Entropy in the Manufacturing World

Complexity lies between low entropy and high entropy

Published: Monday, April 10, 2017 - 12:02

In this article, I will be looking at entropy in the manufacturing world. Entropy is generally defined as a measure of disorder. This general definition can sometimes be inadequate.

Let’s look at the example of a desk in an office. One could say that if the desk appears to be in order (i.e., neat and tidy), then it has low entropy. However, the concept of orderliness is very subjective. To me, if I am able to know where everything is, and I can access each item quickly, then my desk has low entropy. It may not seem “ordered” to an outsider, and he may conclude that my desk has high entropy.

The second law of thermodynamics, loosely stated, is “entropy always increases.” Thus, a desk will always get messier. There is also a probability aspect. There are many different ways the things on my desk can be arranged, but only a very small number of those arrangements can be considered “ordered.” Thus there are a multitude of ways a desk can be seen as more disorderly than the small number of ways it can be seen as orderly. From a probability standpoint, it is always likely that a desk is messy unless there is a consistent process in place to put it back to the “ordered” state at frequent intervals. This line of thinking also shows that the more things you have on your desk, the more likely your desk will be in a state of “messiness.” Interestingly, 5S in lean requires you to limit the number of items in an area to only those items that are needed. You are encouraged to remove all of the extra items.

Entropy can also be explained in terms of the element of surprise. For example, in a brand-new deck of playing cards, the cards are arranged in order. It has low entropy because one knows exactly where every card is. There is minimal element of surprise. If one were to riffle-shuffle the cards once, there is still some form or order maintained in the cards. For example, the order of the cards from ace to king is not disturbed. There may be some different cards in between, but the three of hearts is still above the four of hearts, even though there may be other suit cards in between them. This concept is known to magicians and used in several magic tricks. When the cards are shuffled again and again, however, the knowledge of any form of order is lost, and the entropy thus increases. With a well-shuffled deck of cards, any card presents an element of surprise—new information.

With the same logic used in the previous paragraph, it is very unlikely that continuous shuffling will bring a deck back to the original new deck order. There are always more ways for the deck to be in a different order than a new deck order. In the new deck order, if you are required to produce the king of hearts, it is simple to do it because you know the order of the cards. You can do this fairly quickly. However, when the deck is shuffled, this becomes harder. You will need more time to look through every single card until you get to the king of hearts. Although it is not exactly the same, it is stated that as entropy increases, it causes a decrease of useable energy. Thus, one could say that high entropy states do not yield value. Jeremy Campbell, in his wonderful book Grammatical Man (Ward & Balkin Agency, 2012) states, “At the heart of the second law [of thermodynamics] is the insight that order has value.”

One can understand the need to maintain order in the manufacturing plant. Managers strive to maintain low entropy within the manufacturing system, and they surely do not appreciate surprises. From their viewpoint, painting all cars black makes sense. Producing the same item in big numbers using the principles of mass manufacturing is an attractive proposition for management. A large number of products and components brings disorder and an increase in entropy. Thus, minimizing the variety of products manufactured is an attractive proposition for management.

However, the world has become smaller, and the market is asking for variety. From a complexity science standpoint, one can say that manufacturing processes are ordered or complicated. There are good cause-and-effect relationships, and these can be easily controlled. However, the complexity outside the manufacturing plant has increased with the advent of the information age. A manufacturer in China can sell his goods in America—and vice-versa—easier. The demand for variety from the market is increasing, and a manufacturer can no longer make only black cars if it wants to stay in business.

Management has to realize that organizations are not technical systems, but sociotechnical systems. When you treat an organization as a technical system, you assume that direct, linear cause-and-effect relationships exist, and that you can control the system through hierarchy. The most important requirement in this case is to minimize entropy. Entropy has a negative relationship with efficiency in mechanical (technical) systems.

On the other hand, the goal of a sociotechnical system is not primarily to lower the entropy at all times. Complexity lies between low entropy and high entropy. Complex problems require complex, dynamic solutions. Organizations should become complex, adaptive systems and be able to move between phases in order to thrive. “Everything changes” is the reality, and thus the organization should be able to change and adapt its actions to meet the needs posed by the environment.

The idea of order implies a state of permanence. The organization has to go through phases of permanence and impermanence to be able to thrive. This is similar to the idea of kaizen in the Toyota Production System, where kaizen requires standards. Kaizen, the idea of change to improve, requires order (standards). This is also the going back and forth between permanence and impermanence. In the complex world today, nothing should be set in stone.

I will finish with a wonderful lesson from Shunryū Suzuki-roshi:

“Suzuki Roshi, I’ve been listening to your lectures for years,” a student said during the question-and-answer session following a lecture, “but I just don’t understand. Could you just please put it in a nutshell? Can you reduce Buddhism to one phrase?”

Everyone laughed. Suzuki laughed.

“Everything changes,” he said. Then he asked for another question.

Always keep on learning.

About The Author

Harish Jose

Harish Jose has more than seven years experience in the medical device field. He is a graduate of the University of Missouri-Rolla, where he obtained a master’s degree in manufacturing engineering and published two articles. Harish is an ASQ member with multiple ASQ certifications, including Quality Engineer, Six Sigma Black Belt, and Reliability Engineer. He is a subject-matter expert in lean, data science, database programming, and industrial experiments, and publishes frequently on his blog Harish’s Notebook.