Identify dominant parameters of comprehensive pyrolysis model - Part I: Variance-based sensitivity analysis of wooden materials

When a piece of wood burns, it does not simply vanish into ash and smoke. Before the flames even touch the surface, the material undergoes a silent, invisible transformation called pyrolysis. In this process, intense heat breaks the complex chemical bonds inside the wood, turning solid matter into gases that can ignite and a solid residue known as char. Predicting exactly how fast this happens, how much gas is released, and how hot the center of the wood gets is a critical challenge for engineers designing fire-safe buildings and for scientists trying to understand how wildfires spread. To do this, they build computer models that simulate the physics of the event. However, these models are notoriously complicated, relying on dozens of different numbers to describe everything from how quickly a chemical reaction occurs to how well heat moves through the material. The central question for researchers has long been: which of these dozens of numbers actually matters most? If a model is a machine with many gears, knowing which gears drive the motion allows engineers to build better, faster, and more reliable simulations without getting lost in unnecessary details.

A team of researchers at Wuhan University of Technology set out to answer this question by applying a rigorous statistical method to the study of wood burning. They focused on two common materials: birch wood and medium-density fibreboard, a type of engineered wood product. Instead of guessing which factors were important, they used a technique called variance-based sensitivity analysis. Imagine a model as a black box that takes in many different settings and produces an output, such as the rate at which the wood loses weight or the temperature at its core. The researchers systematically varied every single input parameter within a realistic range and watched how much the output changed. By measuring how much the uncertainty in the output could be traced back to each specific input, they could identify the dominant drivers of the process. Their goal was to separate the essential physics from the noise, determining whether the precise details of chemical pathways or the broad strokes of heat transfer were the true masters of the pyrolysis process.

The investigation revealed a striking hierarchy among the chemical properties of the wood. The researchers found that the activation energy, a measure of the temperature threshold required to start a chemical reaction, was overwhelmingly the most influential factor. In their simulations, the sensitivity of this temperature threshold was generally more than ten times greater than that of the reaction order, a parameter that describes the specific shape or path of the chemical reaction. This finding suggests that for the purpose of predicting how wood burns, the exact temperature range in which the material breaks down is far more critical than the minute details of the reaction path itself. While the reaction order might slightly alter the shape of the curve showing how fast mass is lost, it does not significantly change the overall outcome. This implies that modelers do not need to spend excessive time chasing perfect values for reaction orders; focusing on the correct temperature range yields a much better return on investment for accuracy.

The study also examined the role of moisture, a variable that is often difficult to control in real-world scenarios. Wood naturally contains water, which must evaporate before the wood can begin to char. The researchers tested models with moisture contents ranging from a dry 10 percent to a very wet 50 percent. They discovered that while the chemical kinetics of water evaporation had a noticeable effect on the temperature at the center of the wood, it had very little impact on the total amount of mass lost. Even when the wood was soaked with 50 percent moisture, the rate at which the material burned away was not significantly altered by the precise speed of the water's evaporation. Instead, the heat required to turn that water into steam became the dominant factor. This distinction is vital: if an engineer is trying to predict how long a wooden beam will last in a fire based on how much it weighs, the details of water evaporation are less important. However, if the goal is to predict how hot the inside of the beam gets, the behavior of the water becomes the primary concern.

Perhaps the most significant insight came from comparing the chemical reactions against the physical movement of heat. The researchers found that these two forces compete, but their dominance depends entirely on what is being measured. When looking at mass loss—the amount of wood turning into gas—the chemical reactions were the clear drivers. The speed at which the wood decomposed chemically dictated how quickly it disappeared. Conversely, when looking at the temperature at the center of the material, physical heat transfer took the lead. How well the wood conducted heat and how much energy it took to raise its temperature were the deciding factors for internal heat. This balance shifted dramatically with moisture content. In dry wood, the competition was between chemical speed and heat conduction. But in wet wood, the energy required to evaporate the water became so large that it overshadowed almost all other thermal properties, effectively dampening the influence of the wood's natural ability to conduct heat.

The researchers also looked at how different components of wood, such as cellulose, hemicellulose, and lignin, contributed to the process. They found that the importance of lignin, the component that gives wood its rigidity and turns into char, increased as the heat flux, or the intensity of the fire, increased. At lower heat levels, the wood simply did not get hot enough to fully break down the lignin, making its chemical properties less relevant to the simulation. However, under intense heat, lignin decomposition became a major factor. This nuance highlights that a single model cannot be universally applied without considering the intensity of the fire. The study confirmed that while chemical reactions drive the consumption of the material, physical heat transfer governs the thermal profile within it, and both must be included in any comprehensive model.

Ultimately, this work provides a clear roadmap for improving the computer simulations used to predict fire behavior. By identifying that activation energy is the king of chemical parameters and that reaction order is a minor player, the researchers have shown where modelers should focus their efforts. They demonstrated that simplifying the complex details of reaction paths does not come at the cost of accuracy, provided the temperature ranges are correct. Furthermore, they clarified that the role of moisture is specific to temperature prediction rather than mass loss, and that the interplay between chemistry and physics changes depending on how wet the wood is and how hot the fire burns. These findings do not just refine a mathematical equation; they offer a deeper, more practical understanding of the physics of burning wood, allowing for more efficient and reliable tools to protect lives and property from fire.