The Internet of Things (IoT) has forever changed the way we live our lives. With about 15.4 billion connected devices in 2015 predicted to double by 2020 and double yet again by 2025, IoT will undoubtedly become a huge part not just of our daily lives but even more so of various sectors. This includes the manufacturing industry. In a 2018 report, the International Data Corporation (IDC) predicted that by 2020, 60% of the world’s top manufacturers will rely on digital platforms for as much as a third of their income. The same report said that by 2021, 20% of these manufacturers will automate their processes and improve execution times with IoT.
Digitization within the factory floor has been heavily associated with setting up smart sensors in machines so that they can collect and communicate real-time data to optimize production and predictive maintenance processes. When these cost effective IoT sensors were first introduced, the manufacturing sector was eager to take advantage of its numerous benefits for brownfield operations. On the shop floor, smart sensors, some as small as a pencil eraser, do various things that include observing the conditions of the equipment, tracking machine performance, machine temperature and when paired with the correct backend system it’s also capable of sounding out alerts for potential problems, perform location tracking, and inventory control down to saving energy. Smart sensors are tasked to increase efficiency and productivity that in turn translated into higher revenues for companies. These favorable results from sensorization of the manufacturing sector have led some in the industry to believe that the way to go is to add more sensors. However, when enough sensors are in place and enough data is being generated and gathered, is there really point to adding more? Also are we even collecting the right data that can be scaled at a cost effective manner?
The truth is that digitization is a much murkier problem to solve. The key for brownfield manufacturing digitization is to not take the human element out of the process. The perception that IoT is a silver bullet that can help resolve all issues on the production floor persists among many in the industry. However, this is usually never the case because IoT is simply a feature of an entire system and should when appropriate, be an aid to ensuring safety, efficiency and traceability, but certainly not the end all be all. A combination of constraints such as the costs of sensors, feasibility of implementation and the simple fact that some processes just doesn’t have an available sensor in the market to measure is an indicator for many to put on the breaks to doing a massive IoT implementation.
The first best solution is to, if possible, pull data directly from the machine or equipment. For greenfield operations, this is a much easier proposition to make. New machines should now be able to provide most if not all the information needed for monitoring, maintenance, and analytics. Just ensure that when you purchase these machines the vendor has opened up access. In most cases, there shouldn't be a need to have to install more IoT sensors unless they are older machines.
Many digitization processes are most effectively—operationally and financially—done by a person keying it into a tablet, their phone, or some mobile device. IoT can help in some cases here if there is actually a solution available, but the cost of adapting this system is something that must be considered. For example, the cost of finding a sensor to determine whether a person is working in a particular area or not may be disproportionate to the data’s equivalent value to the company if someone just keyed it in on a mobile device. Simply training personnel to use a mobile phone to scan a QR code when they arrive in an area to track their timing would be way more practical and cost-effective. Of course this opens the age old debate of if personnel will follow this procedure or not but considering the cost vs. benefit ratio of implementing a full sensorized system to capture this, it may be more worthwhile to change business practices and incentives to encourage personnel adoption than to go through a massive capital expense purchase.
In addition, there will always be procedures and processes that cannot be sensorized and hence need manual inputs as well as apps and other digital solutions that workers can interact with. Although smart sensors essentially do the grunt work of collecting and delivering data, workers have to keep track of the data, interpret them, and act accordingly. Time and time again we’ve seen great IoT solutions rolled out but with no data scientist or a lack of cultural focus on real-time data being critical for reporting and analysis, be the catalyst for failure of an implementation. Even if IoT solutions are paired with AI or machine learning tool sets, it requires the entire enterprises’ mindset to teach, listen and interpret the data/system that is collected/recommending solutions. Without this, any system will at best achieve a fraction of the potential benefits.
Although IoT is indeed a helpful feature, it is not a perfect cure and should be used in the right context within a larger system for it to truly be scalable and effective. Deploying IoT technology also require integration with operations and other systems and is not a standalone solution. Otherwise, many of the technology developed will simply be trapped in pilot purgatory—basically it’s a great showcase when the board of directors walk by for a visit, but it will never be cost effective or practical enough to scale factory wide or enterprise wide.
IoT alone cannot solve problems, but it is certainly a powerful and helpful tool. It is the pooling together of different elements, human and machine/IoT, manual and digital, analysts and analytics, that can really help manufacturers create better products, make them faster, and maximize value across the board.