How can we feed AI?
By Dr. Selim Erhan, TLT Editor | TLT From the Editor December 2023
Artificial intelligence (AI) can reduce the time needed to formulate complex products.
Artificial intelligence (AI) has entered many aspects of our lives. The complicated algorithms that require enormous computing power are now in reach for many applications. The complexity of industrial problems is one of the ideal areas of applying this powerful tool. Industrial applications are very complicated because the mechanisms we are dealing with are affected by speed, friction, temperature, chemical interactions, electrical interactions, time, weather conditions—all the way to human behavior. Most of the time we do not think of this complexity because we are accustomed to working in a simpler scope. We must look at what we can manage and hope the rest will fall in place.
For example, let us look at a beverage can. It is not made only out of aluminum. The aluminum part is complicated enough. In metalworking, we already know the complexity of drawing a can, the lubricants that must adhere to a flowing metal and operating under starved lubricant conditions. There also is the speed and afterward the cleaning of the surfaces and more. But the amazing part is that it utilizes three different materials that must work together during its performance. After the can is formed, it is coated on the inside with a special plastic so the acidity of the beverage will not dissolve the aluminum. The outside is painted. When the can is completed, we are looking at three completely different materials—the plastic, the aluminum and the paint—where all must adhere and stay together through a variety of conditions, including mechanical abuse and hot and cold temperatures which at times change very quickly. All three materials have different expansion and contraction coefficients. Each can go through these cycles many times. I am sure a whole lot of people deserve congratulations on a collaboration, which we seldom give any thought to. I am sure there were many trials and errors and experimental designs that slowly reached a successful outcome. AI power reduces this time significantly.
When we look at formulating, we are really looking at a team effort. It is somewhat like assembling a car where different manufacturers design and prepare the different parts, and all comes together at the final assembly line. The formulators, for example, who are designing metalworking fluids take advantage of premanufactured additive packages. In some applications this approach speeds up the formulating process. The difficulty in this approach is that the formulator does not know all the details in the additive packages and the additive manufacturers do not know the end formulation. So, in fact it is a much more complicated issue when we compare it to the example of manufacturing a car. This is where AI helps to reduce the time it takes to formulate these complex products.
However, for AI to work well, it needs as much data as possible. If additive manufacturers can provide as much physical data as possible, they will have much happier customers. For example, AI has been used in rubber formulations for years. Rubber and tires are another one of the very complicated areas where there are many and sometimes conflicting requirements. A tire must adhere to the road so there is some friction to prevent the vehicle from sliding and, when the time comes, to stop at a reasonable distance. But at the same time it must be flexible enough to bounce off the road, because if it sticks too much it will take a lot of energy to keep the tire rolling, which will increase the fuel consumption and engine wear. Then there are wet conditions, dry conditions, very hot roads, very cold roads, ice and snow and rain, which are all different from each other in their requirements. We add speed and acceleration, momentum and lateral forces. There is a varying amount of weight, and of course noise must be minimized, wear must be minimized—and the list goes on. The layers of dissimilar materials in a tire are astonishing. These are only the physical requirements. Then we have the dozen or so different additives which must work with each other at the molecular level. Therefore, tire formulators are happily embracing AI. They feed the physical and chemical data from the additives into complex algorithms, which then navigate the input toward the desired outcomes. This way the formulator can reduce the number of experiments from over 100 to a few a day. Therefore, the more they know about each of the raw materials the better the results from the algorithms.
Let’s take another example, drilling operations. Although not even close to the complexity of tires or metalworking fluid operations, the whole process is very expensive, often far exceeding $100,000 a day. Optimum drilling, maximizing tool life and achieving optimum speed is very valuable. Holes can go to 30,000-40,000 feet or more. They can start vertically and then go horizontal and sometimes go around obstacles, all controlled from ground level with data from sensors and GPS. AI helps and is used. With these powerful tools, we humans must now deal with too much information and not get distracted! Luckily it is the end of the year: holidays, family and friends and relaxation time. We can think of too much information later and hopefully deal with too much food now! Happy New Year!
Dr. Selim Erhan is director of business development for Process Oils Inc. in Trout Valley, Ill. You can reach him at serhan@processoilsinc.com.