3 Steps to Truly Smart Manufacturing

Just 1% of the pie

Imagine your grandma bakes you your favorite pie. Would you content yourself with a thin sliver of it? Heck, no. If anything, the proverbial wild horses probably couldn’t keep you from inhaling it whole – crumbs et al. Why then would you treat real business benefits any differently in your digital manufacturing setup? Why would you stop at mere digitization, turning your back on the truly phenomenal advantages you can leverage from it?

According to Morgan Stanley Research, the manufacturing industry is the single global industry that accumulates the maximum volume of data – 2,000 petabytes annually since 2010 (two times as much as the government, which is the second largest accumulator of data). However, the industry has also been literally discarding an alarming 99% of its data since the same year, effectively sealing off any potential revenue-generating or cost/risk-minimizing opportunities it could have otherwise leveraged.

With the Industrial IoT having bequeathed us a self-sustaining oyster of sorts, it’s a crying shame to let so much of it go to waste – and also ironic considering all it takes is an improvement of 1% to have an annual impact of millions...something that’s possible when you apply the right analytics to your IIoT data.

Having your pie and eating it too

So how exactly does one manage so much data – and get smarter operations at the end of it, to boot? The answer is obvious – what you need to do is create a smart manufacturing setup. Only when a manufacturing process is in itself smart can you hope to achieve truly smart outcomes. So what are the components of a truly smart digital manufacturing setup?

1. The digital core: The first and most crucial component is the digital core – this refers to the big data that’s now part of every industrial setup. Till the advent of big data, factory managers never received information in real time – it was usually retrieved from databases, which by default meant that the data was old. Following the advent of IoT, data is now a commodity that no manufacturer can fall short of. The zillions of sensors installed – capturing every click, every movement, ever change – all of them tell a story. The life of an asset, a process, an entire operation. A truly smart factory gathers pan-organizational data from infrastructural (physical and operational) as well as human assets to digitally map their operations.

However, since less than 1% of all gathered data is currently being analyzed for business insights, most of it remains unstructured, creating or adding to existing clutter – and at worst distracting from more important functions.

2. Smart machines: Next, we need all the information collected to be communicated somewhere so it can be used either for analysis or to trigger related actions. All the data gathered from machine sensors, therefore, needs to be sent on to the next step of the process. This is what facilitates the transparency of the digital setup – the real-time visibility into operations. Thus, connected machines draw information from the physical world into the digital world and then send it back to the physical world to initiate appropriate action. As a result, there’s a continuous loop of autonomous, intelligent activity across the value chain.

3. Smart analytics: Finally, all the data gathered and communicated needs to be analyzed for business/ operational insights. This is the step that helps identify new opportunities for monetization – be it in the form of enhancing efficiency or productivity, or directly reducing costs. In essence, the massive data lakes gathered from the machine sensors are organized, structured, and analyzed for insights that can be turned into business benefits.

In a manufacturing setup, the most common approach adopted for this step is anomaly detection or the process of identifying outliers in a normal range of performance. The first step is establishing what is normal – which in itself is a variable given the dynamic nature of manufacturing conditions. Therefore, historical data, current data, and external influencing factors are analyzed simultaneously to establish baselines for asset performance on an ongoing basis.

When implemented with an unsupervised machine learning-based approach, this process can also help predict future anomalies, thus helping solve a whole hoard of manufacturing issues – a prime example being unpredictable asset downtime. Called predictive maintenance, this process involves gauging the mechanical status of a machine (whether or not it requires maintenance, or whether failure is imminent). Thus, maintenance can be scheduled at a time when the operational impact is minimal. This also inevitably helps reduce maintenance costs, as it helps avoid repairs – which tend to be much more expensive. Similarly, data can be leveraged to achieve a whole gamut of business objectives, ranging from reducing scrap generation to enhancing product output and quality.

So how does a smart factory operate differently from a normal digital factory?

While data collection, connectivity, and automation aren’t alien terms in the manufacturing industry, what’s different about a truly smart manufacturing setup is how these concepts are integrated and applied. For one, the data that’s collected in such a setup is actually analyzed and leveraged to facilitate intelligent operational decisions that save time, effort, and money.

Next, the links in a smart setup aren’t linear, unlike in the traditional setup – they’re interconnected to form a network. So the connectivity, in this case, facilitates holistic (horizontal and vertical) integration of processes, functions, and operations. Thus, the whole system becomes more responsive, more flexible, more agile in terms of adapting to changing requirements. A truly smart factory absorbs pan-organizational asset (machine, process, and human) data to drive informed decisions that help achieve business objectives.

Finally, the automation in such a setup isn’t of discrete, disparate manufacturing processes or operations – it’s the implementation of these processes. Traditionally, automated manufacturing processes meant binary tasks such as switching on a motor or turning off a chiller. However, a smart setup leverages machine learning and artificial intelligence to replicate human decisions – a process that involves the use of complex optimization algorithms.

A smart factory goes far beyond simple process automation and is, in fact, a self-learning, self-optimizing system that continuously adapts to real-time changes in operating conditions.

The Impact

The benefits of a smart factory setup are far-reaching – perhaps the most important one being that it leverages the power of 1%. The effective use of amassed data means that manufacturers can identify every available potential opportunity to cut costs, enhance efficiency, and drive profits – however “small” the opportunity. Also, a smart setup offers complete operational transparency – something that has only been an ideal dream till now. With complete visibility into operations, manufacturers can take intelligent, data-backed decisions to make the most out of their given manufacturing conditions.