Collect, optimize data

Use automation to collect pertinent operations data in the right format for worthwhile outcomes. See three data structure tips.
By Paul Figie, EZAutomation December 7, 2017

Data logging feature in the EZRack PLC from EZAutomation can structure and store data on a USB flash drive plugged into a CPU. The resulting .csv file can be opened in Microsoft Excel and easily manipulated. Courtesy: EZAutomationTwo common sayings about information are contradictory: "You can never have too much information," and "Information overload." It’s challenging to use data collection in real-time processes and to determine what data to collect and how often. A collective effort is required to make collection schemes that work.

It’s common to make data collection schemes that generate files that never get looked at, not because the information wasn’t useful, but because those who had access to that information did not apply the data in a way that helped the process. Other data collection schemes are used daily and make production and processes more efficient. 

Data perception, presentation

Data that isn’t understood or is misinterpreted may be useless, misleading, or even dangerous. Engineers need to have a good understanding of data collected and how it applies to the process. Data must be meaningful and applicable to increase its relevancy to those who are going to use it. It must be presented in a clear format. 

Data: Real time vs. historical

In today’s world, people are used to ready access to everything. Decisions are made on the fly. There’s no need to wait for historical data analysis to make a decision. Data and trends can be accessed through smart phones, tablets, and other devices.

Industrial programmable logic controllers (PLCs) can incorporate Industrial Internet of Things (IIoT) capabilities by using message queuing and the message queuing telemetry transport (MQTT) protocol, which can send real-time data to the cloud and relay it to smart phones, tablets, and other devices. Historical data points to past trends, helps analyze mistakes, and highlights corrective actions to make the process more effective. It takes time and effort to analyze and make a decision.

Historical data can be stored on PLCs that can use a USB drive and then used for later analysis. Combining real-time and historic data can help predict and take corrective action, which will help companies to be more productive.

Data collection in the automotive, oil and gas, pharmaceutical, packaging, and other process industries plays a vital role. It helps personnel of all departments know where they stand, what is their target, and what they have achieved. What if machine-level controllers like a PLC or a programmable automation controller (PAC) have a built-in data collection and storage feature? What if the same controllers could send data to a cloud?

The IIoT is emerging, and implementing IIoT in an appropriately equipped industrial controller will give companies a competitive edge by making data more available for decisions when needed. 

Collective data effort, goals

For effective data collection, assemble a diverse team and get them involved in the process of determining what data should be collected or how it should be structured. What’s obvious to some may be overlooked by others.

Set goals on how collecting data is expected to improve the process or increase production. Doing so helps with buy-in and drives the use of collected data from the beginning. 

Paul Figie is an application engineer at EZAutomation. Courtesy: EZAutomationStructuring data: 3 tips

Measuring production process improvement will keep team members accountable and make it easier to duplicate success and minimize failures. Data structure is important for effective use. Raw data collected over a period of time will be of little value, or may be too cluttered to be used effectively.

Three tips for data structure follow.

1. Data should not be nested. Instead it should be kept simple, perhaps stored in a .csv or a table format for ease of use and manipulation.

2. For different data types, such as Boolean, integer, or string, consider storing each in separate files to help simplify the process of data analysis.

3. Collected data should include an option to send to and store in a secure SQL database for future use.

Paul Figie is an application engineer at EZAutomation; edited by Mark T. Hoske, content manager, Control Engineering, CFE Media, mhoske@cfemedia.com.

KEYWORDS Data acquisition, DAQ 

  • Data can be useful in real time and in historical analysis.
  • Programmable logic controllers can help collect and store data.
  • Data structure matters.

Consider this

How often do you revisit what data is being collected and how it’s being used?