PSI Blog

Chicken or the Egg – How to Perform a Root Cause Analysis in 3 Steps

02 Dec 2020 - Industry 4.0, Artificial Intelligence, Production, Technology

Chicken or the Egg – How to Perform Root Cause Analysis in 3 Steps
©AscentXmedia/iStock; maxkabakov/iStock & Hennadii/iStock; edited by PSI Metals

Recently a joker on Twitter came up with the following creative answer to the eternal which-one-came-first riddle: “I ordered a chicken and an egg from Amazon. I’ll let you know which one comes first!” Unfortunately, metals production is too complex for a simple answer like this. The production processes are meant to work together and build off one another to ensure the high quality of the products, and you cannot fully succeed in one without the successful foundation that another provides. Despite today's state-of-the-art technology, production defects can often only be detected during finishing – when it is already too late. So what if you could identify the root cause of defects and avoid them in the future and thus once and for all answer the question of which came first, the chicken or the egg? 

Quality losses are not only caused by the unreliability of production scheduling and planning, but happen due to the production itself. Learn how to identify the root cause by using your data efficiently and in a process-related manner in just 3 steps!

  1. Establish better reliability in production by collecting mass data – the combination of data collected using new technologies and a higher resolution of these data helps to stabilize product quality!
  2. Use a holistic way of defect mapping – model your data into something that you can really work with and that generates helpful information! 
  3. Perform a root cause analysis – it can help you identify the root causes along the entire process chain!

Step 1: Collect as Much Data as Possible!

Everything starts with the acquisition of production data along the complete supply chain in the Factory Model, from pre-material data to finished goods. Even if you have all the data for a single coil along the entire production process, this says little about the root cause of a defect. It is like the evaluation of last summer by the weather of a single hour. To maintain statistical relevance, it is beneficial to have data from many orders and materials. The more, the better!

PSImetals Advanced Quality Evaluation: The "Digital Twin" of a coil ©PSI Metals

Step 2: Enable Holistic Defect Mapping

Collecting lots of data is the starting point. The next step is bringing the data into a holistic process context, which gives the collected mass data a massive added value. Here we support you by making the data transparent while displaying it in the right context. The system shows, for example, at which production stage and material genealogy node the data was collected, where the exact defects are located including their position on the material and documents them by using images from the surface inspection system.

Visualization of highly aggregated mass data ©BFI

PSImetals can show you all measurements and defects from different production stages, which are projected into the respective coordinate up- or downstream!

With all these data you then can rewind in time and view historical data related to the material in that stage. It is like watching a movie – sometimes you want to rewind to better understand the current scene. It is the same in production, and it is in our system!

Step 3: Perform the Root Cause Analysis

Collecting process data on many materials with different characteristics, e.g. length and width, is of course not everything. You have to make them comparable! And here we use a little trick patented by the Research Institute and our partner BFI – we use the so-called “High Resolution Server” model. The newly developed model normalizes all these different and, from the first glance, somehow incomparable materials into a coordinated system (a normalized grid). Regardless of the actual length of the coils, they are thus mapped into a standardized coordinate-based grid. 

The grid has different resolution levels but usually the main level used is at the same time the highest resolution level. A certain number of cells in the grid is mapped to a theoretical material length, i.e. if the actual length is shorter than the theoretical length, it is stretched, if it is longer, it is compressed to fit into the normalized coordinate based grid. And of course, all defects are also mapped into the grid, which gives us the opportunity to reveal statistical relevance. You can compare it to the analysis of a soccer game by visualizing where on the field the actual game mainly took place.

Defects are mapped consequently to the corresponding location at other production stages (forward and backward) ©PSI Metals

In the next step, the model visualizes thousands of materials simultaneously and highlights the areas with irregularities by displaying coloured areas in the grid – the more irregularities found at this coordinate, the more intense the colour! Finally, it shows structurally the location of clustered anomalies in the products on the so-called “noise map”.


In this way, the information can be easily mapped backwards along the entire process chain and hence allows on any upstream line to correlate the mapped defect positions with corresponding process parameters.

So What?

The combination of this model with the PSImetals factory model allows you to use a wide variety of different types of measurements and to combine them all in a process context or with material genealogy. This combination enables you to keep track of the material and to follow all changes to the product and their effects on the material quality along the process chain. Nowadays, it is no longer sufficient to look at individual pieces. But using our algorithm, which aggregates your mass data by mapping it in a holistic model, makes you smarter and saves you money.

You may not know every detail about every coil, but you will identify the structural problems within your production, as the cause is often more of a structural problem that you can work on, improve or even eliminate. The “noise mapping” model is much faster and better at analyzing root causes because it uses aggregated data than traditional analysis, and offers a quick win for users because the data is already collected in the factory model and just needs to be applied in the right way.

Coming back to our “chicken or egg” problem, Elon Musk recently “closed the case” by posting the following picture on Twitter:

You too have the opportunity to rethink the analytics and thus ensure the best quality of your products and save money. PSImetals Quality offers an intelligent, flexible and fast solution for advanced quality evaluation, so that you can take a deep breath and enjoy the silence of a smooth production!  

Do you want to know more? Follow our Enjoy the Sound of Silence campaign!

Raffael Binder

Director Marketing PSI Metals GmbH

After taking over the marketing department of PSI Metals in 2015 Raffael Binder immediately positioned the company within the frame of Industry 4.0. So it is no wonder that in our blog he covers such topics as digitalization, KPIs and Artificial Intelligence (AI). Raffael’s interests range from science (fiction) and history to sports and all facets of communication.

+43 732 670 670-61