September 30, 2016
If you don’t create cultural and process changes to accompany the internet of things, it’s just a bunch of connected devices.
Systems that draw on a vast network of sensors to gather data, automate analytics and generate system alerts and recommendations can assist humans in making decisions and executing work. However, these systems rely on humans in many ways to:
- correctly install and calibrate sensors
- contribute thorough and consistent input on activities performed
- provide qualitative assessments,
- conduct common sense checks of system output, and
- provide timely and appropriate responses to system alerts and recommendations.
The IoT promises to take machine-assisted operations to the next level by closing the loop between sensors, communications and machine action. However, humans remain key participants in designing, monitoring, evaluating and troubleshooting systems. And, these tasks require changes in the way that people behave that may be viewed by staff as additional work, nuisances, or even threats. Automation often clashes with human nature.
For example, an oil & gas company spends millions of dollars on the management of Electronic Submersible Pumps (ESPs) responsible for extracting oil. Hard to unearth, prone to failure and costly when offline, ESPs are an attractive target for using sensors to predict and prevent failures, thereby reducing associated downtime and costs. The predictive models involved require quality data on a number of key variables, as well as information on previous failures.
The challenge is that the work practices designed for the old system are not working in the new predictive analytics era. Before, delays in submitting tear down reports didn’t matter much, as that data was primarily used for warranty claims and was filed away. Now, organizations want that information as close to real-time as possible to update their models. The .pdf format for reports used to be fine, but now users want the system to automatically read that data, requiring a format that is conducive to ingestion, which .pdf’s are not. The same is true with the old practice of keeping failure histories in local or shared excel files. While this served its purpose in the past, the organizations now require analytics systems to access this data, consistent descriptions, and auditable data trails. The humans involved in these processes, including staff as well as vendors, need to adapt their behavior.
To maximize the value and utility of these systems, organizations need to make cultural and process changes that are commensurate with the technology changes in which they have invested. These systems should be designed with guidance from and support of workers who know the job and the assets being monitored or automated. People must be trained to understand, at varying degrees of depth, what the system is doing and why. They should be empowered to question results or recommendations that do not make sense, and rewarded for the successes that the system helps to drive.
For the IoT to realize its full potential, we need to consider the impact of this new automated element on the humans in the system, and the impact of human decisions and behaviors on that system, at the onset of experiments. Navigating the human element needs to be a central component of an IoT strategy. Workers must be informed, trained and motivated to help ensure that IoT information flows freely through systems and is used to reduce the guesswork of decision-making.