Sia Soft provides custom software development and product delivery across multiple domains to customers worldwide. In the energy domain, Sia Soft collaborates with Elevate Global, as it’s specialized partner for Data Science and renewable energy production forecasting and modelling. UbiOps provides an independent operations layer to run these models in production, so that customers make requests and apply the results in their day-to-day operations.
Why have you built these models in the first place?
Michiel van der Pant, CCO at Sia Soft: “This collaboration started as a team initiative to see how we can contribute to the energy transition. The team is committed to this cause: every winter in Skopje there is still a lot of smog because of all the wood-burning stoves. This needs to stop!
By helping to increase the efficiency of green electricity via (solar, wind) forecasting, we can accelerate the transition to green energy and help make the world a better place. To contribute we not only had to apply the newest technologies but also had to learn a lot about the energy transition market and solar and wind in particular.”
How do you tackle this societal problem?
“First, we collect data from different sources (among other weather data) which we combine with the customer solar panel/ wind data to make the best possible production forecasts. For each model in production, we operate/ train multiple alternative models in the background to check on the behavior of all relevant variables over time. We optimize our results through the automatic validation of competing models. Finally, we continuously monitor our error rates and enable automatic anomaly detection.”
How do you work with organizations that want to solve these challenges?
“We have a team of experts that can help with integrating our solution in any environment, both with preparing and handling the required input data as with the usage and interpretation of the output data. The data of our customers, the weather forecasts and historical weather data, as well as the full end-to-end energy forecasting service implementation, resides fully cloud-based on UbiOps, on infrastructure hosted in the EU.”
“UbiOps enables us to create products for specific customer segments, such as those working with energy data, and serve multiple of such customers. That makes it cheaper to use, while restricting our access to confidential customer data and infrastructure.
In this case we have deployed Elevate Global’s time-tested energy forecasting models to predict solar production for one of our clients. Sia Soft has DevOps knowledge in-house, but we use UbiOps for optimal horizontal scalability and service elasticity, thereby ensuring business continuity.
By managing user permissions, the customer can connect and adjust their database to UbiOps, without accessing the code. We can adjust the pipeline and the code but have restricted access to the customer data. At the same time, our source code stays protected, so our teams can develop a model once and help a variety of customers with it, decreasing the development costs per client significantly.”
Sia Soft connected a first small solar park in the east of Amsterdam to enable a real-time POC.
The first results can be seen in the dashboard below.
Figure 1: Actual solar production (solid blue line) and forecasted production (solid green line) plotted against time (17-12-2020 – 12-01-2021)
Next steps – future, expectations, plans
Sia Soft and Elevate Global continue to develop models for a variety of customers in Europe. The next in-depth use case will be published in a few weeks. That article will provide insight into how the models are developed, deployed, operated and provided to the end customer to use.