Thaioil · Technologist Intern · Jun – Nov 2025

Finding fouling in a crude preheat train

Before crude reaches the furnace, a train of heat exchangers warms it using hot product streams. As deposits build up on the tubes, less heat gets across and the furnace has to burn more fuel. I built a simulation that tells you how fouled each exchanger is, using only the temperatures the plant already measures.

Aspen EDRAspen HYSYSPlant dataEnergy analysis

What I did

Built rigorous models of the preheat exchangers in Aspen EDR, using each exchanger's real geometry.
Fed the models the current operating inlet and outlet temperatures, then found the fouling resistance that makes the model match what the plant is actually doing.
Used the results to estimate the energy lost to fouling across the train and to support planning of exchanger cleaning, so the worst exchangers get cleaned first.

This work was the basis of my B.Eng. senior project. Plant data and results belong to Thaioil and are not shown here.

Try the idea

This is a simplified version of the same method for one exchanger. Enter four measured temperatures and it works backwards to the fouling resistance. The real work used rigorous EDR models; this page uses a plain energy balance and LMTD, with made-up example numbers.

Measured today

Hot in, °C

Hot out, °C

Crude in, °C

Crude out, °C

Crude flow, kg/s

Exchanger (example)

Area, m²

U clean, W/m²·K

Furnace fuel price

Fouling resistance
U, fouled vs clean
Heat duty now
LMTD
Heat lost to fouling
Extra furnace fuel

How this is calculated

Q = ṁ_crude · c_p · (T_crude,out − T_crude,in), c_p = 2.5 kJ/kg·K
LMTD (counter-current) = (ΔT₁ − ΔT₂) / ln(ΔT₁/ΔT₂), ΔT₁ = T_hot,in − T_crude,out, ΔT₂ = T_hot,out − T_crude,in
U = Q / (A · LMTD), R_f = 1/U − 1/U_clean
Clean duty: ε-NTU with U_clean at today's inlet temperatures and flows
Extra fuel = (Q_clean − Q) / 0.85 furnace efficiency × 8,400 h/yr × fuel price

The trend shows an invented year of readings, with a cleaning in July, so today's value has something to be compared with. The cleaning trigger at 0.0012 m²·K/W is an example, not a plant limit.