IIoT technology can often benefit industrial processes without a large capital investment.
James Wiczer, Ph.D. • Sensor Synergy, Inc.
The Industrial Internet of Things (IIoT) has long been viewed as a means of improving manufacturing in a variety of ways. Although there are many definitions and interpretations of IIoT, at the most fundamental level, IIoT technologies make key operational information broadly available. This information often includes sensor monitoring which spans a wide range of manufacturing metrics.
Real-world IIoT sensing applications capture metrics about a variety of parameters including energy use, process temperature, air compressor pressure, oil viscosity, motor bearing vibration, coolant pressure, natural gas flow rates, dew-point temperature, vacuum pressure, impurity concentrations, part count per minute, fluid flow rate, ambient humidity, ultrasonic intensity level, process gas flow rates, and several others.The rate of acquiring IIoT sensor measurements varies with the application and can range from one daily to more than a million measurements per month. For a typical project, the number of sensors deployed can range from a single sensor to more than 30 different sensors. The resulting measurement trends are frequently visualized in web browsers as graphs or tables with users selecting the sensor and time frame to view.
Manufacturers often perform operations in ways that differ significantly from those performed by others in the same industry. This uniqueness in “how it’s done” is an important ingredient in what allows a business to successfully compete with its rivals.
Because each facility has its own unique aspects, its IIoT solutions often require customization that comes at a cost. Management and staff may easily envision how IIoT can benefit their facility, but the customization costs may be a problem.
Even with an initial goal of “picking the low hanging fruit” to demonstrate the value of IIoT remote monitoring solutions, there can be significant hurdles for the IIoT project team. Projects must justify the funds necessary to develop and deploy a solution despite the promise of important benefits like energy savings, improved resource usage, better equipment failure prediction, and improved remote workforce safety.
Sometimes it is tough to prove the worth of an IIoT integration to upper-level management due to the uncertainties of the benefits. In some cases, the true benefits of IIoT may remain unknown until operating data is available and analyzed.For example, suppose an IIoT project requires adding a sensor to a piece of manufacturing equipment and corporate best practices encourage tight integration into the existing controls. There may be issues with, and substantial costs from, integrating this new sensor into the equipment’s existing PLC. This may be an especially difficult hurdle if the addition of this sensor hardware and changes to the PLC software require the modified production equipment be re-commissioned and pass another round of factory acceptance testing.
Once the hurdles of adding hardware to an existing piece of manufacturing equipment are cleared, another challenge can be integration of the new data into manufacturing information systems (MIS) or the higher-level, corporate Enterprise Resource Planning (ERP) systems. The same challenges exist for the cost and security associated with storing data, displaying a visual representation of complex sensor data, and securely alerting key stakeholders about out-of-normal conditions.
Uncertainties in manufacturing environments can lead to uncertainty about the magnitude of the expected financial benefits. Consequently, many seemingly viable, high-quality, IIoT projects never get the green light.
It’s fair to ask whether measurement data from IIoT can be useful without tight integration with MIS or ERP software systems. In many cases, the answer is yes. And in general, it is reasonable to expect lower development and deployment costs for IIoT solutions not tightly integrated to existing systems.
IIoT data analysis can range from simple to complex. A simple approach with real-time online tools may reveal underlying relationships like, “We spend $47/hour on energy to idle this machine on the weekend.” In contrast, complex analysis may add contextual information to get higher-level insights like, “Last month our second-shift crew made 7% more product while using 19% more energy with 5% less down-time compared to our main shift.” In these real-world examples, both results arose from use of relatively simple, stand-alone IIoT hardware/software. But the second example required integration of external data streams during the monthly analysis and reporting efforts. IIoT that is more tightly integrated with the MIS may simplify the task of making complex analyses of measured data.
Contextual factors can supplement our understanding of raw IIoT data and help transform it into actionable recommendations. Physical context refers to information about a physical sensor including calibration factors, operating parameters, and sensor state-of-health. This type of context information may include the physical location of the sensors, properties of the equipment or process being monitored, and details of the manufacturing processes.Economic context refers to financial issues. These can include the cost of the energy a motor consumes, the opportunity costs of shutting down a process, the damage when equipment used to condition the environment for a down-stream process fails, or the economic implications of a missed production target during a shift.
Temporal context includes historical information about events affecting the equipment or process subject to the IIoT monitoring. Temporal context can come from discussions with key stakeholders or analysis of past electronic data from other sources. Historical data can include information about temporary production halts, incomplete staffing during a shift, failure of up-stream equipment affecting the process being monitored, or an unusual event that interrupted production.
For example, consider the average cost of electricity for a facility. It may be expressed as dollars per kilowatt-hour ($/kWh) and represent the economic context needed to interpret IIoT power measurements. Holiday plant-closing hours may be enough to help explain anomalies in IIoT data two months prior. And knowing that the ac current sensor generates “one additional volt of signal for every 100 A of additional current flowing through the sensor” may help interpret sensor signal measurements.
Some of this information may be directly available in a facility’s manufacturing information system. But it may also be entered directly into the setup software of a less expensive, stand-alone IIoT system. To evaluate which is the best approach, the costs for data analysis must also be considered as well as the difference in costs of fully integrated vs. stand-alone IIoT implementations.
Initial IIoT projects
In most industrial facilities, the process of transforming a facility into a smart, remotely monitored, manufacturing “plant of the future” may not always be straightforward, despite many “obvious” benefits.
It may be better to pick an initial project that only requires limited data interpretation or straightforward sensor networks. Consider an IIoT project that monitors electric power consumption. The benefits should contribute to corporate goals like reduced energy costs, minimizing equipment downtime and maintenance, enhancing operator safety, improving operational efficiency, or other important concerns.An initial IIoT effort may be able to exploit simple-to-install, low-cost, stand-alone hardware and software. Think of it as, “Walk before you run.” Once financial benefits have been demonstrated, a more tightly integrated approach may be easier to justify. If the expected benefits do not materialize, knowledge about what does not work may more than justify the relatively low costs.
One often-cited application of IIoT is predictive maintenance. The system predicts future failures based on sensor measurement trends that diverge from normal operating conditions. But predictive maintenance is complicated. It relies on weeks or months of sensor measurements during normal operating conditions to make predictions. Typical sensor measurements used for predictions may include vibration levels, temperature, power quality, lubricant status, filter status, and similar factors. All in all, predictive maintenance may not be a good starting place for demonstrating the benefits of IIoT–especially if in-house data analytics personnel are burdened with other projects.
Our company has found that often the most impactful and profound benefits of IIoT projects were unanticipated. In essence, these unexpected benefits arose from insights about how operations take place on the factory floor–which were not according to “the plan.” Such discoveries typically don’t come out of ERP or MIS systems, which tend to expect known data streams and search for known features in the data.
It is noteworthy that in approximately a third of our IIoT projects, we discovered an unanticipated effect that dominated the process. Synergy often arises from analyzing two or more data streams, synchronized in time, describing different aspects of the same process. This synergy helps identify not only the anticipated results but also the surprise benefits.
For example, sensors at a foundry indicated that one of five 250-hp vacuum pump drive motors unexpectedly failed, but the vacuum pressure in the main manifold did not fluctuate.The original purpose of the IIoT installation was to monitor electric motor power consumption. But the unexpected pump motor failure (from a broken belt) caused us to further investigate the factory’s requirements for vacuum pumping equipment. Additional contextual information revealed that safety valves were designed into the original vacuum system to keep vacuum pressure above a certain level. These valves automatically let outside air into the system to prevent the vacuum pressure from being too strong.
The actual load on the vacuum system required only one or two of the 250-hp motors. The others could be turned off. These three “extra” motors had been running for about six years (8-12 hr/day) before our measurements revealed this “overdesign.” During those six years, the three extra vacuum pumps created extra low pressure in the system. The low pressure was mitigated by allowing ambient air into the system as fast as the three extra pumps removed it!
The waste of energy and financial resources was significant. The cost of the electricity alone exceeded $150,000 per excess motor during the six years of operation.
In another example at a different facility, we discovered that separate heating and air-cooling systems ran simultaneously for three hours during the spring and fall! Relatively simple monitoring again brought a change that saved the plant thousands in energy costs.In yet another case, we discovered that almost a quarter of furnace-heated crucibles at a brass foundry were abruptly halted before completing their 20-minute heat cycle. The crucibles then cooled while other higher-priority crucibles with different alloys were heated in the same furnace. Much of the energy used to heat the low-priority crucibles was wasted.
In several other instances, IIoT revealed operations personnel didn’t fully understand their equipment. In one case, the staff thought a 4-MW furnace was intended to hold a melt at 40% of full power. But the system was designed to hold the melt with only 20% of full power. The problem was that the word “hold” meant different things to management and staff. This disconnect had gone on for three years! Our real-time, IIoT electric power usage measurements exposed the issue.
In its simplest form, an IIoT system collects sensor data from the factory floor, transmits the data to a computer server, and then makes this sensor data available for human viewing and computer analysis. The goal of IIoT efforts is often to make a manufacturing process better with some actionable message for the stakeholders– based on the measurements.
Typically, data-acquisition electronics collect the sensor measurements and transfer the data to a nearby edge computing device. The edge computer formats the data and sends it to a network – wirelessly or via Ethernet connections. A server on the same network receives and processes the data to identify out-of-normal conditions. If necessary, it issues email or text message alerts. The server can be a cloud computer or located at the manufacturing site.
Many real-world IIoT projects are more complicated than the simple explanation provided here, but they incorporate the same key elements. The approach they take typically depends on the skills of the in-house engineering staff and the unique technical requirements of the task.
A do-it-yourself (DIY) approach may be possible utilizing COTS components. Relatively inexpensive single-board computers with easy start software packages can bring data from a physical sensor to an ethernet or wireless port. Several software services can accept sensor data and allow stakeholders to securely display it in browser programs. These software services also allow IIoT users to set up alerts to highlight out-of-normal measurements and provide data backup services.
The manufacturing facility may need special data routers depending on the number and location of sensors and the availability of ethernet or wireless data networks. These routers incorporate extra features for use with manufacturing IIoT and can include rugged physical construction, extra security, mesh networking, battery backup, connections to other networks, compatibility with various data formats, and others. Depending on the facility’s security requirements, wireless networks might not be allowed. Other limits may apply to sending data off site or to cloud-based computer servers. Even if security is not a major concern, the electrically noisy manufacturing environment may not be friendly to wireless communications.Relatively simple turnkey approaches may be best for initial IIoT installations involving a lot of specialized technology. Some of these solutions can be purchased for under $2,000 depending on sensor and data collection requirements.
As an example, Sensor Synergy has developed IIoT solution kits applicable to common manufacturing issues. A representative kit includes power supplies, cables, sensors, data acquisition electronics, edge computer, cloud computer server access, and pre-configured setup files. The goal is to enable easy setup, install, and basic data viewing within an hour.
Some manufacturers prefer to keep data in-house and not connect to cloud computers. As such, they may favor approaches that offer both options. Various software and hardware features can make a packaged solution viable for many applications and include email and text alert notifications for cloud-based installations. Again, as an example, the Sensor Synergy systems can measure up to four analog sensor signals from a wide variety of sensor types. DIY sensor installation and phone-only assistance keeps costs down and speeds time-to-completion.
In addition, other providers also offer IIoT packages for significantly under $10,000. Costs can vary depending on the quantity and types of sensors, the amount of in-house effort available, and the networking issues at the facility.
Even the most promising IIoT projects need to be observed over time to experience the full benefits. In some situations, low-cost, low-risk, stand-alone IIoT technology may improve manufacturing processes and pave the way for more ambitious IIoT projects.