice versa?

?

System Frequency Management and Risks

To understand the above question, we need to understand how NESO manage and control the system frequency. Practically, NESO contain the frequency around in the 49.8Hz to 50.2Hz range. Sometimes the system frequency runs high for a while, sometimes it runs low for a while. Sudden changes in frequency are usually due to unplanned events. A typical morning (yesterday) is shown below – with data take from Elexon. Note that this is a low resolution snapshot, actual with 15s samples – more accurate historical data can be found from NESO. ?

?

NESO produce a key document called the Frequency Risk and Control Report (FRCR) Response Control, which can be found here. There are a few key features to be understood.

• The system frequency is operated within a nominal range of 49.8Hz to 50.2Hz
• The formal frequency limits are 49.5Hz to 50.5Hz
• Under frequency load shedding kicks in at 48.8Hz
• Frequency risks are categorised on probability
• Large frequency events are caused by sudden loss of large generation, load or HVDC interconnectors
• NESO assumes the largest loss to be 1800 MW and the maximum RoCoF to be 0.5Hz/s
• Historically the inertia was 140 GVA.s, but this was dropped to 120 GVA.s in 2022 and is planned to be dropped to 102 GVA.s soon.
• Over-frequency events are much lower risk and less common in the UK

??

Frequency Disturbances

The amount of volatility in the network depends also on the network load, system inertia, and other events that may be occurring. Typically cold / winter days have more disturbances and can become more volatile as the headroom gets used. ?Conventional Thermal and Hydro plants are brought on & off line to match required demand and meet any shortfalls of renewables, but also to maintain the minimum system inertia. This can often be the reason that some renewable generation is constrained off, there is an excess of wind / solar that is not needed, as NESO have to keep on certain thermal generation to meet the system inertia.

For those with a more technical persuasion the magnitude of the system frequency disturbance can be estimated with the swing equation, shown below in different formats. Where the system inertia in GVA.s is shown by the parameter I.

Doing a quick bit of maths, some typical RoCoF rates for system loss events can be seen below. A trip of (for example), the IFA1 or IFA 2 link would typically be around 1GW< so leading to a RoCoF a bit above 0.2Hz/s, depending on the actual inertia at the time.

• For a 120GVA.s -> dP = 960MW loss -> 0.2Hz/S RoCoF
• For a 120GVA.s -> dP = 480MW loss -> 0.1Hz/S RoCoF
• For a 102GVA.s -> dP = 816MW loss -> 0.2Hz/S RoCoF
• For a 102GVA.s -> dP = 408MW loss -> 0.1Hz/S RoCoF

?

What Does that all Mean?

The above sections are all a bit technical, so pulling things together in a bit more detail we need to think about what could actually happen in terms of system disturbance and how it could effect the P28.2 limits. A key point that often gets missed in this debate is the impact of ramp rates – historically many consultants have focused on a 1s ramp from export-import and import-export, as this is conceptually easy to consider. Practically however, the ramp rate doesn’t really matter too much for the P28 SVC limits (±3%), and faster / slower ramps just contain different transients.

?The problem with P28 compliance problems for SVC events, is that we can have events with a much lower RoCoF that could cause the BESS to swing from one extreme to the other and cause voltage regulation problems. DNO tap control relays are not particularly quick acting - and are set somewhere between 45s to 90s, so any change of frequency within this bound can cause a problem for the SVC compliance.

Consider an example situation where the NESO system is sitting a bit high – lets assume a value of 50.2 Hz for ease of calculation. The network experiences an unplanned loss of 400 MW (a fairly common occurrence),?giving a RoCoF of around 0.1Hz/s. The frequency could then drift from 50.2Hz to 49.8Hz, over a period of 4s, and still be within the NESO limits, and NESO would not be overly concerned as the system frequency is staying within nominal range. ?Similar smaller events could also occur ?taking perhaps 10-20s to do so.

In summary, while a ramp of export-import (or vice versa) over 1s is not credible; the same ramp over a longer duration of 10s-30s is credible and would have the same net effect on the SVC limits for the DNO, as the transformer tap changes could not correct it in time.?

?

BESS Trading in Non-Frequency Response Markets

As well as FR services, it is also important to consider that a BESS will operate a lot in trading throughout the day - moving through different ? hourly periods. BESS units are totally free to move between import and export depending on what the market is doing, and in some cases can stack various markets. These are inherently more volatile trading sessions, designed to meet short terms changes in demand and supply.

When this is considered, ?it is likely (almost certain in fact), that there will be occasions when the market prices changes in a way that the BESS wants to go from one mode to the other between trading blocks. So this would see significant power swings - most likely at times when the network is already a bit stressed i.e. morning / evening peaks of following an unplanned outage.?These power swings would therefore be routine and expected to occur regularly within the year. Practically this is more likely to be a more likely occurrence than large frequency disturbances occurring on the system. What further complicates matters for BESS units, is that when trading in these markets they may be operating in a specific power factor (non-unity) leading to larger SVC occurring.

A further complication of the ? hourly trading markets, can occur there are a number of key issues to consider. All trades occur in the same ? hourly block, and there is a requirement to start trading within a target time of 60s. If multiple BESS units are responding to the same market signals, then there is also the possibility of coincident swings occurring at once, stressing any DNO network.

To illustrate the above point, a sample ? hour trading price is shown below (taken from Nordpool). The plot shows a number of steady and small changes in prices, but also some sudden large step changes such as 6am, where the price jumps from ￡40/MWh to ￡80/MWh – which could well have triggered a jump from import to export.

NESO are also actively considering moving to 15 minute trades, and possibly to 5 minute trades, as used by other countries like Australia.

?

?

Solutions

The obvious question to be considered when faced with an SVC issue, is “how do I fix it?” Unfortunately this is not quite so simple to answer. The SVC disturbance in a system is primarily a function of the DNO system strength and the MW & MVAr interchange between the BESS and the DNO. Operating the BESS at unity power factory significantly reduces the magnitude of the voltage disturbance on the host DNO, and should be a first option considered. If this is not possible then the options become more limited, and as noted earlier, simply slowing the ramp rate down won't help much.

A number of DNOs are open to the consideration of voltage control, and this is an approach we would recommend. The use of a voltage control approach allows the BESS to produce / consume reactive power on the network, and can help limit voltage disturbances significantly. However, there are a number of issues with this approach and it is not as simple or risk free as it appears. Scotland use this control method widely, but have been having an increasing number of problems on voltage stability and Sub Synchronous Oscillation (probably) because of it. At a simple level, it is not difficult to see problems occurring where there are multiple BESS units all in voltage control on one BSP / GSP, as the units could potentially start 'fighting' with each other and creating oscillations.

It would be reasonable for a DNO to consider one BESS providing voltage control services for a node, but this would have to be determined and calculated in an equitable manner, such that one BESS was not unfairly treated against others. Mechanisms would be needed to allow a DNO to move a BESS in and out of voltage control mode depending on its availability and the operation of other BESS units in the area. This is technically not difficult, but practically most DNOs do not have the control infrastructure to manage it.

Summary

It has been shown in this article that BESS units can interact with the host DNO network in a number of different ways, which can cause significant voltage disturbance issues. The traditional method of assessing a disturbance by assuming a full power swing from import to export, or vice versa, over a 1s period, is not strictly correct, as it assumes a large frequency event has occurred. Instead consider a ramp over a longer duration of 10-30s would be more reasonable – however the net result for the P28 SVC limit would be the same as the DNO transformer tap controllers would still not operate within this window.

It was also shown that BESS units trading within various ? hourly markets could be the more onerous and routine case that needs to be considered for voltage control issues. These events happen 24/7/365 and due to the nature of the trading system would occur in a coincident manner, potentially stressing key DNO substations. How this is managed and assessed will be a key headache going forward, and the use of some level of voltage control is likely to be needed.

At present the P28.2 standard does not cover this issue well enough - but work is ongoing to provide clearer guidance on this issue in EREP P28, which we have been providing key input to.

An obvious solution to the problem is for DNOs to start considering voltage control of a BESS unit, but this has its own risks and cannot be seen as a simple solution to the network problem, otherwise it may create other issues. If you want to know more, please get in touch.