The first of the Fusion Fundamentals series with Suchi Badrinarayanan, Business Consultant, at Fusion. This series will focus on liberating and normalizing OT data to accelerate enterprise-wide adoption and high business value realization of AI/ML. This first topic is how to Enable Industrial Data Analytics.

Read the video transcript below for your own convenience:

We've generally seen that there seems to be a big divide between the data coming from your sources and those who want to do the analytics related to that data.

Look on the left-hand side of this photo; you can see a couple of different sources you can get information from. Most people are trying to get time series information from the site, corporate historians, or even the operations management systems. But in addition, we've seen a lot of new sources that people are using, for example, IoT devices and instrument analyzers. All of these data sources need to land in a place where the analytics team can access them efficiently, and the data quality and reliability are also intact. 

Generally, what we've heard from people trying to get access to that data is as follows:

1.  Data is not analytics-ready. There's a lot of time spent cleansing the data and connecting the data in the first place.

2.  Only 5% of it is actually useful for whatever use cases are based on that data. 

At Fusion, we're trying to use that data for is really any of the BI (Business Intelligence) or Analytics teams. So on the right-hand side of the photo above, you can see many different ways people can use that data. One way is digital twins, but again everyone is trying to get that data in the spot where they can either do just general BI or use that to make ML or AI models.

When we are talking to customers, there are a couple of things that always come to mind.

The problem of secure and scalable extraction, too many sources and formats, and industrial manufacturing pressures are the three that come to mind. 

The problem of secure and scalable extraction

The first regarding secure and scalable extraction really stands out. That data is typically behind multiple firewalls and needs to go through multiple layers of security. It is necessary to make sure that the data streams that are coming out are incredibly scalable. So that the data can potentially work with millions of tags because the data source is so rich and has that much information in it.

Too many sources and formats

The other thing that we've been recently hearing is that there are a lot of sources of data and a lot of formats of data. So people might have multiple different historians across their enterprise but also might have different formats for that data as well. That makes it even more challenging for analytics people to go and access that data because it's not all in one easy-to-use format.

Industrial manufacturing pressures

Any kind of “do it yourself mechanism” takes way too long, and people who have these use cases need that data and are not usually willing to wait the three months to a year that it might take for someone to ramp up and create this architecture all by themselves.

Closing the gap with Fusion

This is exactly where Fusion comes into play by connecting the data with the use cases. What we’ve created with Fusion is a Cloud-Native Industrial Analytics Data Hub. All of the data that's there and is replicated from your source systems on your left-hand side gets the data there securely and reliably and, most importantly, analytics-ready. 

Whatever use case you have after that, whether it's Pipeline Integrity, ESG reporting, or even Anomaly Detection, all of that data can be accessed directly from Fusion.