Automation is relevant in repetitive processes. In those situations, robots, machine learning and data can secure consistency and accuracy. Business owners and managers often do a direct comparison of the CAPEX investment vs. man-hours, to do the same task. One critical detail that is usually not considered, is the ‘human factor’ and its impact on quality.


Let me give you a real-life example of how the ‘human factor’ in repetitive processes can have consequences: As a frequent traveller, I try to optimize my travel and save time, I therefore tend to use the same car-rental company, the same hotel chains etc..


Preparing for a recent customer visit (pre-COVID-19), I rented a car through the usual rental company and stayed at my usual hotel. When arriving at the hotel, I handed the car keys to the valet, who gave me a receipt which was printed using a handheld registration device. The receipt included details such as license plate no., the time and date of arrival, and even the colour of the car.


The whole staff greeted and welcomed me back, which of course is a custom and trained behaviour towards regulars. The next morning, on my way out, I gave the receipt to the valet, who went to his drawer and picked up a car key and started reading the label, looking very confused. Some minutes later, his manager approached me and asked if I were traveling with Mr. Sam. When he found out I didn’t know Mr. Sam, he apologized many times and informed me that my car had been mistakenly given to Mr. Sam, who is also a regular, and who also rents cars from the same rental company.


In this case, the people involved all made mistakes based on assumptions and previous experiences. The registration device resets the counter every month. Mr. Sam had an old receipt in his pocket with the exact same no. and had mistakenly given the old receipt to the valet. To make things worse, he had also rented the same type of car from the same rental company. However, his car was grey and mine was white. Besides that, the dates on the receipts and the license plate number’s were different.


How could such a mistake happen? One word, habit!

The valet knew and trusted Mr. Sam. They didn’t double check or pay enough attention to his receipt. Mr. Sam didn’t question the valet, because he believed it was a simple (yet repetitive) task. Furthermore, he trusted the valet as he knew him from the many times he has stayed at the hotel. Not even the obvious colour difference triggered a reaction.


Would this issue even exist if the process was automatically controlled by a machine/software? Clearly not! An automated process would have done all the checks in an instant, comparing the dates, the license plate no. and the colour - no matter who the guest was or how many times it had repeated the same task.


As human beings, we are biased by so many external and internal factors whether it's mood, sickness, stress, poor communication and exhaustion, just to mention a few. We are not well- built for such repetitive tasks.


So, what does this all mean for a cement plant?

The cement production process is complex. It requires a large amount of data flowing from the quarry all the way through the process to packing and dispatch.


Operators gather most of their information routinely through different communication protocols, directly from the equipment. However, at many cement plants, many data points are still entered manually by quality- and process staff. 


An operator controls the process using the data provided to him and changes the setpoints and adjusts accordingly.


Starting from the actual sampling all the way to the preparation and analysis, human biases are influencing the process. Given a cement plant contains several sampling points, and requires several types of analysis – hourly, mistakes along the way are certainly possible. In turn, the situation can lead to false conclusions and costly errors. 


Below is an illustration of the material mass flow in a typical cement plant. The analysed mass is less than 0.1g, but represents 200t of material. Having an accurate, automated sampling and sample preparation process is crucial to eliminate false results and to reach the targeted quality levels.

Every tonne of cement produced below standard quality is money out of the window. On the other side, every tonne of cement produced above standards is also costly. It's all about having as tight and precise a quality control as possible.

An automated sampling and sampling preparation process eliminates the risks mentioned above and releases humans from doing daily repetitive tasks.

At FLSmidth, we see these cases daily and assist our customers in optimizing these processes. Our automated quality control solution; QCX® system, takes samples automatically at a pre-set rate and location. The sample is then transported through tubes to the laboratory, where robots prepare and run the different analyses. The results are stored in a central database and accessed by the plant control system; readily available so that operators can use the data to take the best actions.

We have many intelligent solutions available to us today, in diverse areas of life. All for the purpose of avoiding unintended errors or simply eliminating repetitive work. It helps us bypass fundamental human biases and optimize processes. I will probably double check my receipt next time I hand-in my car keys at the hotel, but if I ran a 4000 t/day cement plant, I know who (what) I would trust.  

Stories from FLSmidth