There are three generally accepted strategies for approaching equipment maintenance:
· Reactive maintenance, where problems are dealt with as and when they occur
· Preventative maintenance, in which scheduled maintenance is carried out in an attempt to prevent the most common failures, but since this does not catch all failures it is generally combined with some reactive maintenance
· Predictive maintenance, where the aim is to use diagnostics and other data to anticipate when failures will occur and take action to prevent the problem from even occurring
Of course the most ideal strategy is predictive maintenance, as this results in the highest equipment efficiency and the lowest maintenance cost in the long term. However, this is in fact the least widely used strategy, with most companies instead using a combination of reactive and preventative measures. The main reason for this is that while predictive maintenance is a wonderful concept in theory, it is not always as simple to implement and in the past required a high implementation cost.
The main factors that contribute to the complexity and cost are the need for a large amount of data in combination with advanced analytics capabilities. Equipment will behave differently depending on the environment it is used in as well as the actual usage patterns, so a wealth of data is needed regarding the equipment and its environment. Of particular interest in third world countries are conditions such as power quality, temperature and humidity, which can cause equipment behaviour to be highly different compared to first world countries. In the past, this data would need to be stored locally, and the infrastructure to manage this storage does not come cheaply. Furthermore, in order to accurately predict problems from this data, software needs to be available to perform intensive analytics. Until recently, lacking technology and/or unreasonably high costs of analytical software and sufficient processing infrastructure made such implementations unfeasible for the majority of industrial companies.
But with the emergence and fast adoption of IoT, as well as advances in data science, the intended potential of predictive maintenance is fast becoming a reality. By using IoT as an extension of existing SCADA monitoring systems, various benefits can be achieved which make data storage and analytics far more accessible, including:
· Distributed versus centralised system architecture
· Increase in visualisation possibilities
· Access to data for analytics
· Availability of extended data history
· Increased scalability and decreased cost
IoT as an Extension of SCADA
Until recently, the most common way to monitor and control equipment has been solely using Supervisory Control And Data Acquisition (SCADA) systems. These systems provide an interconnection of various sensors, computers and Programmable Logic Controllers (PLCs), which provides two main benefits:
· The ability to interact directly with equipment by sending commands to the PLCs
· The provision of a visual interface by which technicians, engineers and decision-makers can gain visibility into what the equipment is doing
However, since SCADAs are localised systems and are focused mainly on the current status of the equipment, the amount of data storage and long-term analytics is limited.
This is where the Internet of Things (IoT) can help, and its increasing use in equipment environments has in fact given rise to the term Industrial Internet of Things (IIoT). At its core, IoT has many similarities to SCADA systems, in that the focus is on obtaining data from various sensors and visualising this data. The main difference is the fact that IoT systems store all of the sensor and other data in the cloud as opposed to locally, which moves the main goal from short term visualisations to long term analytics. Therefore, by adding IoT capabilities to existing SCADA systems, the best of both worlds can be achieved.
Distributed Versus Centralised System Architecture
Since IoT systems are by definition based in the cloud, this provides a distributed as opposed to a centralised system. This lowers the cost of data storage since it is not necessary to have separate storage infrastructure at each site, and means that decision-makers can obtain a comprehensive view of all the equipment at different sites as opposed to one system in isolation. This is beneficial for determining which problems are unique to specific sites, and are hence a result of different environments and/or usage, versus which are common to the same equipment type regardless of where and how it is used. It also provides a form of security in terms of data backups – if something happens on site, the data is not lost since it is stored in the cloud.
The distributed nature also widens the possibilities of visualisation – certain levels of data can be made available to particular users without them having to connect directly into the SCADA and this data can be viewed not just on a computer but on any mobile device that has internet connection. Furthermore, the visualisations can be customised to particular users and made cleaner, more graphical and more informative through the web-based interface to facilitate decision-making. Enhancements such as 3D can also be used, which takes the visualisation to a whole new level.
With the data being stored on the cloud, it can easily be accessed from anywhere in the world in order to perform analytics and processing. This is key to following a predictive maintenance strategy, as it is the analytics which will make it possible to predict failures. Trends can be drawn that provide answers to the typical questions that equipment owners and manufacturers typically ask, such as:
· What is the efficiency of my equipment?
· How does this efficiency vary between sites?
· What is the average lifetime of my equipment?
· What are the most common failures that occur?
· What are the specific conditions and timelines that lead to these failures?
The distributed nature of the data means that these analytics can be carried out per site but also for the same equipment across multiple sites, which is critical to gaining a holistic view of failures. In some cases data from a single machine may not show obvious trends, but when aggregated with data from numerous other machines, patterns could begin to emerge. This empowers decision-makers to take steps to prevent failures that would not have been anticipated otherwise.
With the data science field constantly expanding, the insights available from analytics will only increase in coming years and are likely to extend from just predicting failures to developing usage patterns and ideal environmental conditions that will optimise the equipment’s lifespan and Return On Investment (ROI). In terms of forward planning, an IoT system is well suited to these coming advances since all of the data will already be on the cloud and available for analysis. This also means that it is not necessary to have dedicated analytics infrastructure and software at each site. The analytics can even make use of cloud processing capabilities, so it may not be necessary to purchase more powerful infrastructure at all.
The cloud storage makes it possible to view the complete history of a piece of equipment, and compare to other similar equipment. While some historical data can be viewed using SCADA systems, the centralised setup means that storage space is often a limiting factor and therefore only more recent data is generally stored. This is suitable for a SCADA system since the focus is on current data observation, whereas with an added IoT layer there is a shift towards analytics and predictions, for which historical data is a key component. This can be used to identify whether newer trends are emerging that did not exist in the past, such as whether a change in environmental conditions has made a particular failure more common, or whether equipment lifetime has been shortened after changing a particular component or service provider.
Scalability and Cost
While SCADA systems typically use proprietary communication protocols to receive data from sensors, IoT uses standardised protocols. This opens the door for more open communication between many devices and is far more scalable in the long term. With an estimated 30.1 billion connected autonomous objects predicted by 2020, and with manufacturing environments following the Industry 4.0 trend of moving towards ‘smart factories’, monitoring systems have to be designed to allow for seamless integration of numerous additional sensors in the future and IoT makes this possible.
Not only that, but it also achieves this at a relatively low cost rather than using more expensive proprietary hardware and software. With mini-computers such as the Raspberry Pi, a sensor network can be set up quickly and at an extremely low cost, and the cloud storage provides an additional cost saving. This makes it feasible for smaller factory or equipment environments to also adopt predictive maintenance, whereas in the past cost was a prohibiting factor.
As with any solution, there are negatives as well as positives, and there are certain challenges that IoT systems face, such as security concerns in terms of access to data and dealing with problems with internet connection, which means it is not yet the optimal choice for directly controlling equipment. With the rapid evolution of technology it is likely that solutions to these issues are not far around the corner, but the ideal solution is likely to combine both IoT and SCADA elements to develop a system that meets the requirements of both comprehensive monitoring and accurate control.
One of Looksee.do’s core focus areas is remote environmental monitoring, and since we are always abreast of the latest technologies, our solutions are all IoT-based. Contact us to find out how we can give you more visibility of your remote equipment.