A familiar phrase, and a commodity that is harder and harder to come by in the automation industry.  The statistic is that 50% of the automation engineers are retiring in the next 5 years.  Many are NOT being replaced.  Their experience vanishes. 

The HMI/SCADA automation systems we are all using are great at watching our processes, and alarming on variables that are out of range.  But they don’t capture or create experience.  The Historians we put in place also store the complete history of the process, but they too don’t manufacture experience.

But what is experience?

Ask an old-timer, and he will tell you it’s the feel of the control room when he's there.  It’s the vibration of the floor.  It’s the smell of the air.  It’s the clang of the machine.  Yes – all true, and all difficult to measure.  But let’s look a little deeper – What is experience?

Experience is knowing the temperature of a motor under this load condition.  It’s knowing the typical flow when three pumps are running.  It’s knowing the time it takes for an oven to preheat, or the energy that is consumed to heat a building when it’s freezing outside.  It’s also knowing that the process is usually running at this rate.  Experience is understanding the operations of your process, in the current context.  

Experience is what’s missing in today’s automation systems.

What is context?  That one is easy.  It is selecting a Key Variable that others will be linked to.  In Building Automation, Heating Cycle time will be linked to Outside Temperature (ok – and thermostat setting).  Water Heater cycling will be based on Water Use.  Lighting will be based on Occupancy.  In Waste Water Treatment, Chlorine addition will be based on Flow.  Flow may be influenced by rainfall, and time of day.  In Power Generation, it’s the fuel use for this Power Output, or the Transformer Temperature for this Load.  Context is the current state of the process.

When you combine the current Context, with expected dependent variables – that have been learned over time (a very important point), then you have Experience.  This is what’s in the head of your operators and engineers.  Experience is what’s missing in today’s automation systems.

If you had Experience in a database, what could you do with it?

Reference it -  The Experience database becomes your go-to guy for information.  Have a condition you haven’t seen before?  Has this temperature ever been this high?  Yes – when and what were other variables at the time?  What did we do about it the last time?

Empower It – Let the Experience Database watch your current data and let you know if you are out of expected ranges.

Generate Reports with it – Management would like to know that the process is running in its sweet spot.  You are likely to have less unscheduled downtime there.  Maybe you’ve identified contexts that have yielded better operations, higher efficiencies, lower waste, etc.  What were the target variables associated with that operation?  The experience database contains all that information.

Integrate with it – Let it provide you with expected variables at the current context, for use in your existing automation systems.  At this load, an experience database can tell you what the expected ranges are for variables (typically between this and that).  All your HMI/SCADA displays can be updated in real-time with expected values, right next to your current values.  This adds a whole new dimension to alarm management.

Experience lets you be Proactive in driving an outcome.

Catch Anomalies before they become Alarms – Alarms are typically on critical variables.  Why?  Because you can’t possibly manage defining them on all variables, for two reasons – the alarm storms that could occur when there is a problem, but more importantly, because you can’t manage all those varying alarm setpoints.  Imagine if you had an experience database that could watch them all for you, and notify you of any unusual variables?  Now, you are watching the process closely,  and noticing the subtle changes that may lead to an alarm.  You have the opportunity to change an outcome, to prepare.  You are no longer reacting to a “Failure”.  Experience lets you be Proactive in driving an outcome.

An Experience Database exists Today!

Thanks to Aware Technology, licensing and improving upon NASA technology, an experience database comes in the form of a product called Process Data Monitor™ (PDM for short).

This product essentially watches your variables, and builds a database of data, categorized by Process Context.  The result – a database of Experience.  It can tell you the usual motor temperature for this load, and the HVAC Cycling for this outside temperature.  It can tell you when your filters are clogged, by noticing a lower air flow based on the current set of fans.  Most important, it can do this without tedious alarm management and constant system maintenance.  It will learn from usual behavior, and can detect unusual behavior, just like an experienced operator.

Learning can be fast, or can happen over time.  A historian enables learning of history through an import of data.  Six months of history, a year or more can be read in to populate your initial experience.  Or, the system will connect to real-time data, through the use of OPC Standards, and will begin the process of learning in parallel with ongoing operations.  Then you decide when it has learned enough, and when it should start alerting you to change.

[an error occurred while processing this directive]Technically, how does one create experience?

Building an Experience Database is simpler than you might imagine, but it requires some out of the box thinking…  First, organize data into sets that are “correlated” somehow linked in their data movement.  For instance, your car engine, running down the road at 55 mph will be in this gear, typically have this engine temperature, get this fuel mileage, have this oil pressure, charge the battery at this rate, etc.  All those variables will be monitored by PDM, and PDM will store the data into a relational database cluster of variables.  Here is some of the magic.  As a new set of variables comes in, PDM will compare the new data, to its database of clusters, utilizing patented pattern recognition algorithms.  If there is a match with an existing cluster, it increments a match count, stores relevant data like lowest values, highest values and match time, and waits for the next set.  If there isn’t a match, it creates a new cluster (possibly notifying you of the new learning that is happening).  Another bit of magic – if the set is close to an existing cluster, PDM can decide to either expand an existing cluster to accept the new set, or to create a new cluster.

In the end, you have a new database of contexual data.  Something that has not been available to common automation applications.

Alternatives to Learning and Pattern Recognition

Sophisticated systems for mathematically modelling processes have existed almost since the introduction of the computer.  These systems rely of quality data, mathematical equations and continual maintenance to model your process and run along side it, comparing model outcomes to real-world data.  Their cost, both in terms of capital expense, and in terms of commissioning and upkeep, have positioned them as solutions for the most complex of processes or high value assets.  They are often categorized as APC (Advanced Process Control) products and while they have the ability to mathematically calculate expected values, they do not capture experience.  They do not understand past occurances and their frequency.  However, they are an ideal solution for complex simulations and can be excellent in refining control scenarios.  In fact, APC solutions would benefit greatly from being combined with a Learning and Pattern Recognition system such as PDM.

Summary – How have you lived without one for so long…

With Controllers, HMI/SCADA  and Historians becoming a commodity, the next step in automation is to turn data into information, to make these systems and their operators smarter and better equipped to run the systems of tomorrow.  The most useful information is disappearing at an alarming rate – Experience – and it will be systems like PDM, that have the ability to put experience into a database, that will become the next wave of system automation.

Products like PDM, that have the ability to put Experience into a database, will become the next wave of system automation.

About the Author
Roy Kok is the president of Aware Technology.  He has spent over 30 years working in the automation industry, dealing with HMI/SCADA, Historians and layered solutions.  Aware Technology was founded in February of 2011 and has developed its Process Data Monitor product as the next logical extension to any HMI/SCADA and Historian.  You can get more information at www.AwareTechnology.com and can contact Roy at Roy@AwareTechnology.com