Scaling AI and Machine Learning for Improved Operational Reliability

Written by Justin Pivec, Senior Manager, Global Channel APM Sales, AVEVA

Scaling analytics capabilities across organizations provides the framework for asset and operational reliability and continuous optimization of processes.

The effectiveness of your organization’s analytics directly correlates to how well your workforce understands the information they collect, and how that information can be applied to improve the reliability of operations in your process. This framework for handling information is called a Digital Thread, and it’s a way of understanding universal access to operations data as it moves through the organization to facilitate contextualized information for individuals and departments and provides complete visibility into operations and asset lifecycles.

Strengthen your Operational Reliability by scaling your analytic capabilities

Many organizations stop at adopting basic analytics, rather than fully investing in developing a digital thread to span operations. This can occur because perceived complexities and a poor understanding of software capabilities can make development of a digital twin somewhat intimidating. However, at scale, implementation of analytics tied to the digital thread reduce time to value and foster a programmatic approach to how they are deployed, used, and maintained. As these gains extend through the organization, implementing the framework of a digital thread can help foster exponential returns as business units better collaborate and operate based on a single source of truth. For example, an organization may wish to use their digital thread to close the loop on prescriptive guidance by using asset libraries in combination with best practices to prescribe mitigating actions for potential failure risks that can be detected with AI and Machine Learning.

It is important to note that the efficiency gains made by effectively scaled analytics are not the result of a one-time project. To get the most value from investments, scaling your analytics and improving your digital thread should be a continuous improvement endeavor, as well as a part of your organization’s digital journey. As value is proven, analytics can often be easily replicated throughout the business.

Challenges and Best Practices for Scaling Analytics

There are as many reasons to improve analytics as there are data points. For example, analytic solutions may already be used in the following cases:

• Instrumentation
• Equipment
• Production
• Customer service/retention
• Sustainability
• Operational Resilience
• Quality Assurance
• Energy Efficiency
• Asset Reliability

These are often the starting point for many organizations. Unfortunately, when the time comes to scale or improve analytics, some teams may have difficulty clearly identifying the business need or calculating the return on investment necessary to justify increasing analytical capabilities. Measuring and reporting results are critical to a sustainable analytics program, and attention to outcomes will help overcome the challenge of poorly understood metrics. Be sure to consider how the diagnostic process occurs, and how a full digital twin can help improve your diagnostic capabilities.

Another common pitfall includes scaling technology, such as equipment and software, and integrating it with your systems. Integration hurdles can be alleviated by choosing a trusted and reliable software partner like AVEVA that builds solutions with scale in mind, can assist operators in understanding the full capabilities of their software, and provides the communications capabilities for easy integration with equipment and systems. Your deployment model (on-premise, Cloud, Software as a Service, and Hybrid models) may also determine the speed at which you can scale. Scaling with the Cloud is often faster than doing so on premise or at the Edge, so leveraging the Cloud in deployments where it is applicable can help you achieve scale more quickly.

If your organization lacks communities of practice, it’s a good idea to set them in place. These may include a standard template for use cases, training/onboarding processes, ROI calculators, implementation guides, and accessing support.

Resource Management is another area where some organizations struggle. You can avoid recurring problems by defining a monitoring and diagnostic process with a program manager as you become more efficient over time. Outsourced services can conduct the monitoring and diagnostics to supplement where resources are constrained. Ensure that training for the front line, engineering and management personnel is aligned with how the predictive workflow process is executed.

What are the Next Steps?

To help, we’ve put together a basic guide of the steps you’ll undertake as you scale your analytics and improve your operational reliability.

1. Define the results you want to accomplish
2. Define the value of improving your analytics
3. Define the information sources and types of data that can contribute to achieving the result
4. Define the analytic approach or approaches to achieving results
5. Test the analytic approach to confirm it satisfies the demands of the project
6. Scale the analytic solution across the business
7. Measure and confirm value realization over time