I am on the road for a month always a good opportunity to speak with new people, and again the
discussion of access to site information across sites and equipment becomes
critical. This requires different tools and different approach , we have also different people
involved.
Example I was at dinner with an engineer who runs installation/
tuning team for a wind turbine manufacturer. He talked about how his team
follows the installation team, and goes in and sets up the turbine, while I
expected the discussion to center on turning and setup of the turbine. The
discussion was actually on how they set the data gathering equipment, and
making sure the contextualization was in the data, so that he would be able to
convert to information through contextualization. This was not a nice to have
it was now a natural and critical part of the wind farm set up, as they leave
the farm (which is usually in the middle of nowhere) he talked about the second
phase of tuning, an analysis phase. This he does with his team anywhere but not
at site; they capture the data, and start setting situations, and patterns.
Applying past known conditions but he talked about doing this analysis across
10s of wind farms over 100s of turbines from all over the world. This engineer
was unaware of what I do, and whom I work for, so the discussion was very
candid and interesting to me as I had a mechanical engineer who is Gen Y, and
just assumes that this information will be available, and to him it is the
critical part of installation to setup this data acquisition
system, so he is empowered to work from where ever he is.
So the next question I tried to understand the type of
analysis and clients, profile of people using the information. “Self Service”
is the key, and the sense of discovery insight was key, using tools like
trends, mat lab and excel played in for
analysis, but key was a set of tools spreadsheet models and analysts that they
had built up over time. Notice this was not just talking reporting, dashboards
but it was the discovery aspect, the ability to combine different data sources
easily e.g. to compare like turbines.
Everything needed to be “plug and Play” allowing turbines to
be added by not instrumented people, access to information, and the data from the
turbine is not enough the information on the wind farm is key to provide the
context situation the turbine is performing in. So orientation, terrain, and
weather input for that site, both now, history and future is key. So merging
data sources from sites, with other models etc., and then compare from sites to
sites to improve and evolve.
But in the discussion it was clear that he and his team are
ideal for not storing the data local but going to the cloud architecture,
enabling these remote data source sites to gather and push to the cloud,
combine with the weather data for that site already in the cloud, and then consume,
discover and share from tools and models in the cloud. So the virtual team,
virtual sites, can become unified and effective in the collaboration end
evolution.
Time for
New Approach to Industrial Information
It may have been Tom Davenport, noted
professor, author, and analytics expert, who first came up with the terms descriptive, predictive, and prescriptive to describe the three
stages of maturity for analytics use within an organization:
• Descriptive Stage: What happened in the
past?
• Predictive Stage: What will (probably)
happen in the future?
• Prescriptive Stage: What should we do to
change the future?
The first stage ("descriptive) is
the traditional approach with trend analysis, simple tabular reports and
dashboards into the descriptive bucket. When applied effectively, these
technologies provide visibility into what happened - but only up to a point.
Many companies are now also rapidly adopting a third class of descriptive
solution, visual data discovery. The reason is simple - it can significantly improve
the odds that managers and process engineers can find the right information at
the right time.
But, a report of that type is never
going to help answer some important questions that may arise, such as:
"Why is work-in progress in progress for longer than in the past?” Likewise,
an indicator on a dashboard could show current on-time delivery performance. In
practice, dashboards are often more flexible than reports enabling users to
drill down from summary information to detailed data. This can help managers
understand cause and effect. But ultimately, users are still limited to answer
those questions anticipated by an IT specialist when he or she first developed
the dashboard. And that's the crux of the problem: most current BI solutions
still largely depend on IT specialists to create new BI assets (such as reports
and dashboards), or to modify existing ones.
Why does that matter? It matters
because most decisions have a distinct "window of opportunity." In
other words, after a certain point in time, any value to be had from making a
decision just vanishes. For example, the opportunity to maximize a load demand,
only exists while there is a window of demand, the ability to bring up a set of
wind turbines in a timely manner, understanding the landscape across the wind farm,
and the weather model for the next 24 hours. When the demand has gone the
Window of opportunity and data need has gone, so if it took 20 hours to get
that information the opportunity has been lost. In practice, all decisions have
a point in time after which they are no longer relevant.
Clearly, something more is required.
And for an increasing number of organizations that something is a visual data
discovery tool. Visual data discovery tools provide a very visual workspace
that encourages process analysis engineers, managers to manipulate data,
hands-on. They provide an engaging experience to explore data freeform, with
minimal or no help from skilled IT staff. Starting from the first glimmer of a
problem/ opportunity users can investigate freely, follow their train of
thought, and link cause to effect. That is exactly the type of capability
required to furnish answers to unexpected questions – the type of questions
that conventional reports and dashboards often struggle to answer.
Visual data discovery tools typically
provide:
• Unrestricted navigation through, and
exploration of, data, example search
• Rich data visualization so information can be
comprehended rapidly
• The ability to introduce new data sources
into an analysis to expand and follow it further These factors are at the core
of self-service analytics.
As the manufacturing world grows
increasingly fast-paced and dynamic, self-service analytics probably offered as
a service online; that can be consumed from anywhere, “on boarded” fast, and certain
tools only used as needed, will enable a more cost effective, reliable and
powerful industrial analysis environment. It is clear to me domain engineers
like my friend in Wind Turbines need a platform of tools to build their domain
solutions, to deliver a “self-service “ domain solution for Wind turbine
tuning, on boarding to really enable the ability to satisfy the dynamic world.
As we move to “micro grids” where the expert decision makers with lots of experiences
are not available, decisions and actions will need to be enabled through “Self
Service” visual domain tools.
