Future of Manufacturing and Continuing Supply Chain Collaboration

“Without data, you’re just another person with an opinion.” W. Edwards Deming.

This year we attended FutureMSCE event, which brings together professionals from all Slovenia and abroad regarding supply chain collaboration. It was a great congress and there many points where we envisioned ourselves delivering value by solving supply chain problems. Through this post, we would like to present some relevant ideas and provide insight into our vision and how we are solving them through our platform.

Optimization planning

One of the most important problems companies face is accurate planning: the less error is introduced in expected demand, the better we can optimize the whole process and infrastructure, reducing stocks and associated costs. This is no easy task since due to imperfect information we suffer from the bullwhip effect: small fluctuations at the retail level may cause larger fluctuations in demand at a wholesale level. Wrong demand forecasts do not only affect stocking levels but sometimes introduce direct loss if requested materials are perishable or too specific to be reused.

Like any complex problem, it can be addressed from multiple angles. One of them is simplified product design by paying attention to common components across products as well as newly available ones. We may also consider available material sources and locations since this will have an impact on the overall production chain and choice consequences should be properly estimated. Sharing common components where possible reduces stock complexity and results in lower material losses due to non-demanded products.

An important factor in forecasting is the horizon for which we predict: usually, we observe that the closer the horizon, the better the quality of prediction. Thus company delivery performance (on time full process) is no longer a quality we aspire only to satisfy the client but also to reduce the forecasting horizon which translates into higher accuracy and savings.

When delivering a model, we depend on general context, industry-specific factors as well as on data available in the company. But overall good solutions can be developed to improve predictions as well as to assist humans in special cases that cannot be modeled.

Jobs metamorphosis: helping people through the digital transformation

While digitalization is increasing and driving us towards Industry 4.0 jobs may not decrease but requirements will change. Employees are required to have more analytical skills and a broad vision. Since repetitive tasks will be automated they may be engaged in wider decision making since a single person will hold more context but require less prework to get information and get actions executed. Digitizing all factory aspects may lead to new roles such as skills management and optimization, where we learn skills needed, best ways to transfer or teach them as well as optimal combinations of skilled workers through production schedules to maximize output and overall learning.

Towards smart factories

Smart factories aim to have a digital twin of the physical reality so that we can model possible scenarios and run optimizations providing feedback to the physical world regarding actions to be taken (optimal configurations in machines, schedules, skills, layouts; improve demand planning, etc.). Findings become reality through actions. Thus a good platform should have the following components: a link between physical reality and digital twin, usually through connected sensors, which provide real-time data:

an engine, which takes available data and outputs meaningful information, recommendations and actions

  • information can be displayed as insights derived from analytics and AI approaches
  • insights to action mappings should be learned so best actions can be taken. The platform should also learn who should take the decision about the execution: if a human, a recommendation will be presented; otherwise the action will be just informed. In any case, insights must be provided to data available at time and decision rationale that leads to recommend or perform the action.

Since reality is dynamic, criteria need to evolve and be re-validated on a continuous basis and whole platform AI capabilities need to constantly update to avoid staleness and errors derived from that.

Opportunities that evolve from digitalization are infinite and great gains will be obtained by applying human creativity to this cyber-physical world. At QLECTOR we are developing smart solutions for Industry 4.0 and always eager to recruit the best talent as well as partner with the best companies. Ping us if interested — will be happy to hear from you!