A Step Change in Refining Unit Monitoring


← Back to Case Studies
Industry

Oil and Gas
Downstream

A world-class refining organization set out to improve the health of their assets through a new approach to unit monitoring. With the goal of improving overall performance of their FCC (Fluid Catalytic Cracking) and Reformer Units, process unit operations required a deep understanding of how parameters change over time and affect the quality and quantity of production.

Key to this understanding is the ability to rapidly contextualize not only time-series process data, but also combine it with maintenance work orders and safety constraints. This improved context would create a 360-degree view of their assets, giving them a standardized, scalable approach to rapidly identify and respond to changes through ongoing analysis.

With Element Unify, engineers were freed up from the frustrating and burdensome tasks of collecting and preparing data. Tasks that typically consumed 80% of their day, now only consume 20%. This allows them to spend more time delivering their expert guidance, a 4X increase in engineer productivity.

Element completed the work in 3 days, down from the anticipated 3 months, resulting in an efficiency increase of 30X.

In addition, project time and costs were reduced by 75% (for example the first 3 PI AFs were built in 3 weeks vs. their previously successful project of over 4 months) delivering value at rates previously unseen.

Challenge

The customer determined that the simple task of aggregating existing production data presented several challenges that negatively impacted their process unit analysis as they were unable to use relevant and trusted data in a timely manner:

First, data sources were siloed within the refinery fenceline as well as enterprise data centers making it difficult to integrate. These sources included: time-series data from the OSIsoft PI System; process safety data from a varying set of HAZOPs, layers of protection analysis (LOPA), and safe operating limits systems (SOL) engineering spreadsheets; and maintenance information such as functional locations and work orders from SAP PM.

Second, once extracted, data existed in a varied set of formats, complicating the analysis process that required a high degree of intervention from site engineering. The customer had used an ad hoc, manual process centered around Excel as a path to integrate and model the data.

Today we are building calculations on laptops (Excel, R) without any approval process in place, so we could find ourselves in trouble!

VP of Process Optimization

However, that approach proved extremely inefficient and lacked data integrity and governance. In addition, this existing process also required consumers of the data to interpret process alarm & event system data, SAP work orders, and whether they had all the plant data they needed to make timely decisions.

Third, unaudited engineering calculations were developed and run on laptops, creating additional risk.

Solution

The Element solution complemented their existing OSIsoft PI System infrastructure and expanded their use of Microsoft Azure.

The following activities were performed:

  • Connect

  • Deliver Element Unify within the customer’s Azure tenant

  • Configure OSIsoft PI System and file agents to ingest relevant data

  • Manage

  • Design and build equipment-centric Asset Twins based upon their desired attributes of their target equipment, including heaters, furnaces, turbines, compressors, pumps, fans and reactors

  • Transform and contextualize process, maintenance and safety data

  • Share

  • Export metadata in the form of PI Asset Frameworks (AFs) to import back into the OSIsoft PI System

  • Export various hierarchies and raw data to the customers Azure storage for use with Microsoft PowerBI, R Studio, and Microsoft Azure Databricks

Results

For the first time ever, the customer was able to view a complete set of refining data in ways previously unavailable. By using Element Unify, not only did they significantly reduce the time to design, aggregate, contextualize, and share all refining data together, the engineering and data science teams enjoyed an up-to-date, 360-degree view using their visualization / BI tool of choice. As an example, the customer was able to rapidly model 6,200 tags, uncovering 27,845 new data relationships that were captured within the Element Graph. These relationships translate to thousands of new ways to view process unit data, offering additional unit monitoring insights.

In addition, project time and costs were reduced by 75% (for example the first 3 PI AFs were built in 3 weeks vs. their previously successful project of over 4 months) delivering value at rates previously unseen.

Screenshot of static data import
Reduce time to build PI AFs by importing static data imported from relational databases and combine with real time sensor data

With Element Unify in place, engineers were freed up from the frustrating and burdensome tasks of collecting and preparing data. Tasks that typically consumed 80% of their day, now only consume 20%. This allows them to spend more time delivering their expert guidance, a 4X increase in engineer productivity. They were also able to relate production data with safety data. This had two benefits; It increased the timeliness and quality of their integrity reports, and also gave them a way to visualize production data alongside safe operating limits, informing them as to the consequences of exceeding those constraints. Engineering calculations can now be developed with changes tracked in a central location, not buried within desktop applications. Asset Templates can be encoded with variable operating envelopes, excursion limits, or anomalies. Physical relationships can be defined between attributes (e.g. Power = Current * Voltage).

With Element Unify in place as an efficient, cost-effective solution, they can not only continue to improve refining unit monitoring, but also expand their scope to help optimize a broader set of Refinery operations and processes.

In the past, when our engineers wanted to do an analysis, they’d ask for a number of data points to be loaded into SQL. But after 6 months, they discovered they didn’t have enough of the right data, so it was a waste of time for everyone.

Site Technologist