Why controlling residential solar inverters is more complicated than it appears

Last month Tom Geiser published an article about why small-scale solar PV system owners should embrace control to enable greater responsiveness to wholesale spot market prices on the NEM.

SwitchDin exists to help better integrate distributed energy resources (DERs) like solar inverters & batteries into the grid; we are working with networks, retailers and manufacturers to do this. Tom’s article was naturally of considerable interest to us.

We absolutely agree that coordinated control of PV inverters is a good thing for both solar system owners and the broader energy system alike. Establishing control of inverters and other DER will be crucial in maximising the benefits of solar for everyone; many of Australia’s DNSPs are investigating ways to make control a win-win proposition as an alternative to some of the blunt tools currently available – system size & blanket export limits.

We’d like to pick up the conversation to elaborate on this topic from our experience integrating with nearly two dozen brands of inverters.

Tom wrapped up his article speculating that writing a controller for an inverter should be easy:

“A controller would be pretty easy to set up. It needs to speak to the inverter and check spot price data. Ideally it would also be able to send data to AEMO on intentions, potential power, and actual power.”

Tom suggests that writing the control algorithm would be easy – but from experience we know that making something like this work across a diverse fleet of products in real life is hard.

The real-world challenges of inverter integration for control

SwitchDin has a team of engineers whose job is translating inverter protocols for our Droplet controllers with the aim of standardising comms between various inverter brands. Here is a list of some of the challenges that they’ve faced.

  • There is no national or international standard that defines how inverters must communicate with a HEMS controller, the DNSP or the market. Different inverter manufacturers use different communications protocols. Most of these are Modbus and Sunspec, but these include proprietary aspects and there are also a number of fully proprietary protocols out there as well (Victron, for example). There are also others like DNP3, which is commonly used by utilities & billing-grade meter manufacturers.
  • Even within widely used protocols of the same name (Modbus, Sunspec), there are substantial variations in implementation, meaning that each one needs to be mapped out and tested separately for communications & control via an external device (or cloud platform) to be possible.
  • On top of this, the publicness and availability of these protocols varies dramatically from one manufacturer to another. SMA, for instance, publish their Modbus maps on their website, while others hold them tight to their chests, sharing them only with commercial partners – or with no one (such as Tesla). Other manufacturers will provide them on request, but frequently this involves first identifying and getting in touch with the right team member, which itself can be further complicated by language barriers or dealing with the bureaucracy of large corporates. Still others may not have any documentation to speak of (which can be especially frustrating). These issues are probably a good indication of the relative immaturity of this technology market.
  • Certain control registers for some inverters have a limited number of write cycles (particularly in cases where registers are located in EEPROM as opposed to RAM memory); continuing to send control instructions to the inverter after this threshold is reached can literally break the inverter and cause it to stop working, potentially leading to voided warranties and angry customers. Although some manufacturers are starting to make this limitation explicit in their documentation, often the only way to know for sure is to ask. Even then local support teams often just don’t know as it’s not something they have been asked.
  • Different manufacturers offer varying levels of control, which makes implementing a global, tech-agnostic controller a big challenge.
  • Even where controls are accessible through integration, these control options may not be standardised. (This is especially true of hybrid inverters, but also for PV-only inverters to a lesser extent.) For example, when limiting the export power, some inverters provide an export limit that operates at the inverter’s terminals, while others provide export limits that operate at the grid connection point for the site. In some cases inverters do not have the ability to do either, or at least no clear path to triggering the controls functions that may sit in their registers. In addition to identifying the control levers in each inverter, it’s therefore also necessary to characterise the controls to ensure a consistent behaviour across different inverter brands in response to an event. For instance, an inverter’s control settings may take the form of tariff optimisation modes, without access to the individual controls.
  • Inverters that appear to be identical – even down to the model number – can have different firmware versions. And on top of this, depending on the manufacturer & model, updating the firmware version can be a manual process that requires physical access to the inverter and the right tools to carry out an update; this is in direct conflict with the many of our clients’ aims to reduce service calls in the field. On the other end of the spectrum, there could be remote firmware updates that happen without the controller developer’s (e.g. SwitchDin’s) knowledge that can in rare instances ‘break’ the control chain of command.

A big part of our job at SwitchDin is working with the manufacturers to knock down as many of these barriers as possible. This requires not only in-depth technical expertise, but also close, collaborative relationships with the manufacturers to get to the bottom of the less obvious issues.

Emerging trends: Inverter cloud APIs & standards

It’s worth pointing out that all of the above challenges have to do with integrations on the hardware/firmware level; cloud API integrations are an alternative approach. However, with cloud APIs there is currently even less standardisation than with hardware, and additional fees may apply. The use of cloud APIs across multiple vendor platforms also has implications for consistency in reliability, latency and cyber security, as well as data sovereignty (depending on where the vendor’s servers are hosted).

Another approach is to build a framework for standardising inverter control going forward – but at present there is no such framework in Australia. In a policy vacuum (but with a view to what’s coming down the pipeline) SwitchDin has been endeavouring to fill this gap through our work with DNSPs and our engagement across a number of harmonisation activities taking place in the sector. This includes our work with Horizon Power, where we’ve implemented an IEEE 2030.5 client on our Droplets with rollout commencing in the Onslow DER project. This is the first implementation of this international standard in Australia, which SwitchDin is leading efforts to localise.

Why not retrofit them all?

Another important point here is about retrofits. There are already 2.2 million systems installed in Australia, which would mean a lot of retrofits for any real or hypothetical controller (n.b. SwitchDin also has a real one, which we call a Droplet). Retrofits are an especially fraught endeavour due to the sheer amount of variation in brand, model and age – not to mention the lack of a standard for them to adhere to.

The vast majority of the millions of systems already in operation will continue operating away – without any smart curtailment – until they expire. The best approach to most of these is to let them die a natural death – although a case could be made for choosing a selection of popular models that are identified as low-hanging retrofit fruit. For the others, the costs of retrofitting them all would likely outweigh the benefits – to both the owners and the energy system as a whole.

DER control goes beyond wholesale prices

The problems for high penetrations of rooftop solar are not isolated to the fluctuations of the wholesale electricity spot market or system frequency – there are different layers of the energy system which affect one another. The NEM level is important for electricity retailers, generators and aggregators, but the more pressing challenges in Australia are arising on the distribution network level, and this is where much of the interesting work around DER is happening.

For distribution networks, high levels of PV penetration can cause power quality issues and put stress on distribution transformers. In some cases, an inverter (or battery storage system) may be controlled for network reasons before NEM spot prices even come into the picture. Ultimately the goal is to have coordination between both of these layers; determining how to implement this coordination is the goal of the ARENA-backed evolve project (which SwitchDin is a part of). Industry bodies like AEMO & AEMC are exploring how a network services marketplace could fit into this type of framework.

And crucially, this type of DER coordination requires control of prosumers’ behind-the-meter assets, which brings about a whole slew of issues around fairness and equitability over how, when and how often inverters & other assets are controlled by utilities & retailers. It also requires another level of ‘fleet-wide’ management of these assets. All of this complicates the challenge of orchestrating assets

A work in progress

Australia’s energy system is evolving to better incorporate DERs like distributed solar PV, but we’re still at the beginning of the journey. One of the biggest challenges is making prosumer-owned equipment ‘fit’ into the everyday operation of the grid’s physical infrastructure and financial markets. With the huge amount of diversity in PV and BESS inverters (not to mention other types of DER), there’s a lot of work to be done. We’re happy to be one of the companies helping to move the system in the right direction.


About our Guest Author

Joseph Kassouf is a solar engineer with nearly a decade of experience in the Australian renewables industry. His former roles include application engineer, technical sales engineer and product manager for ABB and Power-one covering residential, commercial and industrial products. He is the Product Manager for Newcastle-based energy software company SwitchDin.

You can find Joseph on LinkedIn here.

4 Comments on "Why controlling residential solar inverters is more complicated than it appears"

  1. Most houses already have a built-in remote controlled switching device – the hot water service ‘ripple switch’.

    Worked well for decades.

    The situation with rooftop PV is worse when you consider a black start situation. Imagine having to wait until night time before restarting the grid. Ouch.

    • Yes until we have a level playing field with electron tenderers to the communal grid being mandated to reasonably guarantee them along with FCAS 24/7/365 or keep them this farce will continue. With up to a third of electricity used to heat water for the home it makes economic sense to use rooftop solar for heating electric mains pressure storage HWS with solar diverter controllers like so- https://www.paladinsolarcontroller.com.au/
      Any surplus can then be used to RC heat and cool the home for comfort after nightfall. The unreliables problem is political not technical but it will become increasingly apparent with the demise of coal and with it off peak power pricing.

  2. From a policy perspective, we would want to see any price or control signals from the system or grid operators being set at the consumer front gate. While solar is a challenge at present, the future challenges will resolve around EVs, storage and other in-home devices. The consumer (or their agent) should be the one who decides which appliances respond to these signals. Unless there is a technical reason, the system and grid operators should not be reaching in a operating individual consumer appliances.

    • Anthony, yes to your sensible points. Central control of millions of individual inverters? There has to be a better way. Some would argue the ‘better way’ was several large generators close to load centres with minimum transmission.

      The dilemma is that we are achieving lower grid utilisation by subsidising rooftop PV, while at the same time building more grid by subsidising wind and solar farms, renewable energy zones and interconnectors.

      I can’t see a way out of the inverter problem except retro-fits.

      But properly designed tariffs could help, especially if a penalty tariff is applied to bigger systems and those on a premium feed-in-tariff.

      But if owners opted for a network optimised battery, they’d avoid the penalty tariff and receive a network support tariff.

      That would help all consumers by avoiding distribution upgrades.

      Then EV chargers – I think you have to put them on the ripple switch anyway.

Leave a comment

Your email address will not be published.