Blog Post

Dealing With IIoT’s Firehose of Data


Dealing With IIoT’s Firehose of Data

In the previous post, The Problem of Too Much Manufacturing Data, we examined the problem with treating the flow of data made available through the growth of smart devices — as manufacturers pursue a fully transformed operation.  Viewing data in silos can lead to serious misperceptions about current performance and drive bad decisions.  The solution was to look at the totality of data to see the bigger picture but that presents the challenge of how to actually “see the forest for the trees”, especially as manufacturers turn to automated systems to alleviate labor and skills shortages.  

Humans have an innate ability to process overwhelming large volumes of data by focusing only on relevant information.  If you think about driving a car, we constantly are processing thousands of pieces of data related to the car’s position, performance, direction of travel, surrounding environment, weather conditions, and dozens of other factors.  This is why it has proven so difficult until recently to have autonomous vehicles except in tightly controlled scenarios.  Manufacturers must approach their push to autonomize their operations in much the same way as humans have learned to drink from the firehose of data that we constantly process performing our daily tasks.

Classifying What is Important

The key to handling any crisis is the process known as triage.  Anyone who has seen a medical drama or disaster film is familiar with the concept of triage where limited resources such as those with medical skills, are assigned to care for patients based on the severity of their injuries.  Those with life-threatening injuries are tended to first while those with serious but non-threatening injuries are treated second while those with minor injuries are relegated to being seen last.  The word tirage derives from the French term trier which means “to sort” or “to separate”. 

As manufacturers design their operational-systems architecture they need to build in the capacity to triage the flow of information.  This will enable them to deal with the “firehose of data” that IIoT enabled devices make available to our manufacturing systems.  This means that data needs to be triaged as it is generated so only the most important and critical data is passed along. This is performed in a “push” fashion such as via alarms, alerts, or other critical forms of messaging.  Second-tier data needs to be made available via broadcast so systems that can use the data have it available as needed.  Third-tier data needs only to be made available on request.  Simple in concept but doing this correctly requires systems and software that can contextualize data so that data is processed correctly.  Just as in medical triage, where a patient can quickly change status, such as a person with internal injuries can suddenly crash and require immediate attention, in manufacturing operations there can be data that suddenly becomes much more critical if there is a major process upset.

Contextualize Data on the Fly

Your manufacturing execution system (MES) needs to interact with your operators the same way a modern car’s electronic control module (ECM) works.  The ECM is constantly monitoring fuel flow and current fuel levels to provide an estimate of the distance until your fuel runs out and typically displays this information somewhere on the dashboard.  Most of us look at the fuel level occasionally to judge when we need to refill the tank and occasionally glance at the DTE number.  Rarely do we monitor actual consumption in current and average miles per gallon (MPG) which is usually a display several levels down in the display menu.  However, most of us immediately seek out this display if the “low fuel” light turns on and we get the loud “ding” that we have an alarm.

In manufacturing, you need to structure your MES the same way.  The MES must display the information to the operators that is relevant to maintaining production at the proper speed, quality, and performance. But in the background, it should be monitoring machine performance data such as the energy a machining center is using. And by utilizing an AI-based background processing system, an anomaly alerts the operator that the machining center is operating abnormally and gives the option to display critical parameters such as current draw, speed, tool vibration, etc. This helps the operator diagnose the problem.

You Have the Data. Use It, But Use It Wisely

The key to successful use of cascaded data from smart devices can support manufacturers:

  1. Don’t waste the data. Make sure to collect it if it is available.
  2. Use intelligent (AI-based) processing to triage the data and recognize that not all the data is valuable all the time
  3. Make sure to present data in context.  Showing an operator detailed data without including other information that illustrates why the data is important will slow down their ability to make meaningful decisions.

 

New Call-to-action


ibaset featuredcustomers