Experimental methods in chemical engineering–Validation of steady-state simulation

Abstract

Steady-state simulation (Aspen, PRO/II, WinGEMS, CADSIM Plus) guides equipment selection, operating conditions, and optimization to design chemical processes like Kraft pulping, specialty chemicals, and petrochemical complexes. Ensuring that the simulation characterizes the yields, heat transfer loads, purity, utilities demand, and profitability requires data that represents the physicochemical and transport properties of each stream and unit operation. Here, we present strategies to validate steady-state simulations against plant data and expectations from operators. To build and validate simulations requires real-time data, but errors contaminate measurements and dynamic conditions—start-up, shut-downs, process upsets—compromise fidelity. A pre-treatment step removes incongruous data to build the simulation on process conditions representative of steady-state. Working through the process with experts (informal validation) and comparing simulation results with plant data (formal validation) reduces gross error with an objective to achieve a simulation accuracy to within one standard deviation of measurement variability. A bibliometric review highlights the limited focus on steady-state simulation validation in the field of process engineering. Most articles mention accuracy but neglect to describe how it is evaluated. Despite this scarcity, validation remains a critical factor in various domains of chemical engineering research. Interviews with professionals offer a practical perspective on the applications of simulation in an industrial context like process monitoring, equipment performance analysis, operator training, and decision-making. Finally, a case study demonstrates how to implement data treatment and validation for Kraft mill brownstock washing department: Applying multiple validation techniques increases the value and confidence in the simulation.