Using IIoT for a more energy-efficient pump environment

Industrial Internet of Things (IIoT) case study shows how to use pump power measurements to save energy using sensors enabled by wireless gateway and cloud-based analytics. Secure and relatively simple technologies were used; see eight project guidelines.

By Mike McClurg January 4, 2021


Learning Objectives

  • IIoT project uses wireless communications and sensors to analyze pump energy use. 
  • Energy maintenance savings are among expected results. 
  • Process manufacturers with many pumps could benefit from pursuing a similar project. 

The economics of improving pump energy consumption are compelling, and a case study shows how significant savings can be achieved. More than $250 billion is spent annually to power the world’s industrial pumps; another $100 billion is spent maintaining pumps to keep them running and improve efficiency. It’s not uncommon for a process manufacturer to spend upwards of 25% of an operating budget on power to keep pumps spinning.

Even so, industry solutions to address pump power consumption focus on new pump purchases. The goal for this project was to develop an architecture that bridges the gap between currently-installed environments for pumps and sensors and the emerging world of cloud-based machine learning. The solution enables long-term pump energy consumption data to be stored and analyzed, allowing decisions on pump efficiency, sizing, maintenance and replacements made from factual historical information.

Track pump power: 8 project guidelines

A few guidelines provided direction to the technologies chosen for implementation, including wanting to:

  1. Create a cloud-based digital representation of pump status, energy consumption and workload.
  2. Leverage state-of-the-art visualization and analytics tools, without the need to implement these locally.
  3. Have the ability to spin up and down test environments as well as to retain data for long periods.
  4. Work in today’s world (hands-off, remote monitoring, without a disruptive installation). Given COVID-19 induced restrictions manufacturers need a remote monitoring solution that doesn’t require extensive on-site staffing for implementation, monitoring and maintenance. The solution had to be simple to install and require no local staffing afterward.
  5. Avoid new sensors or proprietary software, leveraging existing sensors prevalent in the processing industry and avoid proprietary solutions that would provide vendor lock-in.
  6. Avoid wireless and power assumptions, such as availability of Wi-Fi on the factory floor, a low noise environment, or 120 V power. Many manufacturing environments are large, noisy and may lack reliable Wi-Fi capability. Most control environments don’t have 120 V power outlets.
  7. Avoid messing with the manufacturing execution system (MES), programmable logic controller (PLC), supervisory control and data acquisition (SCADA) system or information technology (IT) resources.
  8. Make the project secure and simple. That means having the data and control flow from the measured pumps without a mechanism to control or influence the process. By keeping this data flow in one direction, there’s no ability to alter the manufacturing process. Similarly, attaching non-disruptively to the sensor network eliminates the need for software changes, time consuming data dumps, or IT support for data acquisition.

Analysis of pump power data

Measuring pump motor power provides an accurate view of energy consumption. While this is great for instantaneous control of a pump (shutting off a pump in error conditions such as dry-running, cavitation or mechanical failure), it’s also valuable information when tracked over time. Longer-term analysis of power data can indicate increases that signal:

  • Increased workloads
  • Need for potential maintenance, including alignment and lubrication.
  • Under-sizing.

Similarly, decreasing levels of power could indicate potential oversizing of motors/pumps, or changes in the pumping process that need to be addressed.

Output from the power sensor is a 4 to 20 mA analog signal corresponding to instantaneous power.

The implementation attached a simple temperature probe to the pump motor. The temperature sensor also provides a scaled 4 to 20 mA analog signal. This thermocouple/transmitter was attached to the motor via a small metal strap.

IoT network gateways

The wireless gateway takes the 4-20mA analog inputs, converts them to a digital signal that is then transmitted via a long-range wide area network (LoRaWAN) using The Things Network protocols. The sensor gateway takes a snapshot of two analog data inputs every 20 seconds. Using LoRaWAN enables the devices to communicate over long distances (up to kilometers as opposed to tens of meters when using Wi-Fi) and in factory environments where interference from machinery would otherwise disrupt a Wi-Fi signal.

Wireless networks and data analytics

The Things Network (TTN) enables user-defined network topology based on low-cost gateways, with either inside or outside deployment. This deployment installed low-cost inside gateways, which were installed in minutes. The TTN network provides integration to stream data to a variety of analytics engines. Results are shown in Figure 2.

Data stream from pump motor

The implemented architecture built a continuous data stream of pump motor activity. Tracking motor power and temperature levels with time stamps created visualizations and analysis for:

  • Oversizing (% of motor capacity)hp measured versus faceplate motor hp
  • Increased energy usage where the area under the hp curve represents kWh or energy consumption.
  • Drift in power results. Increases in motor power and temperature for consistent workloads may indicate:
    • A pump installation issue
    • Pump wear or maintenance needs
    • System problems such as control valve issues.

Digital twin, energy cost savings

Next step in the deployment was using captured data and creating a digital twin, or virtual representation of the pump motor over time. Future capabilities will leverage machine-learning capabilities in the data analytics environments to automate anomaly detection and remote warnings.

This proof of concept can show immediate cost savings through improved energy awareness and motor efficiency information. Additional benefits of increased uptime and improved manufacturing quality will come as the data model builds and learning matures. All process manufacturers with large installed pump environments could benefit from pursuing a similar project.

Mike McClurg is chief marketing officer, Load Controls Inc. Edited by Mark T. Hoske, content manager, Control Engineering, CFE Media,

KEYWORDS: Industrial Internet of Things (IIoT), pump analytics


How many pumps should be supplying energy information and aren’t?

Author Bio: Mike McClurg is chief marketing officer, Load Controls Inc.