The maintenance of things: asset management and the IIoT

The so-called Fourth Industrial Revolution is promising enormous advantages in plant and asset reliability and uptime by making it possible to predict the future condition of assets before they fail.

The transition from preventive maintenance to predictive maintenance will be enabled by increased amounts of data that is more accurate and that can be analysed to produce reliability predictions — avoiding unscheduled downtime and therefore improving the business bottom line.

The impact of the Industrial Internet of Things (IIoT) will be to help:

• enhance asset performance
• lower maintenance costs
• improve asset reliability
• enrich business performance.

The IIoT, utilising collaboration with big data and cloud capabilities, will have a significant impact on asset management. It will become possible to gain insights into the real-time health of physical assets. IIoT provides enhanced connectivity within the plant, with more smartly connected devices, while ensuring a cheaper, less time-consuming and smoother data gathering process.

IIoT concepts

The intention of the IIoT concept is to collect data from devices in the plant or in the field and then processes the data using sophisticated analytic software systems. Doing so is intended to enhance business processes and create operational improvements that were previously unavailable. However, achieving the transformational potential of the IIoT is not as simple as deploying so-called ‘smart’ plant technology. Getting significant benefits from the IIoT requires a collection of software, hardware, communications and networking technologies, all working together in a way that is optimised to the needs of the business.

The IIoT concept is based around the more loosely defined Internet of Things (IoT) that is making itself known in the wider non-industrial world. For industrial organisations, there is some advantage already inherent in the nature of the wired and networked automation technology that is already found in process plants and factories. Industry already has standards for the interconnectivity of smart devices: the question is whether those standards, and the current implementation of connectivity in our plants, can support the extra data connectivity and processing that is necessary to take advantage of the IIoT.

Currently the wider Internet of Things is in an emerging state, with no widely agreed-upon standard for systems, networks or interfaces. Multiple communications technologies are used and a variety of embedded intelligence technologies, as well as sensor and actuator solutions, are available, and there are as yet no agreed-upon standards of interoperability that take into account the higher-level applications that the IoT concept seeks to achieve. And in the industrial world, the IIoT has a unique set of requirements relating to operational, safety, security and regulatory issues that don’t necessarily apply in the wider commercial world.

Components of the Industrial IoT

The IIoT depends on four aspects or parts:

• So-called ‘smart’ assets
• Suitable data communications infrastructure
• Analytic systems and applications
• Interpretation of the data.

Smart assets can include instrumentation and sensors, equipment, machines, systems or other assets that include their own processors, memory and communications capabilities. Such assets will generate more data and some (such as robots) will eventually operate autonomously.

Data communications between assets and other entities will often leverage wireless network technologies such as IEEE 802.15.4 or Wi-Fi, in addition to industrial Ethernet and (to some extent) wired fieldbus technologies.

The additional data provided by smart assets requires that powerful new analytics applications be developed to take advantage of it. Information generated by these applications (hosted locally or more likely using a cloud infrastructure) will lead to new business models and better optimisation of assets and processes.

By having access to more and better data, organisations will increasingly make decisions based on the analysis generated by these resources. People will also continue to become better connected to others and to plant equipment, machines and systems through social and mobile tools and applications.

Asset maintenance models

Similar to the IoT, there are no generally agreed-upon models for the best asset maintenance strategy or methodology. Research by ARC has shown that there are many different versions of maintenance strategy models. Industry lacks a true standard to build upon, making it difficult to compare solutions — leading to confusion and delaying the application of solutions. A clearer interpretation of terms such as ‘condition-based’, ‘predictive’ and ‘prescriptive’ maintenance is needed to be able to clearly define where the IIoT comes into play.

The ARC Advisory Group has published an asset-management maturity model¹ in which the impact of the IIoT has been taken into account to classify maintenance maturity into five types or levels, as shown in Table 1: reactive, preventive, condition-based, predictive and prescriptive.

Table 1: Asset maintenance models.

Obviously reactive maintenance is an undesirable approach, unless the asset really is not consequential in any way for the business and its failure will not have significant impact.

Preventive maintenance is the more commonly found cyclic system of working that has been in place in process plants and factories for many years: lubrication of moving parts and battery replacements are simple examples, but this applies to any asset that requires regular testing, checking, calibration and potential part replacement on a semiregular basis (eg, safety valves, pumps and instruments).

In recent years, many organisations have tried to move toward a more condition-based (CBM) approach to minimise downtime that results from performing maintenance where often it is not needed, thereby reducing downtime cost and business impact. To do this requires the monitoring of a specific parameter (such as pump vibration or valve stiction) and collecting the data over time to establish a trend, making it possible to optimise the maintenance intervals for specific types of equipment. For static equipment such as tanks, boilers and piping, this is usually a regular manual inspection. Some equipment and software vendors have released systems specifically for logging and automating these tasks and providing predictive reporting or alarms.

Predictive maintenance can be seen as a more advanced extension of CBM, in which multiple variables and advanced computer analysis provides higher accuracy and more advanced warning to prevent failure. Advanced predictive maintenance analysis systems emerging today utilise a ‘digital twin’ model of assets and systems to compare measured and trending values with model-idealised values. Prescriptive maintenance takes this a step further with deeper diagnostics and guidance on repair and urgency.

Predictive and prescriptive maintenance are the areas where the benefits of the IIoT concept will be of most advantage.

Benefits

While predictive and prescriptive maintenance will require more engineering investment to achieve, they have the potential to create economies of scale that result in extensive business savings. That is, once particular devices and systems are modelled, and if those devices or systems are in extensive use, then the models apply to all instances in the form of a template.

Various claims have been made about the effect of organisations moving from preventive to predictive maintenance: some reporting up to 50% savings in maintenance labour and materials, and some claiming it is possible to achieve nearly zero unplanned downtime. Such savings have a flow-on effect in the business to better production, greater safety and greater customer satisfaction through on-time delivery.

How the IIoT can impact asset reliability

The key to the benefits of the IIoT are contained in one word: information. With an increasing use of networked devices, more asset information becomes available. If the information is structured in the right way, the information coming from a particular machine or class of assets, along with other process or environmental information, can be combined with advanced analytics to offer new opportunities for improving asset reliability and uptime.

Greater information content means creating the potential to identify specific components and failure modes more in advance and more reliably, with fewer false alarms. Being able to identify a problem long before it can cause a major failure will have a significant impact on business performance.

Types of data

When we read about the IIoT and big data, we come across the concept of structured and unstructured data. To these we should add semi-structured data:

• Structured data: Data (usually numeric) that has a predefined structure or model and can be easily organised, understood and interpreted. Examples from OT would be historian data and the data stored in CMMS systems, while in the IT world, spreadsheets and relational databases are good examples.
• Semi-structured data: Sensor-based data that is yet to be organised in a useful form.
• Unstructured data: Data that is typically non-numeric and does not have a predefined model, such as images, video files, Word documents and PDFs.

Existing automation control systems work exclusively with strictly defined sets of structured data coming from field sensors and instruments, and process that data for the purpose of control of the process and real-time information for operators.

Increasingly, additional data is becoming available from technologies such as smart field instruments that record more than just, say, pressure or flow measurements, but also provide information about the accuracy and calibration of the instrument, or of process conditions that may affect their accuracy. Additional sensor technologies are being provided within machinery (such as smart bearings and braking systems) or being added as additional sensor elements (such as vibration sensors). The data becoming available from field and plant technologies is now growing far beyond the immediate real-time needs of the process control system itself. Smart industrial assets often include an embedded computer, typically with processor, memory, operating system and communications capability.

Other types of sensing technologies can now also come into play: RFID technology in equipment, and in tools being carried by technicians, as well as wearable technologies that enhance worker safety or provide health and security functions by tracking worker location and exposure to hazards.

Some of the unstructured data mentioned above would come to the fore when creating a prescriptive maintenance system. Documents in formats such as PDF and Word (like operator manuals and specification documents) contain information necessary to prescribe setting and behaviours around the plant’s various technologies. Such information needs to be both accessible in real time for the technician in the field and also available to an expert software system that can use it to assist with maintenance modelling.

The resulting myriads of new structured, semi-structured and unstructured data are of no use without the networking, processing and analytics technologies to take advantage of them.

Horizontal and vertical communication

In the traditional communication process control system, data is exchanged ‘horizontally’ between controllers and field devices. However, IIoT integration means moving larger amounts of data in a ‘vertical’ fashion — that is, up through the data networking hierarchy to the business systems and the internet.

The trick here is to collect the extra data for vertical integration without impacting the horizontal operational activities and by not introducing cybersecurity risk issues though the process of IT/OT integration at the higher networking levels. The large quantities of additional data, from many devices, needs to be aggregated in some way and pre-filtered.

Different approaches to this problem have been offered in the marketplace, and the best solution may depend on the requirements of the plant. For example, some vendors propose a gateway solution in which the non-process control data is forwarded to a gateway for initial processing and aggregation before being forwarded ‘upstream’ — one advantage of this solution is that there need only be gateway devices communicating outside the control network, making cybersecurity more manageable. Another method is the use of more powerful and advanced controllers (PACs rather than PLCs) that perform both process control and IIoT data processing. However, if the only way to collect additional data in a brownfield site is to add extra sensing devices via a wireless infrastructure, then it will more likely be on a separate network in any case.

Whatever the method chosen, serious consideration must be given to all aspects of the network infrastructure that will support vertical integration.

Big data analysis provides more intelligence

Smart devices and improved connectivity lead to the generation of much larger amounts of data. As more things become smart and connected, the volume and variety of data that must be collected, managed, stored and analysed will only increase.

To gain benefit from big data, organisations will have to take full advantage of the advanced analytics solutions that are now becoming available. These analysis applications need to be able to bring useful, actionable information to the right people, applications and systems in an appropriate time frame and context, so that industrial organisations can optimise.

The main purposes of such data analysis are firstly to provide feedback for humans to interpret and act upon, and secondly to enable real-time logic to trigger the appropriate automated responses in mission-critical systems for routine processes.

In the early days of automation, reports simply indicated what happened, and it wasn’t until the 1990s that multivariable systems were developed that provided insights into why something happened. In the early 2000s, more intuitive dashboards began providing performance data about what was happening in real time. Now, as a result of the integration of more advanced data analytics technologies from the IT world, along with the greater quantity of available data, we are seeing a trend towards predictive analytics, offering the potential to predict the future to enable appropriate actions to optimise plant and business performance.

References

1. Rio R 2017, ‘How an IIoT-enabled maintenance maturity model works’, Plant Engineering, <http://www.plantengineering.com/single-article/how-an-iiot-enabled-maintenance-maturity-model-works/f2ecbee2bcfac8a5dd7b2bb9d7f056a8.html>

Image: ©stock.adobe.com/zapp2photo