In our work with clients, we observe many benefits of shifting towards Smart Manufacturing. It offers a new way of monitoring production and getting valuable insights that help make more educated business decisions and becoming more efficient, both in the long and short term. The old way of managing production just does not fit today’s demands when it comes to product quality, traceability, customisation, environmental demands (both from consumers and governments), and optimal production processes.
Often the manufacturing process underlies internal or external obligations regarding the quality of the process itself and the produced part in particular. If it is detected that the production process is running under the quality threshold, it must be stopped immediately and adjusted to ensure the produced goods meet the end customer's expectations. The automobile industry is very well known for its high-quality standards. Another client of ours, a Swiss company in the field of sensor technology, needed to ensure through their software that every car part they produced complied with strict rules. One of the products they make is airbags. In case of a car accident or malfunction, all the production details of that specific airbag need to be available, even if the airbag was manufactured several years ago. The data that need to be available include production parameters, material used, production conditions (e.g. temperatures), and information about quality control (e.g. measured specification of that piece). In case
doubts arise about the safety of one produced batch, they can then find all airbags produced in the batch and recall them.
Smart Manufacturing brings the ability to optimise the production, both in the short and long term, be it reacting to changes quickly to avoid downtime or increasing the efficiency of material consumption.
One optimisation task of an MES system that most manufacturers already have in place is accessing and ordering information directly from ERP systems like SAP and calculating which machine will be used for the order and what parts need to be produced. An MES can also schedule and execute the orders.
In the long term, it is possible to implement more sophisticated optimisations like:
• reducing waste material;
• increasing the efficiency of material consumption by optimising the production processes themselves;
• adapting production parameters automatically in order to ensure quality standards;
• and improving employee assignments, which doesn’t necessarily mean fewer employees, but more skilled, better trained employees and a different nature of work.
Machine Learning also offers another large benefit of intelligent production: the ability to perform Predictive Maintenance instead of reactive maintenance. The main goal of Predictive Maintenance is to predict when a tool used in production will break down and to schedule its maintenance before this actually happens. This leads to a smooth production with fewer lastminute interruptions should a tool break down during production.
To detect defective tools, the production is monitored in real time alongside the historical data being analysed. Drops in quality can be associated with a given tool and indicate the end of a tool’s life. This can save companies a lot of time, energy and resources, especially in manufacturing.
Cost savings in Smart Factories can be achieved directly and indirectly in different areas. When we support our clients in their journey towards a smarter production, we analyse the current situation in the company, identify potential and easily accessible gains and help them make decisions that bring them the fastest and smartest way of how to become a Smart Factory. Here are some of the main areas we’ve identified that can see the most significant cost reduction:
The benefits of Smart Manufacturing can even be advanced by using AI and data science techniques. Whether it is about calibrating the parameters of a moulding machine, performing Predictive Maintenance or optimising long-term production, a suitable machine learning model needs to be trained on high-quality data before it can be deployed to production, which is a long term endeavour. How do you develop and integrate data science capabilities into your organization? Essentially, there are three options: buy an off-the-shelf data science solution, hire an in-house team to develop one, or outsource it to an external data science team.
Utilising data science capabilities
What option suits your organisation best depends on the following factors as outlined by the research company Gartner:
We typically work with tools such as Microsoft Azure Machine Learning and Google TensorFlow to support our clients with data science capabilities.
Although buying an off-the-shelf solution tends to be the least expensive option, it offers little customisation to meet specific needs. A custom-built solution will require a much bigger budget and more resources, but it will prove to be a more suitable solution for Smart Manufacturing transitions because each factory is unique in terms of processes and machines.
Complete outsourcing might be a good fit for companies that have robust data science capabilities in-house but a lack of time to allocate them to a data science project.
At ERNI, we adapt to your needs and help you analyse which solution is the best for you, taking scale, budget and purpose into account. We deliver solutions that are best suited for your company and integrate the proper data-science knowledge throughout your teams.