Food production is truly a sector that operates under the mantra of “reinvent the everyday, every day”. The sector is constantly evolving, whether manufacturers are innovating new product ranges that meet changing consumer tastes or switching packaging materials to extend shelf-life or reduce waste. And these are just examples of substantial shifts; food manufacturers are also regularly making smaller challenges by refining recipes, adapting processes or adjusting ingredient and material supply lines.

Despite — or perhaps because of — the environment of constant change, food processors can benefit more than many other manufacturers from carefully targeted use of data collection, visualisation and analysis solutions. After all, yesterday’s optimisation isn’t particularly optimal if today means a new stock-keeping unit (SKU), a new critical ingredient supplier or a new recipe.

The approach that many businesses take to becoming data-driven is to extensively map out their digitalisation journey, with each aspect comprehensively planned. This doesn’t generally support the flexibility needed in food manufacturing.

Rather than taking this approach, modern solutions make it possible to build or buy micro-applications that share common data infrastructure and even app-building or visualisation tools. This means that impactful new capabilities can be adopted through fast initial works that create re-usable building blocks. Later works then become incremental, rather than potentially having different systems creating overlapping capabilities.

Micro-apps in practice

We can see how this micro-app approach can be put into action by considering one of the most common challenges in food processing: managing the effect of variability in key ingredients, so that yields are maximised with minimal re-work or ingredient waste. It’s likely that a manufacturer would already have some of the information needed to address the challenge. The question is, how can you quickly supplement what’s in place?

It’s a safe bet that the factory has automation and maybe supervisory control and data acquisition (SCADA) systems, so there is an abundance of machine-generated data to tell us about the details of how processes are performing. Focussing more closely on yield performance, we can assume our manufacturer has a lab system where in-process and finished good tests give very clear indicators of how well a product is being made.

From Novotek’s experience, the most common gaps in tackling yield issues come from two areas. The first is supplier quality data, which is often provided either written down or in an electronic format that doesn’t mesh with existing systems. This makes analysis more difficult, because there’s no actual database to work from.

The second area is that the variations in raw materials that affect yields may actually be within the specifications defined for those materials. As such, there may not be an obvious fix. It’s likelier that material data needs to be analysed alongside several process performance and quality performance data points. Understanding the relationships between more than two or three variables will probably mean adding a new kind of analysis tool.

Micro-apps can be highly focussed on the core capabilities required. In this case, the micro-app would provide three core functions. First, it would provide a simple means to capture ingredient quality data as it’s received, into a system that also holds the specific material characteristic specifications and limits – all on a “by-lot” basis. It would also offer a machine learning tool that can help clarify how the range of material quality variation can be managed in relation to what machine settings or recipe adjustments might allow for good final yield and quality results.

Finally, the micro-app would be able to alert production staff to make recommended changes to a recipe or process as different raw material lots are staged for use – an automated monitor of yield/quality risk from material variation. This could be as simple as a new smart alarm sent back to existing SCADA, or a notification on a smartphone.

Industrial software vendors are adapting their offers, in recognition of the trend towards micro-apps aimed at specific business processes. So, the software licensing needed to enable material data collection and quality specification monitoring on a key process would be built around a low user count and narrow set of underlying configuration and integration points, rather than a comprehensive plant-wide project. That can mean starting investments in the low thousands for software and some deployment work.

Some of Novotek’s customers are now progressing through projects defined by such very specific functional needs. Our job at Novotek is to ensure that any new solutions serve the purpose of being able to act as supplements to other such micro-apps in the future.

Next stages

A strategic advantage of micro-apps is that the planning and execution stages are less time-intensive than a far-reaching, plant-wide digitalisation project. Food engineers can do several things to begin reinventing their everyday processes. For example, food manufacturers can deploy predictive downtime applications on key processes. These are apps that can even take into consideration whether the products made have their own impact on failure modes.

Each micro-app reflects an opportunity to make the overall food manufacturing operation more adaptable. This means that innovation in products, processes and business models can be done, all the while knowing that refining and optimising the “new” won’t be held up by tools and practices that are too difficult to adapt from the “old”.

In the dairy industry as in many others, margins are everything. Business leaders always want their operations to be as efficient and effective as possible, with the highest possible uptime and the lowest possible operational expenses. This means low product wastage, high energy efficiency and lean processes.

However, dairy manufacturers and processing plants face an additional pressure. Raw dairy produce has a very limited lifespan, so it’s vital that it is treated in a timely fashion to prevent potential risks to public health. Unsurprisingly, regulation is very stringent on factors such as hygiene and the correct temperature for milk to be stored at pre-treatment. Each of these are defined in the UK’s Dairy Products (Hygiene) Regulations 1995.

This is one of the key distinctions between dairy processing and many other industrial segments. Improved process speed and operational efficacy isn’t simply sought after to increase throughput and profitability. It becomes an objective because it directly correlates with the safety of products. If raw milk spends too long at a temperature above 6 degrees Celsius without being properly treated, it becomes at risk of harmful bacteria growing.

Whether a dairy manufacturer is setting out to reduce their product wastage or lower energy usage, one thing remains a constant. No matter how many shiny new machines, automated systems or wirelessly connected widgets a business invests in, maintenance will be the key to getting the most out of hardware investments. In spite of this, it’s often the part of the process that many engineers begrudge, due either to it being time-consuming, labour-intensive or highly complex to maintain certain systems.

Digitalising maintenance

Fortunately, advancements in industrial automation systems over the past decade have gone some way to addressing the challenges conventionally associated with maintenance. As more industrial assets — whether that is the pump in a milk pasteuriser or the SCADA system controlling a packaging line — are able to connect with manufacturing execution systems (MESs) and enterprise resource planning (ERP) systems, dairy plant managers are presented with an increasingly reliable and scalable means of determining maintenance schedules.

This eliminates some of the guesswork from maintenance schedule, which means that plant managers can devise maintenance schedules that are more strategic. For example, a SCADA system might pick up a dwindling performance in a motor. If the SCADA system is connected to an MES, the plant manager can easily access this information remotely and make maintenance of that motor a priority. For perfectly healthy systems, they can be maintained only when their performance data indicates they need it.

The only downside to this is that if a plant contains thousands of connected data sources, it can simply be too much data for a plant manager to reasonably analyse to determine the ideal maintenance schedule. More advanced MESs resolve this problem, a good example being GE Digital’s Predix, which incorporates machine learning artificial intelligence (AI) into the system.

This AI can identify correlations and trends in data sets, allowing it to alert managers when maintenance should be conducted on a system. In effect, the AI learns the indicators of poor equipment health and can facilitate a shift to a predictive maintenance model, reducing unnecessary labour and time usage.

Although this makes life easier for plant managers, it does little to simplify the maintenance process itself for engineers. This is where AR comes in, by using the data within the MES and a purpose-built industrial internet of things (IIoT) platform to reimagine maintenance.

Bringing AR to dairy

A contributing factor to the complexity of maintaining some systems is a matter of design. Engineers need to know the most efficient and easiest way of accessing the components that need attending to, and this is not always an easy task — not least because it requires prior knowledge of the specific parts that are under-performing.

Take a milk pasteuriser for example. If the problem is that the flow rate of milk is lower than it should be, it could be a problem with the centrifugal pump, the valve or even the flowmeter measuring flow. With enough performance data from each of these parts, maintenance engineers can easily know which to inspect. And what better way to access this data than a digital overlay showing real-time performance data of each part?

This becomes possible with an industrial AR application like those available through PTC’s ThingWorx 8 IIoT platform, offered by Novotek UK and Ireland. This platform allows dairy engineers to build apps that are specifically designed for their plant and applications, ensuring that the app is suitable for any set up. Of course, the key to achieving this effectively is to work with an industrial AR expert like Novotek to ensure everything goes smoothly.

With that in place, maintenance technicians and engineers can either use AR headsets or their mobile phones to access the application. Simply by holding their phone up to a pasteuriser, engineers could see real-time performance data and could zoom in deeper to see specific parts. With a virtual representation of the pasteuriser’s centrifugal pump on the screen, engineers can inspect and identify if it is the part causing problems. If it is, the AR app can show the easiest way to access and maintain it.

And if a problem is particularly puzzling and the maintenance engineer isn’t sure how to address it, AR applications make it possible for specialist technicians to remotely view and advise on the issue. This encourages the sharing of specialist knowledge and improves the effectiveness of overall plant maintenance.

The value offered by AR is applicable to almost every connected system in a dairy plant. Let’s say that a dairy plant’s manufacturing execution system highlights that a rotary evaporator, used to standardise the dry matter of milk in the early production stages, requires maintenance. As the evaporator consists of several components, a maintenance engineer could use AR to see a virtual representation of the components in the evaporator and identify which needs attending to.

By using a purpose-built AR application, the engineer can view real time system data from the ThingWorx IIoT platform and see which components are performing inefficiently. In this case, it could be that the evaporator’s compressor requires lubrication. The engineer can then resolve this in the least disruptive way possible, minimising the impact that necessary maintenance has on production. Crucially, this technology maximises uptime and improves overall equipment effectiveness in the most efficient, effective and easy way possible. If a dairy manufacturer is looking to make their operations as lean and efficient as possible, AR seems like the ideal tool to help achieve precisely that.