Predictive analytics for the largest Oil and Energy Company in CEE

The Management of a leading Oil and Energy company wants to increase the efficiency of operations in and reduce costs of logistics. Priority has been assigned to the following areas: product quality management, predictive maintenance and lead time processing.

Product quality management – One of the company’s challenges is the need for constant monitoring of key fuel quality indicators. The company’s production, warehouse and sales network is extensive and the system based on laboratory tests is expensive. The company’s management wants to use predictive models for ongoing monitoring of fuel quality parameters. Experts intend to better understand the causes of fuel quality deterioration in order to better control the process of fuel production and distribution.
Predictive maintenance – The company maintains thousands of gasoline pumps at various locations. Their continuous work is of key importance. The company intends to use predictive models to determine factors which affect service life of gasoline pumps and estimate their optimal service time.
Lead time processing – The company transports millions of tons of products by railway fleet. There is a need for better management and optimization of the railway transport and improvement in lead time optimization.

Building of analytical data warehouse
In order to address analytical challenges we designed, built and developed the analytical data warehouse which runs ongoing ETL (extract, transform and load) processes integrating large amounts of structured and unstructured data from various IT systems including: production, transport, sales, laboratory and sensors data.
Reporting system
As part of the project, we implemented an interactive reporting environment with personalized reports providing the user the possibility of performing independent in-depth analysis.
Analytics environment
A distinctive feature of the project was the development of a dedicated data analytics environment enabling data preparation and visualization, building and implementation of predictive models.
Building predictive models
We have built accurate and stable predictive models which are used for optimization of key areas of logistics.

We delivered over a hundred accurate and stable predicting models which are used for estimating key indicators of petrol quality.
We have implemented an on-going fuel quality assessment mechanism for thousands of storage containers.
Our predictive models are a trusted source of information with regards to detection of irregularities and control of fuel quality.

Better petrol quality monitoring

Lower cost of petrol quality management

Fast identification, verification and elimination of potential threats

We have built predictive models explaining factors impacting petrol pump filters’ lifetime, e.g. base fuel, components, additives, filter type, volumes.
Predictive models enable optimization of filter servicing costs at petrol stations.
Due to better understanding of factors impacting life expectancy of petrol pump filters the company management can make better decisions with regards to application of product components.