Renewables Market Development
Due to the increasingly serious energy crisis and environmental pollution, the utilization of renewable energy resources have become the only viable solution for ensuring secure and sustainable energy supply. In recent years, the installed capacities of renewable generations such as wind power and photovoltaic are rapidly increasing, transforming the power generating resources from predominantly fossil power generation to high penetration of renewable generations [ref].
Wind and solar photovoltaic technologies are expected to continue their growth trajectory as generation costs involved with these sources are becoming more competitive, thereby encouraging more projects and customers to be set up in the coming years. The emergence of newer markets will further drive uptake for wind and solar energy sources, as it can be seen in the following graph [ref].
In recent years, it is commonplace in most developed countries under the Paris Agreement for the governments to provide subsidy incentives on consumer solar installations, so that countries and social housing corporation solar installations meet their CO2 reduction targets. Thus, it has become more and more attractive to environmentally sensitive customers to become prosumers of electric energy (consumer + producers) by installing photovoltaic panels on their roofs or gardens.
How Photovoltaic (PV) Panels Work [ref]
A PV module is an assembly of photovoltaic cells mounted in a framework for installation. Solar PV cells are made from layers of semiconducting material, usually silicon. Photovoltaic cells use sunlight as a source of energy and generate direct current electricity. More specifically, when light shines on the material, electrons are knocked loose, creating a flow of electricity.
A collection of PV modules is called a PV Panel, and a system of Panels is an Array. Arrays of a photovoltaic system supply solar electricity to electrical equipment. Modules and arrays come in a variety of shapes and sizes. Most PV systems are made up of panels that fit on top of the roof, but they can also be installed on the ground, or fit solar tiles.
The electricity generated is direct current (DC), whereas the electricity used for household appliances is alternating current (AC). An inverter is installed along with the system to convert DC electricity to AC.
There are several major factors that affect the Solar Production of the installed PV module:
The quality and the capacity of the installed solar panel/array. As it is natural, the larger the capacity of the installation, the more electricity is going to be produced. However, important roles also play the material and the connection of the photo-voltaic cells.
The weather conditions. The cells don’t need direct sunlight to work, they can work on a cloudy day. However, the stronger the sunshine, the more electricity generated.
The photovoltaic panel’s pitch angle towards the sun. The roof space should ideally face south, unshaded, and at a pitch angle of about 30 or 40 degrees. East- or west-facing roofs could still be considered, but north-facing roofs are not recommended.
The surrounding environment. Any nearby buildings, trees or chimneys may be potential objects that could shade the area that the PV module is installed, which will have a negative impact on the performance of the system.
The added benefits for a prosumer that has an PV module installed are the following:
Cutting electricity bills. Even though the installation of a photovoltaic module/array installation requires an initial capital, this investment will reduce the electricity costs significantly. One is able to find out how much you could save by using the Solar Energy Calculator.
Cut your carbon footprint. Solar electricity is green renewable energy and doesn't release any harmful carbon dioxide or other pollutants. A typical residential solar PV system could save around 1.3 to 1.6 tonnes of carbon per year.
Worthwhile Investment. The installation of a photovoltaic panel is one of the main contributors for improving your housing energy label, which leads to an increase in the value of your home. Additionally, in most countries, there are net metering and tax incentives towards that direction, paying a portion of the initial investment. To this end, investing in solar is much more attractive than having money on your savings account.
Daily Solar Production Examples
We have gathered a few example figures of solar production from different types of days, in order to provide an overview of real-life production cases. For presentation purposes, an installation in Central Europe with a solar capacity of 2000 Watts was selected.
Spring day, with no clouds
No residents present
Daily Solar Production: 14 kWh
Peak production: appr. 1600W
Summer day, with clouds
Heavy appliance usage during daylight hours
Daily Solar Production: 11 kWh
Peak production: appr. 2000W
Cloudy Winter day
Heavy usage of appliances during daylight hours
Daily Solar Production: 2 kWh
Peak production: appr. 400W
Contribution - Solar Production Monitoring Cases
As mentioned above, it is important for a customer that has invested in a solar panel installation to be able to monitor his/her solar production on a daily basis or even in real time. Having this information available enables (a) better comprehension on how and when the installed solar panel functions and produces energy, and (b) taking preliminary actions against maintenance or installation problems that may arise in the future, based on the metrics of historical efficiency of the installed panels.
In case an energy service company wants to provide information about solar production to its customers, there are actually three ways that allow for solar production monitoring, each one having its pros and cons:
Retrieving information from the Inverter using Software. Most of the solar panel providers are willing to provide (near) real-time measurements from the solar panel production. The most common information provided is the solar production level (in Watts) at a 1/10/15 minutes granularity. However, in order to retrieve this kind of information, a software connection with the API of the installed inverter needs to be implemented. The good thing about this service is that you have a very accurate depiction of the actual solar production. On the other hand, there are too many models and types of inverters out there, each one having its own API / App, making it near impossible to create a unified approach to retrieve data information from all of them. Also, the time granularity plays an important role in removing the solar production time series, in order to retrieve the actual consumption time series of the installation.
Installing a second metering equipment. In this case, another metering equipment has to be installed directly in the solar panel inverter, providing information in real time. This case is perfectly accurate, since the metering equipment retrieves only the solar production measurements and the time granularity is controlled from the metering equipment, which in most smart metering equipment is 1 second. However, as it is obvious, this solution is more intrusive and costly than the first one.
Implement a software solution on the Analytics side. The final approach for extracting useful information from the solar panels is to create a smart service that is capable of extracting the solar production time series using the NET consumption measurements available from the main smart metering equipment. This is a non-intrusive approach, without any additional cost on the customer side. The only drawback is that the results may not be perfectly accurate, since the prediction error is dependent on the quality of the implemented smart solution and the retrieved data measurements.
NET2GRID Solar Production Services
At NET2GRID, we made the strategic decision to go with the third approach. To this end, we have implemented an intelligent, non-intrusive, inverter-agnostic service which provides an overview of an installation’s solar production on a daily basis in a very cost-efficient way. Our approach utilizes a number of state-of-the-art AI techniques in order to estimate the actual solar production time series, without using additional metering equipment or utilizing the inverters’ API, while at the same time resulting in a mean accuracy of 90% on our solar production predictions. Additionally, we are supporting solar production on a total installation level, meaning that the approach is independent of the amount of installed panel capacity or having multiple different brands of inverters (apps) being installed over time, since, lately, it is really common for people to continuously extend their solar investment.
Our algorithm utilizes the following inputs:
The installation’s Net daily total consumption time series (the Net time series of an installation represents the overall time series as it is retrieved from the Grid, meaning the time series that can be acquired by subtracting the solar production from the total consumption of the installation), as it is retrieved from the smart meter equipment of the installation, in 1-second granularity. In case of a three-phase installation, providing all three phases can make the algorithm more robust and accurate.
(Optional) The weather conditions that are present at the time of analysis. More specifically, our approach utilizes measurements from Solar Irradiance, Temperature, Cloud Coverage, in case they are available.
(Optional) The historical values of the solar production time series from days that have presented similar weather conditions in the past month.
In order to estimate the solar production time series as accurately as possible, the following steps are applied on the inputs:
A time-series analysis and prediction algorithm in order to make a first rough estimate of the solar production time series based only on the input NET consumption time series.
A neural network machine learning algorithm, that is responsible for taking into consideration the weather conditions, on the location of interest, at any given time (e.g. solar irradiance, temperature, cloud coverage etc.) in order to co-relate and correct the output of the previous step.
Final normalization of the output is done by using the historical data from the same installation, in order to make sure that abnormal values are removed from the time series.
In order to assess the quality of the implemented approach for solar production analysis and prediction, we did an experiment with multiple end-users from our European customers with PV panels installed. The solution’s performance has been evaluated against solar production ground-truth data coming from 20 installations. The data collection was realized for the summer period of 2019, i.e. May 2019 until September 2019.
The following images present some examples of the daily comparison between actual and predicted solar production. It is easy to see that the predicted solar production time series follows closely the actual one, meaning that the resulting solar production estimation will be pretty accurate.
The following graph shows the overall assessment results for all the test installations, using two metrics: (a) the mean of daily errors, meaning that the prediction error of each day is taken into account separately, and (b) the total monthly error, which is the difference between the aggregated monthly actual production in comparison to the predicted one. The results indicate an estimation error that was consistently lower than 15% from reality, whereas regular NILM performance (appliance and activities detection) was of the requested accuracy, i.e. more than 90% on a monthly basis.
The low level of error in the estimation of the solar production by our approach ensures that we will be able to reform the overall consumption time series pretty accurately. This will allow for detecting end uses of energy intensive appliances even during the daylight time, when solar production is taking place.
The NET2GRID solar production metrics service is able to provide diverse assets for various stakeholders, from both sides of the business, meaning both the consumers, as well as the energy suppliers and the DSOs/DNOs:
Consumers: The end-customer now has at its possession an all-in-one service/app, which allows for an in-depth overview of his solar production specifics on a daily/monthly basis. No need to check multiple Apps at once, no need for installing an intrusive additional metering equipment. The service provides a lot of production insights that are not available from any other energy service provider, such as efficiency and peak details, which can aid comprehension over one's solar installation capabilities.
Energy Suppliers: The most interesting aspect of having an aggregated or personalized overview over customers solar production is being able to use that information for managing and scheduling Demand Response programs towards the customers that are viable to follow them. In addition to this, this service can easily enable the marketing and provision of new ancillary services from the suppliers to the customers in order to improve or optimize their solar production. Examples of such services are cleaning, maintenance, relocation services etc.
DSOs / DNOs & TSOs / TNOs: The Grid Operators are able to have an aggregated overview of the actual production specifics of the residential consumers that are integrated on their operational Grid, being able to make better forecasts and manage their assets and production needs based on their customers requirements. The fact that grid operators can also check in real time what the solar production looks like might unlock useful opportunities for actions that can reduce losses, increase financial benefit and prevent energy production loss. For example, when grid is likely to overproduce for some period of time, the TSO may ask for solar curtailment, where the customers get compensated to switch off their solar panels for some time, or use as much as possible their own solar panels without net metering taking place.
Solar Production App Screens and Metrics
The NET2GRID’s Ynni mobile application (white label app) presents the results of the daily/monthly solar production analysis in a very easy-to-comprehend and intuitive way. The following screen captures from a demo account are highlighting our point.
The first screen shows how the per hour energy solar production is separated during the day in the Detailed Insights screen. This way, the users can easily see their daily solar production specifics: the estimated daily solar production, what time the solar production started and what time it stopped, as well as the hours of the day that the production has surpassed the consumption of the installation, meaning that the overproduced energy was returned back to the Grid.
In the same screen you can select an individual appliance from the list and see plotted the specific energy consumption in orange and how much of that appliance consumption was used from the solar energy being produced.
The second solar screen is the main solar production metrics screen, which gives a more in-depth overview of the solar production metrics for that specific day/month.
More specifically, the metrics that are presented in this screen are:
Generated Solar Production: The value of the overall estimated daily/monthly solar production, in kWh. The percentage shows how much of the overall installation consumption was covered by solar production.
Solar Production To Grid: The value of the estimated daily/monthly solar production that was returned to the Grid, in kWh and as a percentage of the overall solar production of the installation.
Solar Production Used: The value of the estimated daily/monthly solar production that was used by the installation for self-consumption purposes, in kWh and as a percentage of the overall solar production of the installation.
Peak Production Value: The peak of the active power value for production during the day/month, in Watts. This is the maximum value that the solar production reached at the period of interest.
Peak Production Timestamp: This specific date and time that the aforementioned peak value appeared on the solar production time series. of the second that the peak value of solar production was monitored for that specific day, in dd/mm/yyyy, hh:mm format. It must be noted that even though it is expected that this local time that will be represented by the peak timestamp would be similar each day, this may not always be the case. This has to do with the fact that this metric can be heavily affected by the weather conditions taking place at a specific date and place.
Sun Hours Monthly Efficiency Percentage: This percentage shows the efficiency of the solar panel for a specific day/month in comparison with the actual maximum capacity of the solar panel for that specific month. This is a different metric, since the maximum capacity during winter months is expected to be lower than at summer months.
Sun Hours Annual Efficiency Percentage: This percentage shows the efficiency of the solar panel for a specific day/month in comparison with the actual maximum capacity of the solar panel.