Agfa plant in Leeds has developed into a best in class manufacturing facility based on the principal of using data to eliminate waste from its manufacturing process.  But it was not always like that!

 Last year, I met with Graham Cooper Site Manager at Agfa Graphics Leeds Production facility and his Production Manager Charles Meldrum. The discussion turned to what does Industry 4.0 mean, or Digitisation of manufacturing really mean? Their answers were so illuminating and humbling, that we wrote up a case study in the Manufacturer magazine.

Agfa’s data story started in 2001 when they faced a major corporate re-organisation. No longer would the Leeds plant sell the lithographic plates which lie at the heart of modern printing presses. They would now be selling a semi-finished product that would be finished elsewhere in the supply chain. Up to then, their main method of validating quality was to do destructive testing of the finished product. This approach was no longer possible, and they needed to find new ways to manage in-process quality.

First, they reviewed their individual processes, creating control plans as required.  However, faced with a very complex manufacturing process with many interdependent control parameters, this approach was not good enough. They knew their plant operations had over 8000 digital inputs but had never really used this resource. Now that their survival depended on becoming a benchmark for ’lean’ production, their first instinct was to find a way to view the interactions going on in their plant.  Using some basic visualisation tools that already existed in the Distributed Control System, they began to visualise how the material flowed through the plant and how key performance variables such as energy, flow and acidity varied along the manufacturing process.  To their amazement they started to see patterns in the plant operating characteristics, which not only allowed them to manage their process, but to also identify extremely obscure problems that had always existed, but the route cause never identified.

As their understanding of the plants operation deepened, they started to be able to identify problems and advise their sister plants in the downstream operations. They also began to identify and solve problems in their upstream supply chain. For example, they were able to identify that one supplier’s material handling equipment was dropping the aluminium coil in the same place, and damaging it, such that the manufacturing process was influenced. The result was that their relationship with external suppliers and internal customers became much closer and more inter-dependent. Technology and data allows a far greater understanding of the value stream, which is a well-articulated second principle of Lean.

More profoundly, their culture moved from ‘How do I stop that fault from happening’ to ‘How do we understand what happened, how do we get the process under control, how do we eliminate the problem’. Their teams managing the line went from being ‘machine minders’ to ‘process optimisers’.  The key to their success has been to present data in a very easy to use and visual format that their teams can easily action and improve.

And now that a data driven problem solving culture is in place, the Agfa team are going back to the machine learning analytics technologies that they first started to investigate in 1994. The difference now is that their level of maturity in understanding data is at a completely different level.

So why is this story so humbling? I think the answer is that it is so obvious!

Start with the business problem, figure out how to use data to solve the problem and present it in an easy to access format for your people to action and improve. In many ways it goes back to the now tried and tested Lean tools such as the Lean business improvement cycle of Plan – Do – Check – Act (PDCA), which is still highly relevant to solving business problems. It is data that ‘oils’ the improvement process.

So why in today’s world are we making it appear so complex with fancy jargon and words that appear to mask common sense! This is one of the reasons that Si2 wanted to work with Data Science experts,  The Data Analysis Bureau on a series of round tables held last year as part of the Industrial Data Summit Conference in the UK. We wanted to help manufacturers cut through the hype and understand how to move from Business Problem to Data Solution. Talk to any Data Scientist and they will tell you that the most frustrating projects are where they are asked to find patterns in data without any indication of what to solve. It is like asking, how long is a piece of string?

The feedback from the round table discussions was fascinating and identified 3 critical steps:

1.     Articulating the Business Problem

Almost without exception, we all concluded that it is important to start with the business problem area in mind. It is also the first principle of Lean Manufacturing is Value: “specify value from the customers perspective”.  When working with data, it is important to not only understand the business value in your immediate area of responsibility, be that the factory or customer, it is important to look across the whole value chain from component to end user. Often data in one area of the chain, can have significantly more value associated with it in another part of the ecosystem. We also discussed looking beyond the immediate product or service but try to understand your potential impact on the Total Cost of Operation.

In the discussions we observed there were varying levels of completeness of the business problems described. However, in nearly all cases the orders of value (i.e. pounds!) were often missing. Understanding the relative values of the different components of the Total Cost of operation is critical to identifying priorities.

2. Define the Data Problem

The next challenge we discussed was how to turn the business problem into a business data hypothesis. This would describe an expected or speculated relationship that we hope to determine through the analysis of data. For example, the hypothesis for a predictive maintenance solution might be: ‘We can identify the failure patterns for hydraulic system as well as general machine performance using pressure, oil contamination, temperature and humidity data from the PLC, such that we can predict failures and recommend corrective actions’.

Why is this important? Data Scientists cannot tell you patterns that interest you without knowing the area of interest! Hence converting the business problem into a hypothesis is a key part of the process and applying the scientific method which is question led and iterative.  But the hypothesis does not have to be correct. It is very likely that it will change as more knowledge is gained about the data being analysed or definition of the business problem evolves. One must expect a certain amount of iteration from business problem to data problem as our knowledge expands, and this in turn helps deliver optimal business value. The next step to defining the data problem is to understand your data maturity and approach. We do this using four parameters, also known as the four Vs:

Volume: How much data do I have to analyse
Variety: How many different data types am I looking at
Velocity: What is the speed at which the data is being acquired
Veracity: How accurate, complete and robust is the data

The 4V’s together with the Business Data Hypothesis will help us define the most suitable analytical techniques that we should consider using.

These first two steps form the crucial “plan” element of the PDCA cycle. It is critical to be very clear about the business problem and the data required to understand it better

3. Pilot before Scale Up

Now that the data problem is defined, managers can understand where they may have organisational and infrastructure gaps for their project, and from this be able to identify the first steps of their roadmap to a data solution. It is important that these early steps include a pilot of the solution. The goal is to quickly understand if our solution is likely to be successful, and the actions to be taken to scale up across the organization.

Here the PDCA steps of “do – check – act” come into play as there is rapid prototyping or piloting of a new solution, checking the results against what was expected and using this information to learn more about the process. The final solution can then be adjusted prior to implementation and/or replicated across multiple solutions.

This case study highlights how customer value is enhanced, and rapid improvement can be achieved by harnessing the power of technology through data and applying Lean principles and practices to difficult supply chain problems.

Some further examples of how Industry 4.0 and Data is enhancing lean principles include:

  • Value from the customer’s perspective: the use of big data and analytics help us understand how customers perceive the value of our products. Most importantly we can use this to offer our customers enhanced experiences and offerings.
  • Supply Chain Planning: we are able to determine demand patterns far more accurately, which reduces overall supply chain inventory. Logistics can be optimised to reduce costs, reduce fuel and contribute to reducing environmental impact.
  • Quality improvement through reducing variation: real time quality control using sensors, digital visualisation and control systems.
  • Visualisation of process conditions: we can now view the current state of processes in real time through apps on our phones in a way that provides targeted and pertinent information for decision making.

Often in business we take it for granted that we have all the capabilities in place to be able to run our business. However in today’s world, where the use of technology is rapidly evolving, it is very easy to become ‘out of data’ both from a business mindset as well as technology capability. To help leaders understand the strength’s and weaknesses, The Si2 has worked with the Data Analysis Bureau to develop a short 10 minute maturity survey which will you help you identify your strengths and weaknesses as you move from Business Problem to Data Solution. There are just 10 questions and you will get personalised feedback at to your situation and what you can do. Use this link to access the assessment now at https://tdab.outgrow.us/data-readiness-tool

Nick Frank  is Managing Partner at Si2PARTNERS and can be contacted at nick.frank@si2partners.com

Join our community  to keep up to date with the latest thinking in Service Business Growth