Introduction
In 2012, I installed a photovoltaic system at my home. Today, the system appears to be out of date: its peak power is just 2.35 kW (9 panels of 250 W each), and it lacks storage. Given how far the industry has advanced since then, this comes as no surprise. Over the last few years, I’ve wondered several times if and how I could upgrade the installation, but I have yet to find anything truly good. This is mostly because adding a storage system, the most obvious solution, would have required
- To change the inverter because the current one cannot manage batteries.
- To take on a considerable expense1.
Until today, the only improvement I’ve made is to join the energy community in my neighborhood — albeit as a consumer rather than a producer, despite the presence of the system2 — and reap the benefits.
However, I have realized that there are some new solutions on the market that can be really intriguing when mixed with legacy electrical systems like mine. In this blog, I will outline my experience in the hopes that it would be useful to those who are considering doing something similar. In particular, I’ll try to focus on some of the more subtle and less obvious aspects that — as I’ve learned the hard way — should be taken into account before making a purchase. By reading this post, you can save time on research and erroneous commissioning attempts.
First upgrade: installation of a plug-and-play storage system
The first upgrade I made was to install a plug-and-play storage system, namely the Zendure SolarFlow 2400AC+. These novel goods are based on lithium batteries, which are virtually always linked with new solar systems nowadays, but they have several distinguishing features:
- They include an inside microinverter since they are intended to be used mainly as retrofits for existing PV systems that lack storage.
- They exchange energy with the electrical grid through a normal outlet.
- They are modular in the sense that additional battery modules can be added to expand their storage capacity, even at a later time.
The fundamental goal of this addition is clear: to increase the self-consumption quota — that is, the ratio of the energy produced by the PV panels used by the household to the total energy generated by them — which has been about 30% over the plant’s 12 years of operation. To provide a more concrete idea of the quantities involved, I report some relevant data:
- The system has a yield of about 1100 kWh/year per installed kW, so in 12 months it produces about 2500 kWh.
- Of these, about 800 kWh are self-consumed.
- Therefore, approximately 2700 kWh are potentially available for further self-consumption, which until today have been fed into the grid2.
- Taking into account that the household consumes about 3200 kWh per year, if self-consumption reached 100%, only 700 kWh would be bought from the provider.
Setting it up
The physical installation is straightforward; basically, you just need to connect the storage system to a socket. However, be aware of some important caveats that might be overlooked when buying:
- The system is app- and cloud-centric; therefore, the availability of an Internet connection and the use of the app to configure and monitor it are crucial3
- Since the SolarFlow 2400AC+ only supports WiFi connections, make sure there is a WiFi signal at the location where it will be placed. Don’t take it for granted when it comes to residential buildings.
- There is a security and privacy issue to consider, too: if you must connect the device to the internal WiFi network (such as when the installation point isn’t reachable by a guest WiFi signal or you can’t set up isolated VLANs), it may access the internal traffic of the home, which should generally be avoided.
- To fully leverage the potential of the storage system, it needs access to two real-time measurements: the total energy consumption of the house and the energy produced by the panels. That’s why I bought a Zendure Smart Meter 3CT along with the SolarFlow 2400AC+. More on this later.
Physical connections
At the end of the day, the physical connections were made as illustrated in the following diagram.
It is important to emphasize that this is an atypical installation, in the sense that there is no direct connection between the panels — to be more precise, between the output of their inverter — and the SF 2400 AC+. This makes things a little less intuitive, as will become clearer later when looking at the app screenshots.
One fundamental thing to observe is the connection of the 3CT smart meter: it has 3 inputs as it can also operate with three-phase systems. In cases like mine,
- The IL3 input must be used to measure the power between the house and everything outside, that is, the electrical grid and the inverter of the PV panels.
- The IL2 input must be used instead for the output power from the inverter of the panels.
Pay attention to the direction of the current transformers: the red arrows indicate the correct orientation. If this is not followed, the software will not function properly.
In the diagram, there is also a 3EM meter from Shelly. Since I had it available, I took the opportunity to install it to monitor another variable, namely the power in the line towards the electrical grid. In this case as well, the orientation of the CT matters; otherwise, the power entering the house appears negative and the power exiting positive, which is contrary to the convention commonly adopted.
The house’s electrical box looks like this when the work is done:
Qualche delucidazione è d’obbligo per evitare fraintendimenti. Ci sono degli elementi che non sono legati a quanto descritto in questo post e che erano presenti in precedenza:
- RPi, USB/RS485 converter, SDM120: please refer to this old post.
- Digiwatt meter, Digiwatt’s CT: some years ago, I subscribed the service provided by https://digiwatt.energy/. By analyzing household consumption data with Machine Learning algorithms in the cloud, it should understand how energy is used by different appliances (like the refrigerator, washing machine, oven, etc.). I subscribed to see how the algorithm would improve over time, but I haven’t seen any significant changes to date.
Creating the logical system
After making the physical connections, the app is used to add the devices …
… and then you can create your own Home Energy Management System (HEMS0 in the following screenshots):
In the first screenshot, the 3EM is also visible. As mentioned before, this does not need to be part of the HEMS and in fact, it was not included in the definition of HEMS0:
I have nonetheless added it to the app so that I can monitor it without having to open the Shelly app as well:
The screenshot shows a negative power because, at the moment it was captured, the system was injecting 1168.8W into the grid.
Here is a screenshot of the Shelly app taken a few seconds apart:
The visualization of HEMS0’s energy flows (see the second screenshot) is useful for quickly understanding how the system works. However, in installations like mine, one thing could be improved for better readability: in the example, the power from the panels shows as 0, even though the screenshot was taken when they were producing over 1 kW. In the smart meter configuration, you must choose what the measured variable from the various inputs is:
It would be correct, therefore, that in the summary display of the HEMS, this is reported correctly, even if the panels are not directly connected to the SF 2400AC+ as in typical systems.
Operating modes
Here things get really interesting. The software allows for several operating modes to adapt virtually to all possible needs. In summary, these modes are divided into two categories: automatic and manual.
As soon as I received the SF 2400 AC+, I connected it to the home system without having installed the smart meter. Since I did not have the two measurements explained previously, I had to opt for one of the manual modes, specifically the one called “Smart Meter Mode.” In practice, you manually switch when you want the SF to charge the battery and at what power, and when you want it to supply energy, also determining the power in this case.
With regard to the power output of the SF 2400 AC+, it is important noting that, according to current regulations, the maximum power that can be supplied by default is limited to 800 W. To exceed this and reach the technical limit of 2400 W, the installation must be certified by an authorized professional (who, in the case of Italy, will issue a “Dichiarazione di Conformità”).
Auto Mode
After installing the smart meter, I switched to “Auto Mode,” which aims exactly to do what I purchased the storage system for: minimize the energy produced by the panels fed into the grid, that is, maximize self-consumption.
This mode aims for this goal by using a simple algorithm, as can be inferred by observing its operation throughout the day, which follows the rules shown in the screenshot above, without considering possible, future events.
After a week of use in this mode, during which there was only one cloudy day, these are the results obtained:
Comments:
- Due to the previously described “bug,” the energy balance of the panels is zero because their output is unfortunately not counted in this calculation. Consequently, it is not possible to see how much solar energy has actually been used to charge the battery. However, throughout the week, I periodically observed the feed-in to the grid during daylight hours and can confirm that the system is doing its job: the draw from the grid to charge the battery is practically zero, so of the 17.78 kWh injected into the battery4, it can be reasonably assumed that these come entirely from the panels.
- Regarding self-consumption, the displayed figure needs to be corrected. As seen in the screenshot, when I took it on Saturday, the battery was already charged, and the system hadn’t started discharging it yet because the energy produced by the PV system was more than sufficient to power the household loads. Therefore, the figure to consider is approximately 15.3 kWh (17.8 – 2.5 kWh) . Of these, 12.9 went to domestic consumption, which is equal to 85%. The value is undoubtedly good, but there is room for further improvement.
- Given the size of my system, certainly a 2.5 kWh storage does not allow me to make the most of the energy produced. However, I didn’t want to buy another 2.5 kWh module immediately to first verify that the product was suitable for my use case. Since I have no more doubts about this, I plan to increase the storage capacity in the future, but I’m still considering how because in the meantime, another idea crossed my mind.
ZENKI
Needless to say, the most intriguing automatic mode is the one called ZENKI, which is also based on the use of — guess what? — Artificial Intelligence. By combining this with the ability to understand consumption habits and weather forecasts, it is not unreasonable to think that even that remaining 15% of potential self-consumption could be exploited. However, I will only be able to find this out with time, having activated this mode on the same day I am writing this post.
Terms of service
To complete what has been said regarding privacy issues, I will include the agreement that must be accepted when installing the Zendure app.
Zendure AI Data Training Agreement
Introduction
This agreement stipulates how Zendure uses and processes customers’ historical personal data and device data to train artificial intelligence (AI) models, enhance energy management efficiency, and predict future energy usage and solar power generation trends.
In this agreement, “Zendure” and “we” refer to Zendure USA INC. “Customer” or “you” refers to users who use Zendure products or services.
Please read this agreement carefully and ensure you understand its entire content before using Zendure AI services. Once you start or continue using Zendure AI services, it indicates that you agree to our use of your relevant information in accordance with this agreement.
1.Data Usage Instructions
The customer explicitly authorizes Zendure to:
· Use the customer’s system historical energy data, geographic location information, and electricity price information to train Zendure’s AI models and conduct energy scheduling to enhance customer benefits.
2.Data Security and Confidentiality
Zendure commits to:
· Implement security measures in compliance with applicable laws and regulations， including the General Data Protection Regulation (GDPR), to ensure the security of the Customer’s data and prevent unauthorized access, disclosure, alteration, or destruction of data.
· Strictly limit data access permissions, granting data access only to necessary personnel responsible for data processing and AI model training.
· Not disclose the customer’s personal data and device data to any third party without the customer’s explicit consent, unless explicitly required by applicable laws, regulations, or judicial authorities.
3.Data Ownership
· The customer retains ownership of their original personal data and device data.
· Zendure independently owns all rights and intellectual property of AI models, predictive analysis results, data aggregation outcomes, and any derivative data results developed based on customer data.
4.Consent and Withdrawal
· The customer agrees and accepts Zendure’s processing and use of their data in accordance with this agreement.
· The Customer has the right to withdraw consent at any time by submitting a written request via email to support@zendure.com. Upon receipt of the withdrawal request, Zendure will promptly deactivate the AI mode for the Customer’s system (which will then revert to the default non-AI driven automatic mode) and cease any further use of the Customer’s personal and device data for AI model training and analysis. Following the withdrawal, Zendure will, within the limits of applicable law, cease the further use of the data.
5. Limitation of Liability
· If Zendure has fulfilled the reasonable security and confidentiality obligations stipulated in this agreement, Zendure is not legally responsible for any indirect, incidental, or derivative losses arising from the data processing and AI model training under this agreement.
6. Applicable Law and Dispute Resolution
· The signing, effectiveness, interpretation, performance, and dispute resolution of this agreement are governed by the laws of Germany.
· Any disputes arising from or related to this agreement should first be resolved through friendly negotiation between both parties. If negotiation fails, either party may file a lawsuit with the court having jurisdiction in Zendure’s registered location.
7. Agreement Updates
· Zendure reserves the right to update or modify the terms of this agreement at any time without prior notice to the customer.
· After the agreement terms are updated, if the customer continues to use Zendure’s services, it will be deemed that the customer accepts the updated agreement terms.
8. How to Contact Us
If you have any questions about this agreement or need further clarification, please contact us through the following methods:
· Email: support@zendure.com
· Official website:www.zendure.com
One final advice
If you are considering making an upgrade similar to the one described in this article, once you have broadly identified the physical parts you need, I strongly suggest that you create a mental image of the installation in order to try to foresee how and where you will place them, taking into account all the hardware elements (cables included) that you will need to understand:
- Whether the thing is feasible.
- What impact does it have in terms of interventions (addition of electrical sockets, wall drilling, cabling, etc.)
Future (possible) work
- Increasing the SF 2400 AC+’s output power to 2400 W.
- Integration with Home Assistant.
- Adding second photovoltaic system.
- Using the SF 2400 AC+’s backed-up socket to replace my old UPS unit powering my desktop computer.
I don’t promise I’ll do these task. Stay tuned and fingers crossed 🤞
- For the sake completeness, it should be mentioned that the system is still eligible for subsidies under the fourth “Conto Energia” program for another six years and has already more than paid for itself. ↩︎
- The “Conto Energia” provides an incentive for all the energy produced by the system, regardless of whether it is consumed on-site or fed into the grid. For the energy fed into the grid, there is an additional income from “selling” it to the utility, which then reuses it. However, the rate at which it is “sold” is significantly lower than the rate at which it is purchased from the utility. This is why maximizing self-consumption is generally the more economically advantageous option. ↩︎
- To address these issues, you might also consider integrating with Home Assistant to become cloud-independent, but I haven’t explored that option yet. ↩︎
- Also due to the “bug,” the app incorrectly indicates that this energy came from the grid. ↩︎
Credits
Featured image by Evgeniy Alyoshin on Unsplash.
