Simulation Decomposition for Agriculture and Engineering: Seeing Variability, Cause, and Effect

What happens inside complex models when inputs vary?
Monte Carlo simulations can show the range of outcomes, but not the reasons behind them. Simulation Decomposition (SimDec) fills that gap — by revealing how different input combinations shape the output in an intuitive, visual way.

In “Analysis of Agricultural and Engineering Systems using Simulation Decomposition” (Liu, Leifsson, Pietrenko-Dąbrowska & Kozieł, 2022), researchers applied SimDec to three very different problems — each showing how visual analytics can make uncertainty understandable instead of abstract.

🔗 Read the paper: https://doi.org/10.1007/978-3-031-08757-8_36

Three Systems, One Technique

1️⃣ A simple model problem
The team first tested SimDec on a mathematical function with three inputs and one output.
By dividing each input into “high” and “low” states, they created eight possible combinations.
The resulting stacked distribution (shown in Figure 3 of the paper) instantly revealed which variables dominated the output — a feature invisible in ordinary Monte Carlo plots.

2️⃣ Ultrasonic nondestructive testing (NDT)
Next, SimDec analyzed an ultrasonic inspection setup for a quartz block with a void.
Input parameters such as probe angle, position, and depth were varied randomly.
The decomposed histogram (see Figure 6, page 7) showed a clear diagonal trend:
large probe angles led to lower reflected signals, while smaller ones produced stronger echoes — confirming that the system’s behavior was driven mainly by probe angle.

3️⃣ The Iowa food-energy-water system
Finally, the method was used on a simulation of nitrogen surplus in Iowa agriculture.
By varying July temperature and precipitation, SimDec visualized how different weather scenarios influence nitrogen export.
The decomposed output (Figure 9) showed that regular temperature and rainfall contributed most to nitrogen accumulation, while extreme conditions produced rare but large deviations — valuable insight for environmental management.

Why It Matters

Across engineering and environmental science, models are getting more complex — and less transparent.
SimDec helps make them interpretable by:

• Visually connecting input variability to output distributions,
• Revealing dominant factors and their interactions,
• Reducing computational cost (only one simulation run is needed).

The authors note that combining SimDec with global sensitivity analysis (GSA) or surrogate modeling could further strengthen uncertainty evaluation in large-scale simulations.

In Short

SimDec turns data from Monte Carlo simulations into readable stories — of inputs, outputs, and the invisible chains between them.
Whether it’s sound waves in solids or nitrogen in soil, the approach helps scientists see complexity clearly.

Reference:
Liu, Y.-C., Leifsson, L., Pietrenko-Dąbrowska, A., & Kozieł, S. (2022). Analysis of Agricultural and Engineering Systems using Simulation Decomposition. In Computational Science – ICCS 2022 (pp. 1–10). Springer.
https://doi.org/10.1007/978-3-031-08757-8_36

‍