Partnering to Push Digital Transformation Limits
Industry leaders share how they’ve put their heads together with the Seeq product team to create solutions that scale, empower, and maximize access and profit.
Long-standing collaboration was a common thread among the presentations in Conneqt 2023’s Digital Transformation track. Presenters highlighted the outcomes that are possible when you combine innovative thinkers at some of the world’s top manufacturing organizations with an agile technology company’s software development and user experience (UX) expertise. Outcomes achieved include how:
- Flint Hills Resources is efficiently monitoring 8,000+ parameters on 4,000+ assets
- Williams is avoiding inflated receipt meter readings, preventing costly gas replacements at delivery
These leaders, and the iteration behind their fit-for-purpose solutions, have helped guide the path of the Seeq product roadmap.
Flint Hills Resources Masters Monitoring at Scale
The monitoring capability team at Flint Hills Resources (FHR) has a goal of automated anomaly detection for early identification of process and equipment anomalies to enable profitable action. Their vision is to monitor all equipment and processes, not just specific assets. They partnered with Seeq in creation of a suite of Add-ons to Seeq Workbench to:
- Construct use case specific asset groups for rapid scaling of analyses
- Mass implement univariate models on data in asset groups
- Manage, prioritize, suppress, or add context to model alerts
What began as a wide scale pump vibration monitoring project has expanded to the monitoring of 8,000+ variables across 4,000+ assets. As FHR’s monitoring activities increased so did the need for a performant, sustainable solution. This partnership proved to be of tremendous value to both FHR, whose needs were defining the first iterations of the solution, but also to the rest of Seeq’s customers who have benefited from the productization of these ideas, like Asset Groups.
The Seeq UX team embarked on a journey with FHR to improve the user interface (UI) that was used by the monitoring analyst to view which model alerts were present, investigate, and respond to them, resulting in a new solution. Gathering user feedback through usability testing led to some key advances like customizable and default user views, quick trend capability, notes, and logic-based suppression. While FHR’s monitoring capability team continues to improve upon their workflows, they note the solution “has really been a step change in our performance within our monitoring capability.”
Williams Systematic Approach to Reducing Measurement Errors
Williams faced a significant financial hit each time the meters into their facilities overestimated flow. This led to a mismatch in the stated and actual delivery volumes, the make-up of which is replaced by Williams at the market rate, many multiples higher than their feed prices. When your pipelines handle 30% of all the natural gas consumed in the United States, these losses from receipt meter measurement errors add up fast.
Detecting meters which are overstating flow (due most commonly to freezing events) means William’s analysts can adjust the flow calculations of the meters to reflect the actual flow and reduce the likelihood of a receipt-delivery meter mismatch. Williams’ journey to tackle this problem began in 2015 using a rules-based approach to identify events. While the initial results were positive, the logic required became increasingly complex, and the false positive and false negative rates were too high. Even with layers upon layers of logic, they were missing events that experts could detect when looking at the data.
Enter the machine learning approach. Using the Seeq ML algorithm, part of the Seeq Process Health Solution, Williams experts provided the model data from previous known freeze events for training and the model output a new signal calculating the probability that a meter freeze was occurring. The Williams team examined early model outputs, and their feedback was used to retrain the model to get it ready for scale. They leveraged their existing asset hierarchy groundwork to apply the model across their fleet of meters. Their daily interaction with the data now starts with a review of a prioritized exceptions list, flagging meters with freeze probability >70%. Williams’ next steps are to build from the collaboration between Flint Hills Resources and Seeq to implement a robust decision support workflow solution ahead of the next freeze season.
Key Takeaways
- Measuring and monitoring alone do not necessarily translate to value. There must be a program or process in place that encourages people to act based on the information at their fingertips.
- These partnerships would not be as successful as they’ve been without the groundwork in place for rapid scaling. Whether asset hierarchies are built in Seeq, or via one of our technology partner’s solutions, the resource investment to develop and tune them to the needs of the use case enables value to multiply quickly.
- Partnering with software companies lets manufacturers take advantage of teams of developers and UI/UX experts to implement an intuitive solution, and the ability to personalize the way they interact with data.
The Seeq product team is always busy innovating with our customers and technology partners on the next big thing in time series advanced analytics, AI, and ML. View more success stories from Conneqt 2023 here.
Are you interested in sharing your Seeq success story with your industry peers? Submit your abstract for Conneqt 2024 today.