CPE 613 Lecture 32: Dynamic Simulation and Digital Twins

Characterization of Dynamic Systems

Foundational Principle of Dynamics: All chemical processes are essentially dynamic because any disturbance in an input propagates through the system to the outputs.
Steady-State Processes: These are designed to reject disturbances and are intended to operate near one or more stable operating points.
Inherently Dynamic Processes: Certain processes are intended to be dynamic by design, including: * Batch Processing: Operations where materials are processed in discrete lots. * Semi-batch Processing: Operations that involve continuous feed or removal of one or more components while others remain in the vessel. * Storage: Systems designed to hold material, making them susceptible to volume and composition changes over time. * Processes with Cycles: Examples include Pressure Swing Adsorption ( $PSA$ ), which relies on cycling pressures. * Transitional Operations: The start-up and shutdown phases of otherwise steady-state processes are inherently dynamic operations.

Selecting Simulation Software

Design Focus: Most simulation programs featuring a dynamic mode are primarily designed for controllability studies, with some extensions for batch processing.
Steady-State Simulators for Dynamics: To use a steady-state simulator for truly dynamic processes, the user must simulate iteratively and approximate an Ordinary Differential Equation ( $ODE$ ) solution.
Software Specifics: * ASPEN Dynamics: This tool is essentially built for control studies. * HYSYS: Now part of the ASPEN suite, HYSYS includes a dynamic mode capable of solving $ODEs$ numerically.

Digital Twins and Operator Training

Definition: A digital twin is a simulation system utilized to model the effects of various input changes on a chemical system.
Interface: The software outputs results on the same screens (or exact replicas) used in the actual plant control room.
Primary Application: The main use for digital twins is Operator Training.
The Technical Challenge: The simulation must solve dynamic expressions at a rate consistent with the actual propagation speed of a disturbance through the plant to maintain realism.

Modern Applications and Renewable Energy

Intermittency Management: Dynamic simulation is increasingly used to understand plant operations under variable energy availability, a phenomenon known as intermittency.
Renewable Source Integration: Intermittency is a byproduct of using renewable resources such as wind and solar power.
Design Mismatch: Chemical plants are not traditionally designed to fluctuate their production rates based on weather conditions.
Predictive Optimization: "True" dynamic simulation allows engineers to determine the optimal set-point profile based on near-term energy supply predictions.
Case Study (Professor's Lab): A project involves the production of ammonia ( $NH_3$ )—essential for fertilizer to feed the global population—using $100\%$ renewable resources.

Requirements for Dynamic Simulation

Performing a dynamic simulation requires more extensive information than a steady-state simulation, specifically for balance equations. Required data includes:

Physical Dimensions: Unit operation volumes and specific liquid holdups.
Pressure Management: For processes involving vapors, pressure controllers and valves must be included in the model.
Control Infrastructure: * Identification and sizing of control valves to determine minimum and maximum flowrates. * Valve failure modes: Must specify if valves are fail-open or fail-closed.

Case Study: Monomer Purification and Storage

System Description: A holding tank for diolefins is situated before a separation scheme.
Safety Hazards: Diolefins tend to dimerize exothermically, making their storage potentially hazardous. The reactions involve the transformation of monomers into dimers.
Inherent Dynamics: This process is dynamic because: * Feed and discharge rates vary based on upstream conditions and tank levels (head pressure). * Batch processing rates downstream of the tank may fluctuate. * The system is subject to traditional disturbances even during intended steady-state operation. * Operator errors can cause significant deviations.

Dynamic Response Analysis Questions

When evaluating a system like the monomer purification tank, dynamic simulation helps answer specific questions:

What are the temperature, pressure, and concentration profiles during the start-up phase?
What are the impacts of small fluctuations in feed temperature or feed concentration?
What occurs if the entire feed line is blocked?
What happens if the tank is flooded with feed, such as during an emergency release from a previous unit?

Quantitative Case Study Data

Steady-State Baseline

The following table represents the simulation results at steady-state:

Component	Inlet	Outlet
Diluent	$80000$	$80000$
MA (Monomer A)	$20000$	$19810$
MB (Monomer B)	$15000$	$10694$
DA (Dimer A)	$0$	$15$
DB (Dimer B)	$1000$	$5136$
CD (Co-Dimer)	$0$	$346$
Temperature ( $^{\circ}F$ )	$215$	$226$
Pressure ( $psia$ )	$100$	$100$

Start-up Scenario

Initial State: The tank begins filled with diluent.
Action: The tank is filled with diluted monomer while the outlet remains open.
Observation Profile (0 to 6 hours): * Temperature: Rises from the initial point to approximately $226\,^{\circ}F$ . * Pressure: Increases over the $6\,hr$ period toward reaching a steady state.

Feed Flowrate Reduction

Scenario: The inlet flowrate is decreased by $50\%$ after the system reaches steady state.
Result: Temperature and pressure both increase.
Explanation: The diluent outflow rate becomes greater than its inflow, and reactants are simultaneously created by the back reaction, contributing to heat and pressure buildup.

The "Big Problem": Blocked Feed Excursion

Scenario: An operator panics and closes the inlet flowrate completely (Blocked in Feed).
Consequences: This causes a severe "excursion" or runaway scenario.
Quantitative Excursion Data (0 to 6 hours): * Temperature: Rapidly increases from $226\,^{\circ}F$ to over $300\,^{\circ}F$ . * Pressure: Spikes from $100\,psia$ to approximately $180\,psia$ to $200\,psia$ .

Conclusions and Summary

Simulation Capability: Either steady-state or dynamic simulators can perform dynamic studies, but dedicated dynamic simulators provide superior options for solving $ODEs$ .
Universal Dynamics: While all processes are dynamic, inherently dynamic systems like product storage are simulated with different methodologies than steady-state processes facing disturbances.
Human Factor: Dynamic effects are often counter-intuitive. An untrained operator's initial response may be the exact opposite of what is required to stabilize the plant. Simulation and training are vital for accident prevention.

Question of the Day

Question: Whether you solve a dynamic simulation using a simulator like ChemCAD, a dynamic one like HYSYS, or your own $ODE$ solver, there is one parameter that you set which defines the accuracy of the simulation. What is it, and what factors should you consider when selecting it?
Discussion Point: The parameter in question is typically the Time Step (or Step Size). Considerations for selection include the speed of the process dynamics (time constants), the numerical stability of the solver, the desired precision of the results, and the available computational power.