Efficient maintenance, repair and operations (MRO) processes are important in every industry to achieve the best possible return on assets (ROA). This is especially true for capital-intensive industries such as oil and gas, chemicals, mining and aerospace. For example, for an airline, MRO can be as much as 15% of operating costs.
MRO involves many functions in the organization, such as systems to gather, store and analyze data, managing documentation of procedures, resource planning, managing the supply chain, and scheduling deliveries to customers. It must also ensure that operations remain compliant with all the relevant standards and regulations for quality and safety. Increasing the maturity level of these processes reduces maintenance costs but also brings other business benefits.
This article looks at how the combination of sensor technology, the Industrial Internet of Things (IIoT) and advanced data analysis enables companies to improve their MRO strategies.
MRO Maturity Levels
The goal of MRO is to make the most efficient use of the resources in the organization to improve productivity and profitability. This can be achieved by minimizing downtime and optimizing the scheduling of work. Data from all parts of the system (for example, machines on a factory floor, or subsystems in an aircraft) can be used to monitor the health of the system and reduce the time and costs associated with performing maintenance that may not be required.
An effective MRO system, supported by IIoT technology, reduces the time required for maintenance by ensuring that it is only performed when required; for example, because there are early warning signs of component failure. More effective maintenance processes can extend equipment lifetime, which also contributes to ROA.
The maturity levels of an MRO process can be characterized as follows (source ARC Advisory Group):
- Reactive: Run until it fails and then repair or replace. Appropriate when failure is unlikely, non-critical and easily fixed. Example: car radio.
- Preventative: Service on a fixed time or usage interval. Used when probability of failure increases with asset lifetime or usage. Example: replace engine oil every 5,000 miles.
- Condition-based: Check for bad trends or other rule-based logic using a single data value. Includes inspections and manual data collection. Suitable for assets with a random or unpredictable failure pattern. Examples: oil pressure, coolant temperature or on-board diagnostics (OBD) indicators.
- Predictive: Use equipment-specific algorithms or machine learning. Uses multiple equipment and process data variables (multi-variate analysis) and, typically, automated data collection. Appropriate for critical assets where unplanned downtime has significant business impact. Example: battery management system in electric cars.
- Prescriptive: Identify issues and the repairs required using real-time data, cyber-physical modelling and multi-variate analysis. Use with complex assets requiring advanced skills for problem diagnosis. May need knowledge of process dynamics. Example: dealership-level diagnostic equipment.
The strict safety requirements and highly competitive nature of the aircraft industry have led to the development of novel business models and MRO processes. Other industries have different drivers such as cost or customer satisfaction that can also be improved with a more effective MRO strategy. More efficient and predictable production, without unexpected delays caused by system failures, improves product quality and customer satisfaction levels.
Even relatively small improvements in MRO efficiency can provide a competitive advantage.
IIoT/Industry 4.0
The technologies behind the IIoT can be used to support the goals of MRO by enabling continuous data gathering, analysis and communication.
Realizing that these technologies could transform industry, the German government developed a high-tech strategy called “Industrie 4.0” to promote the computerization of manufacturing systems. This was seen as the fourth industrial revolution, going beyond automation and robotics to create networked systems that exploit data fusion, communication and machine learning. This allows control and monitoring of subsystems to be devolved to local processing nodes. The concepts are now being applied to a wide range of industries beyond just manufacturing.
This uses technologies such as:
- Creating a virtual model of the system to simulate the effects of changes
- Using machine learning and AI to enable intelligent planning and autonomous decision making
- Implementing real-time communication between sensors, actuators, processing nodes and users
- Providing technical help to those managing and maintaining systems
Such systems can be used to support a move from preventative maintenance, which is performed on a regular basis whether there are any faults or not, to a more efficient data-based approach. This uses many sources of information about the system in order to implement predictive (determining that a failure is possible) and prescriptive (identifying the cause of failure) processes.
A simple reactive or preventative maintenance model can result in either unplanned downtime or scheduled maintenance of systems even when not really necessary. This can be a huge cost to the organization; for example, closing down an oil refinery can cost over 20 million dollars per day (and the shutdown and restart procedures themselves can take days). Providing more detailed information can improve the scheduling of MRO processes, and may also make it possible to avoid a complete shutdown by isolating specific subsystems that need work.
A data-driven methodology using machine learning and AI technologies can increase the process maturity level by predicting problems and even prescribing solutions.
Prescriptive maintenance makes better use of assets by minimizing downtime, keeping the system in optimal working condition, and reducing the costs compared to purely preventative maintenance.
One of the challenges of this approach is managing the large amount of data that is generated. For example, an aircraft produces several terabytes of data from the many sensors throughout the system; the engine alone may generate 5,000 data values per second.
Complex systems require many types of sensor, which have specific interface and processing requirements. Some of the sensor data will need to be processed locally, at the edge, in order to provide a real-time response. Other data can be sent to the cloud for more in-depth analysis using machine learning or AI algorithms.
The data can be used to detect early signs of potential failure well before they actually cause a problem. As well as predicting a potential failure, the system can also determine the reason. This information can be used to schedule maintenance. In an aircraft for example, data gathered in-flight can be used to order the necessary parts and schedule maintenance when the plane lands. This reduces the time that the aircraft is on the ground, which is an undesirable cost to the airline.
This requires continuous real-time monitoring of all parts of the system. This is made practical by IIoT technology, so every subsystem can be equipped with sensors, processing and communication capabilities.
IIoT Platforms
The number and variety of devices and subsystems that are being network-enabled as part of the IIoT is increasing rapidly. Many of these require new types of sensors and increasing amounts of processing power in order to perform real-time data processing at the edge.
Sensors are used to measure a variety of environmental parameters, ranging from temperature and pressure to acceleration and orientation. Many of these produce analog outputs, requiring signal conditioning and analog-to-digital conversion.
These requirements need to be balanced against power consumption limits and size constraints for mobile devices.
There are several wireless communication protocols available, each of which has different speed, range and power consumption characteristics. Choosing which to use depends on the requirements of the system; for example, the amount of data to be transmitted and the power budget. The table below summarizes the basic attributes of some commonly used standards.
| Power (Battery Life) | Data Rate | Range | |
| Wi-Fi | High (days) | 11 Mb/s to 450 Mb/s | Up to 100 m |
| Bluetooth | Low (weeks) | 723 kb/s | Up to 100 m [1] |
| IEEE 802.15.4 | Very low (months or years) | 250 kb/s | 10 to 300 m |
| [1] Bluetooth 5 increases the maximum range to 400 m | |||
Table 1: Wireless communication technologies
Wi-Fi is optimized for high-speed data transfer and is not suitable for very low power applications. It may still have a place in IIoT networks as the link between the system and the cloud.
New networking technologies, such as Zigbee, 6LoWPAN and others based on IEEE 802.15.4 support very low power data transmission. They also use a mesh topology, where every node can communicate with any other, allowing large networks to be created that are not limited by the range of the wireless protocol.
Meeting all the requirements for IIoT nodes can only be achieved by using a system-on-chip (SoC) solution. This integrates most of the required functionality on a single chip, including the CPU, analog sensor interfaces and wireless communication. Designing a custom SoC needs expertise in silicon design, verification and manufacturing. Most companies, lacking these skills, will choose to work with a third party, such as S3 Semiconductors, which has been developing such systems for over 20 years. S3semi has developed an extensive portfolio of interfaces and supporting technology for sensors that can be used to design custom SoCs using the SmartEdge platform; see Figure 1.
IIoT Platform Examples
As MRO processes have become more advanced, companies need to apply “big data” techniques to analyze the vast amounts of data from the many sensors in their systems. This is a new challenge for many companies.
There are now several companies providing services and tools for application development and deployment, management of IIoT nodes and communication, and data analysis.
In Japan Yokogawa Electric is developing an IIoT infrastructure to provide remote access to system data. An initial project is to monitor the status of pumps, made by its partner Iwaki, that are used in chemical and food plants. Operating data such as the current being drawn by a pump, discharge pressure and flow rate, and temperature will be transmitted to the cloud via an IIoT gateway so it is available to facility managers from any location.
Air France-KLM has created an Engineering and Maintenance division (AFI KLM E&M) to provide MRO support services to other companies in the airline industry. The company has developed a software tool called PROGNOS, which uses data automatically downloaded from the aircraft to predict failure of components. The results are transferred to the maintenance center, allowing the engineers there to plan any required maintenance. Initially developed for engines, the system is now being extended to other subsystems of the plane.
Conclusion
MRO processes are increasingly data-driven and can exploit IIoT technology to collect and analyze data. The amount of data that is available from sensors embedded in assets has grown dramatically. Innovative solutions are required to convert this data to information that can drive decision making and maximize the return on assets.
Large companies are now investing in IIoT hardware and software in order to move from scheduled maintenance processes to more efficient predictive and prescriptive models. A custom ASIC is the most effective way of integrating the sensor interfaces, communications and processing required for your specific application.
The availability of semiconductor foundries, a wide range of high-quality silicon-proven IP and specialized design houses, means that the custom ASIC route is within reach of most companies.
ARC Advisory Group Maturity Levels: https://www.arcweb.com/industry-concepts/asset-performance-management-maturity-model
Edel Griffith is the Technical Marketing Manager at S3 Semiconductors at division of Adesto. She has over 20 years of experience in the semiconductor industry in both R&D and technical Marketing roles. She has a Degree in Applied Physics and Electronics from the National University of Ireland, Galway and an Executive Diploma in Strategic Digital Marketing from Dublin Institute of Technology, Ireland.
