CLIENT STORY

Driving savings through real-time optimization

UPM Energy’s data platform journey

Utilizing real-time data to tackle volatility in energy costs

The world of energy optimization is full of problems that data can help mitigate. Energy price volatility has increased significantly during the last couple of years. For industry buyers, electricity can account for up to 30% of the cost of the final product, depending on the manufacturing technology. 

Reducing energy costs can have a big impact over time - financially, operationally, and strategically - improving e.g. margins and ESG positioning. 

The future looks much brighter for companies that have the capability to divide processes and workloads according to the electricity price and avoid high-cost operations. 

Energy optimization helps find the best scheduling for orders, so that processes consuming the most electricity are run during the hours electricity prices are at the lowest. 

The possibility to do energy optimization is linked to the ability to harness real-time data.

Recordly and UPM Energy case study 2025

The problem

UPM Energy needed a robust data platform to enable modern ways to create optimization and forecasting use cases using ML (machine learning) for its internal operations. 

Other needs, such as establishing a working support function and decreasing operating costs, further directed the development of the data platform.

Moreover, the new platform needed to drastically improve UPM Energy’s visibility into data pipelines and the overall status of the solution. 

The techy part - Recordly's solution

Together with UPM Energy, we built a universal data platform on Azure that optimizes both energy consumption and costs with real-time data processing. 

The project team of eight experts, including data engineers, ML architects, data architects, DevOps developers, and a project lead, completed the project by combining their specialized skills into one coherent delivery. 

The data engineers built the pipelines that bring real-time energy data into the platform reliably. ML architects ensured that ML models could be developed, deployed, and monitored robustly through Databricks’ MLflow.

Data architects provided the foundation for data traceability, auditability, and performance optimizations, ensuring the platform remains trustworthy and scalable. DevOps developers established CI/CD pipelines with Infrastructure as Code and Databricks Asset Bundles, securing smooth and automated operations. 

Guiding the entire effort, the Project Lead coordinated work across roles, aligned the technical vision with UPM Energy’s business goals, and ensured the project stayed on track from start to finish.

UPM Energy (1)

Key components

1. Universal data platform designed for scaling across the entire organization

2. Cloud-native foundation that centralizes compute and code, making collaboration more efficient

3. Mission-critical readiness with high availability and continuous operational accuracy


UPM Energy’s new solution is built on Databricks’ platform, a collaboration environment for any kind of data. Their data was made available to the whole spectrum of Databricks features within defined data access policy limits. This enabled data usage with Databricks MLOps. 

Databricks’ MLOps (MLflow + Unity Catalog) provides useful tools and a collaborative development environment for advanced analytics. Among these, optimization models play a key role. 

By applying mathematical optimization (MO), a method for finding the best possible decision among countless alternatives, UPM Energy can automate energy consumption planning based on real-time data instead of assumptions.

This improves quality by relying on the latest real-time data instead of individual assumptions. Once results are ready, they’re quickly available where they’re needed, ensuring critical information is accurate and operative 24/7.

This project stands out as a unique and compelling Databricks use case due to Recordly’s approach to combine the platform’s mathematical optimization orchestration and tracking.

Why it mattered to UPM Energy

Together with Recordly we set out to rebuild some core infrastructure. But, as often when working with true professionals, we succeeded and exceeded expectations.
 
- Jarkko Kuismanen, Platform Owner, Digital Solutions, UPM Energy

The impact

Immediate cost savings

Electricity invoices of industrial operations can be significantly reduced

Continuous value creation

Utilizing ML and MO algorithms bring ongoing decision-support benefits that accumulate over time

Enabled simplifications of processes

The use of Databricks enables streamlined yet effective data flows

New and smarter ways of working

The transition to a cloud-based, centralized environment makes collaboration smoother and more efficient across teams

Scalability

The solution can be scaled to many other use cases across the global organization

Established frameworks 

Enabling future use cases to be implemented quickly and effortlessly.

Recordly and UPM Energy case study

Looking ahead

Three optimization models are now in production, with more on the way.

Importantly, the project has proven replicable: use cases outside the original scope have already been successfully implemented, confirming that the solution works like a template for the whole organization.

Are you looking to optimize your organization’s energy consumption, supply chain or production?

Let’s talk about how real-time data and optimization models can help you achieve the same kind of scalable, long-term impact. Contact us or schedule a discovery call below. 

<span style="font-size: 14px;">SEE OUR SERVICES</span>

Ask more

Further reading

<em>Improving</em> insights and strengthening data governance

Improving insights and strengthening data governance
Lumo Analytics improved data governance, cost visibility & performance with a smooth Databricks Unity Catalog migration.
Read on
<em>Shaping</em> Neste's data journey over five years

Shaping Neste's data journey over five years
From proof-of-concept to global data platform; Neste’s five-year journey shows how trust and teamwork create lasting business value.
Read on
<em>Revolutionizing</em> renewables innovation with AI-powered knowledge discovery

Revolutionizing renewables innovation with AI-powered knowledge discovery

Neste embarks on a new journey, using Generative AI in innovation and technology development to augment information discovery and creation processes with a new AI-powered knowledge assistant solution.

Read on