Data-driven operation in industry

What decides which manufacturing data are worth collecting?

Sándor Gáspár: “The most important point is that the client should define the data collection and analysis objectives already at the outset. Data storage is no longer expensive nowadays, but data processing, analysis, and integration into processes can be quite costly, which only pays back if we know the exact objectives in advance and can define the expected benefits accordingly. Together, we usually review which business indicators are not performing well, and from this, we can identify which data can help us move forward.

This indicator is determined by data such as the mechanical availability of machinery in production, capacity utilisation, and production quality. Companies are already aware of the importance of measurement, and, when procuring new machinery, it is always important to ensure that the data required to obtain these indicators can be extracted efficiently from the equipment. The design and commissioning aspect, however, that in addition to generating the three main indicators there should also be a possibility to measure additional parameters, which allows more detailed analyses and answers to the ‘why’s and the OEE improvement opportunities, is less frequently addressed.

It is important that the development brings positive changes in the performance of the company as soon as possible, as this can reinforce the need for further similar development needs. Our methodology starts by looking at the indicators that describe the performance of the defined area of operation, then considers the processes that drive them, and attempts to identify what is measurable in these processes or what prerequisites are needed to make them measurable. Overall equipment effectiveness (OEE), for example, is a fairly important metric for any manufacturing company.

Why does it seem more difficult to digitalise industry than banking or commerce?

Sándor Gáspár: “In the case of manufacturing companies, we often have to start from quite a distance, because IT often stops at the boundary drawn by the performance of system administration tasks, such as the support of existing IT systems. This is therefore an issue of organisational maturity; a customised Industry 4.0 project cannot be approached as "deliver me the turn-key solution because production takes up all of our time". As I mentioned earlier, it is not possible to be successful without the formulation of objectives and active participation. Integration is also a big challenge, because it is often difficult to develop physical processes to do the job that the AI had set out based on the data at hand. For service companies, it is obviously easier to transform processes because of the higher level of digitalisation.”

Péter Géró: “It is no coincidence that modern IT first appeared in the administrative workflow of industry, and on the production line we at most find similar systems in machines for remote maintenance, but networking is not carried out in many places for the reasons mentioned above. Where there is a higher level of production line digitisation, often the full-fledged completed solutions of foreign parent companies are adopted by domestic subsidiaries. Domestic industry has therefore not yet accumulated the competence that is an everyday commonplace in other sectors.”

What are the levels of digitalisation of a company's activities?

Sándor Gáspár. “There is a classical approach to the digitalisation of manufacturing companies that distinguishes three levels. The first is the level of corporate governance, where demand is coordinated with procurement, production and delivery to optimise the cost of raw materials or the inventory - this is the classic ERP level. The second level is production, the MES level, which includes production planning and -execution. Here, orders are broken down into production plans to maximise productivity. The third level is the SCADA level, where we talk about the settings of individual machines; reducing scrap, scheduling maintenance, and balancing energy consumption at plant level.

You can't really skip these levels, so I would definitely consider the ERP level as the entry level. We need to see the whole picture of the company's operations to penetrate to lower and lower levels with digitalisation. Where there is already pressure to see not only the past but also the future, we provide predictive analytics solutions - predictive maintenance is a classic example, but nowadays it is also very useful to analyse problems in the supply chain, for example, to predict delays in deliveries. Here, the decision is still made by the human operator based on a proposal derived from the analysis results. But other processes can be fully automated. One such area is quality control, where artificial intelligence can screen out and automation can subsequently remove rejected off-specification scrap from the production process.”

Péter Géró: ”The industry is very diverse in terms of where each particular client is now. Often, at the ERP level, we already see very advanced solutions, so decisions are based on factual data, but orders are still submitted to the company by e-mail, weekly and daily production planning is done in Excel, or stock levels are updated daily in Excel worksheets too. So, the company still needs to pull data from different systems and software to operate, and too much depends on the human factor. Most of the inquiries are now submitted to us at MES or SCADA level for data-driven automation.”

All modern manufacturing devices can collect data. What makes it difficult to turn this into business value?

Péter Géró: “It is a typical observation that the coordinated operation of various equipment is not solved, so individual pieces can only optimise their own running at best. Originally, manufacturers of production equipment built data acquisition systems into machines to provide information on the status of the machines themselves. These data can often be used for other purposes as well, but some manufacturers may even refuse to give access to these systems, or we are talking about software with so different incompatible approaches that are very difficult to link together efficiently. Unfortunately, the way to eliminate these access and integration barriers is not known locally by manufacturing companies, so they cannot use their data to significantly improve their operations. But there are solutions, and we know and propose them.

In which cases should we choose scalable or ready-to-use solutions?

Péter Géró: “For some time now it has been a well-known design method to build any solution on a scalable platform with an easy-to-adapt interface, and the importance of doing so has even grown over the years. Based on client requests, we often build on open-source foundations, which have the advantage and purpose of creating flexibility. Plug-and-play systems are also gaining ground in digitalisation, and there are many such MES systems, or even quality control (AOI) or warehouse management software. We also use similar ones, but given that most of them are rule-based, you can reach the limits of their capabilities quite quickly. “Off-the-shelf” solutions are not our main strength, Stratis rather delivers customised solutions tailored to the business and the specific task, with full responsibility.”

What is more important in a decision: The cost of acquisition or the return on investment?

Péter Géró: “It depends on management culture, and the picture is rather multi-faceted. The problem is that much of domestic industrial investment is dependent on tender invitations, which in turn are centered primarily on the cost of acquisition, whereas long-term competitiveness is created by the total cost of ownership. There is a willingness to look at investments on a return basis, but in the case of digitalisation improvements, the returns are more cross-cutting and harder to capture than for a machine tool. Consequently, the total cost of ownership is also more difficult to estimate. The solutions we deliver typically involve a return on investment period of 1.5-2 years, which is why we help clients with the economic calculations at the beginning of projects.”

When is failure inevitable?

Péter Géró: “I have to refer back to the beginning of our conversation. If someone can't articulate the goal s/he wants to achieve, there is a good chance that failure is inescapabée. So we take the first steps together, where Stratis acts as a consultant. Projects based on a jointly shaped digitalisation development plan have a good chance of success on their own, but as we are talking about a set of interdependent projects resting on each other, it is not after the first step that you should expect to see truly breakthrough results. Many people only realise in retrospect that digitalisation requires different skills from staff members and managers alike. If the employees concerned are afraid of using the new system, perhaps sticking to the so-called "tried and tested" methods and sabotaging the changes, then again, failure is unavoidable. It is therefore not possible to jump several steps at once, and the system introduced must also be adapted to the level of maturity of the organisation. As the industry is also characterised by high fluctuation, an unwieldy system will not deliver the expected benefits. We are in a race to remain competitive and the environment is changing dynamically. A business can be very fragile if it sits back, relaxes, and does not evolve digitally, so that after a while it does not have enough factual information about itself, its customers, or its competitive environment. A project may end, but the development of a business will never be complete. Digital development can never be finished, only temporarily stopped at most.”

Author: László Molnár

Source: jovogyara.hu