[Jeong Sun-woo Column] Plant Factories and Automobile Factories

[Jeong Sun-woo Column] Professor, Department of Horticulture and Life Science Engineering, College of Agriculture and Life Sciences, Chonnam National University

The word factory carries the expectation of mass-producing products of a consistent standard. Just as cars of the same model roll out one after another from an automobile factory, people expect a plant factory to produce crops of uniform quality at a set time. Because the influence of weather can be reduced and light and temperature, water and nutrients can be controlled, the expectation is that if cultivation conditions are properly set, production results will also become consistent.
 

Environmental control and growth control are different

Yet the objects handled by the two factories are different. Cars are assembled from parts according to a design, but plants respond to their environment and build their own bodies. Light and temperature affect not only growth rate but also leaf size and orientation, as well as stem length. These altered forms in turn change the conditions under which plants receive light. A plant is both an object passing through a production process, and a living organism that continues to change during that process.

Think of lettuce. Even if the brightness of the lighting is the same, the distribution of light received by young lettuce and lettuce nearing harvest differs. As leaves grow and overlap one another, upper leaves shade lower leaves, and even the leaves of neighboring plants are affected. Even plants sown at the same time receive different amounts of light if their leaf size and arrangement differ. The facility’s lighting settings remain the same, but the environment experienced by the plants changes. Controlling the environment consistently and making plant growth consistent are not the same task.

Therefore, stable production requires not only the ability to set the environment to desired values, but also the ability to predict how plants will grow under those conditions. It is not enough to check only the current state. We need to know what consequences today’s decision to reduce lighting or change planting density will have for growth and harvest several days later. The question shifts from ‘Can we control the environment’to ‘What kind of environment should we create’.

The surest way to verify this is to actually grow the crop. But the conditions to be tested quickly multiply. If light intensity, temperature, carbon dioxide concentration, and planting density are each divided into three levels and every combination is compared, there are 81 cases. If variety, growth stage, and repeated experiments are also considered, the number of possibilities grows even larger. Space must be prepared for each condition and one must wait until harvest, and in a facility that is already producing, losses from failure must also be borne. It is difficult to test every choice first with actual crops.
 

What can be learned from an automobile factory’s virtual factory

Automobile factories also do not verify every problem only on the actual production line. BMW explained in a case it unveiled in 2025 that it uses a virtual factory to first review whether a new vehicle body collides with surrounding equipment as it moves through the production line. This means finding problems on a computer before changing the actual equipment. What plant factories can learn from automobile factories is not only automated production equipment. They can also learn methods for testing the consequences of choices before actual production.

In agriculture, the foundation for such virtual testing is the crop model. Crop models based on physiological processes express photosynthesis and respiration according to the environment, as well as growth and the distribution of matter, in mathematical formulas. They calculate how crops will respond when different light or temperature conditions are entered. Candidate conditions that are difficult to test entirely in actual facilities can be reviewed first inside a computer, and important conditions among them can be selected for experimentation.
 

Crop models·FSPM, calculating growth before experiments

In particular, when deciding the position of lighting and the spacing between crops, it is advantageous to consider even the spatial appearance of the plants. This is because even if the total leaf area is the same, the distribution of light differs depending on the height and angle at which leaves are positioned, and the degree to which they overlap one another. A tool that can be used in this case is the Functional-Structural Plant Model, known as FSPM(Functional-Structural Plant Model). It is a model that links the three-dimensional structure of leaves and stems with physiological processes such as photosynthesis.

The key is not to draw a plant so that it looks plausible. It lies in calculating how much light reaches each leaf, and how much of that light is used for photosynthesis. In models that implement growth as well, the distribution of light that changes as new leaves appear and existing leaves enlarge is recalculated. They handle the relationship in which the environment changes physiology and form, and the changed form in turn affects physiological activity.

If these calculations are expanded into a simulator, questions such as when to irrigate during cultivation, or at what stage to adjust lighting can be examined. If calculations of equipment power consumption and operating costs are connected to this, the criteria for judgment also change. The goal becomes finding conditions that reduce costs while meeting the target size and harvest timing, rather than the conditions that make plants grow the fastest. This is the direction of linking growth prediction to production planning.

Having a simulator does not mean eliminating actual cultivation experiments. It means deciding which experiments should be conducted first, and making the knowledge obtained from those experiments usable again in the next judgment. Research that measures changes in leaves and roots, physiological responses, and final yield is as important as recording the environment. A relationship is needed in which actual cultivation verifies the model, and the model guides the next cultivation test.

Plant factories and automobile factories are similar in that they aim for uniform and stable production. However, having the same goal does not mean it can be reached in the same way. For plant factories to achieve productivity worthy of the name factory, they must understand plant responses as living organisms more thoroughly, rather than treating plants like parts. When facilities that control the environment are combined with tools that predict changes in plants, the automation of cultivation can lead to production stability.