4 Process Simulation

Process Simulation

© 2012 G.P. Towler / UOP. For educational use in conjunction with Towler & Sinnott Chemical Engineering Design only. Do not copy

Chemical Engineering Design

Process Simulation • Once we have established the block flow and started filling in the main vessels, heat exchangers, pumps, etc. we want to develop a mass and energy balance for the process so we can start evaluating the process in more detail • The simulation is also the starting point for equipment design, as it will set the flow rates and duties for process equipment • In most companies, mass and energy balances are developed using a process simulator such as AspenPlus, ChemCad, ProII or UniSim. Each program has its own idiosyncracies, but they have many common features. Examples will be given in both AspenPlus and UniSim. • Note that the simulation should come after you know what’s in the PFD, but often it’s an iterative process to get both



Process Simulation • Structure of process simulators • Components and physical property models • Modeling reactors • Modeling separations • User models • Recycles & convergence • Optimization © 2012 G.P. Towler / UOP. For educational use in conjunction with Towler & Sinnott Chemical Engineering Design only. Do not copy


Structure of Process Simulators Equipment sub-routines

Convergence & optimization sub-routines Physical property data

Executive Program

Thermodynamics sub-routines

Cost data


Graphical User Interface (GUI)

• The user manipulates the program through a GUI that is set up to look similar to a PFD • The executive program determines the calculation sequence and calls the other subroutines Chemical Engineering Design

Example: UniSim Simulation of GE LM6000 Engine

• Features that will be described are common to most other simulators © 2012 G.P. Towler / UOP. For educational use in conjunction with Towler & Sinnott Chemical Engineering Design only. Do not copy


Using the GUI: Basis Environment

Click here



Basis Environment

Enter components

Enter reactions

Select property package

Specify stoichiometry



Basis Environment

Enter reactions

Specify conversion



Using the GUI: Object Palette

Click here



Object Palette

Unit operations General reactors Separator models

Spreadsheet

User can select operations from the palette and drag and drop to the PFD

Adjust, Set, Recycle

Dynamics functions © 2012 G.P. Towler / UOP. For educational use in conjunction with Towler & Sinnott Chemical Engineering Design only. Do not copy


Using the GUI: Workbook View

Click here Brings up all the basic stream data such as temperature, pressure, flow rates, etc. in one screen



Windows Can Be Configured to Show PFD & Workbook



Editing the Flowsheet in the GUI

Right click on any vessel or stream icon and you get a menu that allows you to select from similar icons, hide the stream or operation, rotate it, rename it and generally tidy up the drawing to look more like a proper PFD © 2012 G.P. Towler / UOP. For educational use in conjunction with Towler & Sinnott Chemical Engineering Design only. Do not copy


Sub-Flowsheets

You can define a sub-flowsheet and use it as a way of grouping several operations away from the main flowsheet. This is particularly useful when you need several unit operations to model a single piece of process equipment. Sub-flowsheet



Generating Mass & Energy Balance Reports Report manager is on the Tools menu

Define a report


Select all streams, conditions and composition only


UniSim Design Stream Report





Entering Components: Pure Components • Pure components • Component library has thousands of pure components • Mostly organic compounds, but some inorganic compounds

• Rules for selecting pure components • • • • • • •

Always include any compound that has a specified limit in the product Always include any compound that has a specified limit in any process feed Always include anything formed in side reactions or consecutive reactions Always include anything with significant HS&E concerns Usually include anything that is present at >2% (by mole or mass) Usually do not include isomers unless required by the process Usually try to have < 40 pure components

• What is the basis for these rules? © 2012 G.P. Towler / UOP. For educational use in conjunction with Towler & Sinnott Chemical Engineering Design only. Do not copy


Pseudocomponents Crude Oil Boiling Curve Volume % distilled

100

50

0 50

1050

Temperature (F)

• Petroleum fractions can contain ~ 104 to 106 components, many isomers, many compounds that cannot be isolated and identified • Instead, use a pseudocomponent that represents all the compounds that boil in a given temperature range © 2012 G.P. Towler / UOP. For educational use in conjunction with Towler & Sinnott Chemical Engineering Design only. Do not copy


Pseudocomponents Crude Oil Boiling Curve Volume % distilled

100

50

0 50

1050

Temperature (F)

• Example: this pseudocomponent represents all compounds that boil between 300F and 350F, making up roughly 8 vol% of the feed • Simulators have default pseudocomponents, but user may need to add more around critical cut points © 2012 G.P. Towler / UOP. For educational use in conjunction with Towler & Sinnott Chemical Engineering Design only. Do not copy


Solids and Salts • Solids • Some simulators recognize solid phase pure components when they are formed • Phase equilibrium with solid phase is often not well predicted: check the model carefully against the literature • Solid phases of mixed composition usually have to be defined as user components (e.g.: cells, catalysts, coal, paper fibers, etc.) • Some of the simulation programs have good models for solid handling operations, including modeling the effect of particle size distribution

• Salts • Ionic compounds in the presence of water must be treated as electrolytes and require special phase equilibrium models



User Components • Users occasionally need to add components that are not included in the component library • Examples: • • • • •

Complex molecules for pharmaceutical APIs Specialty chemicals Proprietary compounds Advanced solvents Electrolytes



Defining User Components In the Basis environment, select Hypo Components

Create Hypo Component

Enter or estimate properties



Defining User Components Using UNIFAC Groups

Select UNIFAC groups to build up the molecular structure. The program will then estimate properties using group contribution methods



Physical Property Models • All the simulation programs have a range of physical property models • Model selection depends on the system chemistry – see Chapter 4 • Be careful: if the physical property database does not have the model parameters then they may be estimated using methods such as UNIFAC, but estimated parameters should be confirmed experimentally • Models are often inaccurate when predicting LLE, SLE, SSE • When user components are present, models will be near useless unless some experimental data is fitted © 2012 G.P. Towler / UOP. For educational use in conjunction with Towler & Sinnott Chemical Engineering Design only. Do not copy


Phase Equilibrium Model Selection • Chapter 4 has a chart to help with model selection: N Y

Start

T < 250 K

Use G-S

Y Y

H2 present

Use P-R or R-K-S

Hydrocarbon C5 or lighter

N Use B-W-R or L-K-P

Y

Use G-S

Y

Sour Water

Y Electrolytes

Y P < 200 bar

N 0
N

Use R-K-S

N Y

P < 350 bar N Need more experimental data

© 2007 G.P. Towler / UOP. For educational use in conjunction with Sinnott & Towler Chemical Engineering Design only. Do not copy

Use electrolyte

N

N Y

Use sour water system

N

H2 present

N

Y

Y

N

T < 250 K

Use G-S or P-R

Polar or Hydrogen bonding

N

N

P < 4 bar T < 150ºC Y

γi experimental data Y

Use UNIFAC to estimate interaction parameters

Two Liq phases

N Use Wilson, NRTL or UNIQUAC

Y Use NRTL or UNIQUAC

Select model that gives best fit to data


Physical Property Example (Based on a real classroom incident)

I couldn’t get that ethanol water distillation to meet specifications using the Wilson equation, but it worked just fine when I switched it to ideal solution!

• Why? © 2012 G.P. Towler / UOP. For educational use in conjunction with Towler & Sinnott Chemical Engineering Design only. Do not copy




Reactor Models •

CSTR, PFR – –

•

Gibbs reactor – –

•

Solves for defined reactions in sequence to specified conversion function

Yield reactor – –

•

Calculates equilibrium only for defined reactions More useful than Gibbs, as all species seldom reach equilibrium

Conversion reactor –

•

Brings all species present to equilibrium at specified temperature or duty Be very careful to define all possible species if this is what you want

Equilibrium reactor – –

•

OK if you know the kinetics and don’t have many side reactions or contaminants Can be combined to model real types of mixing

Allows user to specify any kind of yield pattern Allows reactions of pseudocomponents, solids, changes in particle size distribution, etc.

Real reactors can often be built from a combination of model reactors, e.g. conversion then equilibrium



Example: Steam Methane Reforming Furnace Reactor

Shift Reactor(s)

Compression

PSA H2

Steam

CO2

Methane

Fuel

CO2 Removal



Steam Methane Reforming Chemistry • Methane reforming: CH4 + H2O → CO + 3 H2 • Conversion of methane is typically about 95 to 98% • Strongly endothermic • Conversion increases with temperature, steam to methane ratio

• Partial oxidation: CH4 + 0.5 O2 → CO + 2 H2 • Strongly exothermic • Reduces hydrogen yield and requires expensive oxygen feed

• Water gas shift reaction CO + H2O → CO2 + H2 • Equilibriates rapidly at temperatures >450 C • Weakly exothermic • Equilibrium favors hydrogen at low temperature © 2012 G.P. Towler / UOP. For educational use in conjunction with Towler & Sinnott Chemical Engineering Design only. Do not copy


Autothermal Reforming Process • Feed methane, steam and oxygen to reactor • Partial oxidation reaction provides heat to drive conversion of steam reforming reaction • Reduces cost of reforming furnace



AspenPlus Simulation of Autothermal Methane Reforming Process

RGibbs

REquil



Autothermal Reforming Reactor Model Results





Distillation Models • Shortcut columns • Assume constant relative volatility • Useful for setting up problems, getting initial estimates of minimum reflux and number of trays and checking feasibility of specs • Not good for non-ideal mixtures • Use to initialize complex columns

• Rigorous Columns • • • •

Solve stage-to-stage Allow column sizing Can be used for absorbers, strippers, distillation, extraction, etc. Allow intermediate condensers, reboilers, side streams, side strippers, etc.

• Prebuilt complex columns • For Petroleum fractionation • Can be customized to different configurations



Distillation Example (Example 4.6)

• Separate 225 metric tons per hour of an equimolar mixture of benzene, toluene, ethylbenzene (EB), orthoxylene (OX) and paraxylene (PX) • Feed is a saturated liquid at 330 kPa • Toluene recovery in distillate should be > 99% • EB recovery in bottoms should be > 99%



UniSim Shortcut Model



Shortcut Column Specifications Note: Toluene mole fraction in bottoms = 1/300 = 0.0033 Ethylbenzene mole fraction in distillate = 1/200 = 0.005

External reflux ratio = 1.15 × minimum reflux (as an initial estimate) © 2012 G.P. Towler / UOP. For educational use in conjunction with Towler & Sinnott Chemical Engineering Design only. Do not copy


Shortcut Column Results



UniSim Rigorous Model



Rigorous Column Specifications

From shortcut model Component recovery can be specified

With good estimate of reflux ratio and number of trays, convergence is fast © 2012 G.P. Towler / UOP. For educational use in conjunction with Towler & Sinnott Chemical Engineering Design only. Do not copy


Generating Column Profiles It is often useful to plot column composition profiles to see whether the column is efficient



Column Composition Profiles



Examples of Bad Profiles

Feed tray too high


Feed tray too low



Toluene in bottoms

Reflux too low (toluene recovery 72%)


Reflux too high (toluene recovery 100%)



Too few trays: toluene recovery = 24.5% © 2012 G.P. Towler / UOP. For educational use in conjunction with Towler & Sinnott Chemical Engineering Design only. Do not copy


Column Sizing in UniSim • Tray sizing is under tools/utilities • Default options (shown) may need changing • Column must be converged with the utility enabled



Column Sizing Results



Common Causes of Column Convergence Problems • Infeasible specifications • Make sure specs on distillate or bottoms purity can be achieved (see Section 17.6.2) • Make sure that specifications can mass balance with two products

• Poor initialization • Use shortcut column to confirm R > Rmin, N > Nmin • Remember stage efficiency is typically 0.7 or less • Remember to allow for some pressure drop across the trays

• Poor initial estimates • Most simulation programs default to the Inside-Out algorithm, which is very fast when given good initial estimates. Use simple specs (e.g. distillate flow rate and reflux ratio) to converge an initial simulation, upload the column temperature profile from this as initial estimates and then change to the real specs and the column should converge quickly.

© 2007 G.P. Towler / UOP. For educational use in conjunction with Sinnott & Towler Chemical Engineering Design only. Do not copy


Complex Columns: AspenPlus PetroFrac Model of Crude Distillation



Other Separation Models • Some simulation programs include models for other separations such as extraction, crystallization, solids separations, etc. • All simulators have a “Component Splitter” model – Allows user to specify recovery of each component – Can be used to model any kind of separation process





User Models • User may need to add custom models to the simulation – Detailed reactor models – Novel unit operations

• Most simulators support two ways of doing this: – Spreadsheet tool – Custom model operation



UniSim Spreadsheet • The UniSim spreadsheet can be used to build simple user models of operations that are not on the palette • Allows import and export from cells to streams • Functionality is basic • AspenPlus has full MS Excel © 2012 G.P. Towler / UOP. For educational use in conjunction with Towler & Sinnott Chemical Engineering Design only. Do not copy


UniSim User Unit Operation Define connections to PFD

Select from palette

Enter code





Processes With Recycle Feed B Feed A

1 3

2

4

Recycle of B

Reactor

6

Lights 5

7

8

Product

• How do we break the recycle loop to solve in sequential mode? © 2012 G.P. Towler / UOP. For educational use in conjunction with Towler & Sinnott Chemical Engineering Design only. Do not copy


Possible Tear Strategy Iterate to convergence

Estimate Feed B Feed A

Update

1

2

4

3a

3b

Recycle of B

Reactor

6

Lights 5

7


8

Product


Tearing at the Reactor Outlet Feed B Feed A

1 3

2

4

Recycle of B

Reactor

6

Lights 5a

5b

7

8

Product

• Which tear point is likely to converge better? © 2012 G.P. Towler / UOP. For educational use in conjunction with Towler & Sinnott Chemical Engineering Design only. Do not copy


Convergence Problems • Results that are unconverged or “converged with errors” cannot be used for design • If convergence is slow then: – Check specifications are feasible • Use hand calculations or simplified models

– Try increasing number of iterations – Try a different algorithm • Default method is usually Bounded Wegstein – can change bounds on acceleration parameter – see Ch4 • Try Newton method if there are many recycles or specifications to meet

– Try to find a better initial estimate • Use hand calculations or a simplified model to initialize the problem

– Try to find a better tear stream – Creep up on the solution © 2012 G.P. Towler / UOP. For educational use in conjunction with Towler & Sinnott Chemical Engineering Design only. Do not copy


Model Simplification Techniques • Complex models with many rigorous columns and recycles can be difficult to converge • A simplified model can be used to initialize tear streams in the complex model • Models can be simplified by: – Using fewer components – Using simpler unit operations (e.g. replace columns with separators) – Eliminating complex user models (replace reactor models with Yield or Conversion reactor) – Reducing the number of specifications (allow some variables to remain not quite converged) © 2012 G.P. Towler / UOP. For educational use in conjunction with Towler & Sinnott Chemical Engineering Design only. Do not copy


Gas Recycle Make-up gas

Purge

Feed Reactor Product

• Don’t forget the purge stream • No purge, no converge! © 2012 G.P. Towler / UOP. For educational use in conjunction with Towler & Sinnott Chemical Engineering Design only. Do not copy




Setting Constraints Using Controllers An “Adjust” controller can be used to control the air flow to give a target turbine inlet temperature



“Adjust” Specifications



“Adjust” Solving Parameters The parameters tab can be used to set bounds to give the desired solution



Flowsheet Optimization • Most of the simulators allow optimization inside the program • AspenPlus manual recommends: • 1. Converge the flowsheet first • 2. Carry out a sensitivity analysis and only optimize the variables that have high impact on the objective function • 3. During the sensitivity analysis, see if the optimum is broad or sharp

• Off-line optimization (or near optimization) is usually much easier – see Ch12.



Tips for Process Simulation • For a good (i.e. useful) process simulation, you must have: – Good component properties – A good phase equilibrium model – Flowsheet design that respects the 2nd law of thermodynamics

• It is not essential to have – Reaction kinetics – Detailed models of every unit operation

• Benchmark the simulation against lab, pilot plant or operating plant data whenever possible to increase your confidence that what you see in the virtual world agrees with reality



Questions ?



4 Process Simulation

Recommend Documents