acm

2000 AspenTech. All Rights Reserved.

Introduction to Aspen Custom Modeler

PresenterPresentation NotesNOTE ABOUT THE NOTES:- most are from version 10.0 and I never had the time to review them- they are for internal use only

3 2000 AspenTech. All Rights Reserved. ACM intro

Day 1

Overview of Aspen Custom Modeler

Description of the graphical user interface Workshop

Review of specification

Modeling language description Workshops

Non-isothermal reactor workshop


Day 2

Arrays and sets

Component properties

Submodels

Snapshots

Tasks


Day 3

Procedures

Scripts and Visual Basic Scripts Automation

introduction to PDE

introduction to Estimation

Information on solver settings


What is Aspen Custom Modeler?

A component of the Aspen Engineering Suite which provides: A language for writing equation based custom models A graphical user interface for both the model developer and the

end user


What is Aspen Custom Modeler?

Solvers containing powerful numerical analysis techniques for the resolution of: Steady state Dynamic integration and initialization Steady state and dynamic optimization Parameter estimation

Task language

Visual Basic Scripting and Automation


FAQ: What is the difference between Aspen Custom Modeler and Dynamics? These products are for 2 different types of situations:

Aspen Dynamics : Running "off the shelf" models based on Aspen Plus simulation flowsheet

Aspen Custom Modeler: Create and run your own "custom" models

You can use the Custom Modeling feature simply by activating the "Custom Modeling" option in Aspen Dynamics program (if you have the both licenses)


Aspen Custom Modeler and Dynamics

Feature ACM AD ACM+AD

Run GUI and calculations

Create new models

Use the Dynamics library

Call Properties Plus yes yes yes


A Simple Example

Objective : a quick tour of Aspen Custom Modeler

PresenterPresentation Notes


Buffer Tank Flowsheet

Fin, Cain

Fout, Caout

height

diameter

level

time

time ?


Buffer Tank Dynamics

How to tackle this problem ? Carefully state what are the objectives : i.e. what is the outlet

concentration as a function of time, in response to disturbances in the feed composition and flowrate

Analyze the system to enumerate the different quantities describing it (variables), and the relations that link all theses quantities together (equations)


Modeling in Aspen Custom Modeler

Write the variables and equations Variables

Decide on units, values, bounds : "Variable types" in ACM Equations

Modeling assumptions : "Model" and "Blocks" Specifications

Develop the simulation model Model, stream and port types in ACM Build the flowsheet and run


Exploring the Graphical User Interface



Aspen Custom Modeler (ACM) files: myflowsheet.acmf

ACM language file (ASCII file) myflowsheet.acmd

ACM document (binary file) mylibrary.acml

ACM library file (binary)

Note: File extensions are displayed if you have let Windows Explorer

options do so.



To start ACM: From the Start menu Create a shortcut

To start ACM and open a file: Double click on the file Drag and drop the file onto the ACM program icon

To exit ACM : File/Exit Allow a small delay before restarting


Running Aspen Custom Modeler

Client/Server architecture Client : Graphical User Interface (GUI) running on Windows

NT/95/98 Intel PC Servers : simulation and task servers

"sim_server" "am_task_server"

Running on the same PC in background May be run on another PC Windows NT Intel


Exploring the Graphical User Interface menus help run mode

flowsheet

simulation messages

content of selected object

explorer run buttons

specification status


Simulation Explorer operation is similar to Windows Explorer

Expand/collapse + and -

Single click on the icon to select

Content of selected object a star = double click

launch action

PresenterPresentation NotesShow the GUI life, ie expand the models folder, and then pick the equations of the model


Simulation Explorer

Flowsheet

Custom Modeling : your working space for new definitions Model types Stream types Port types Variable types Parameters types Procedures Tasks



Simulation Explorer

Libraries - "read only" type definitions Modeler System Library


Flowsheet Window Building the flowsheet with Drag and Drop

PresenterPresentation NotesTo build the flowsheet, you need to select the block model you want to use. Then, with the left mouse button down, move the mouse on the flowsheet area and release the mouse button at the point where you want to place the block.

To connect the blocks, you will use the stream types. Select the stream type you want to connect and with the left mouse button down, move the mouse on the flowsheet area. The ports compatible with the stream type selected will be displayed. Move the mouse close to the port you want to connect and click on the left mouse. Then move the mouse to the place where you can to connect the other end of the stream and again click on the left mouse button to select the connection.


Running the Simulation

Allowed only when the system is correctly specified Check status button or the status window

Select run mode Specification status depends on run mode

Run Pause

Restart Rewind

Reset

Step

PresenterPresentation NotesShow the status buttonShow the run mode


Help

Extensive on-line help available "How to" help Reference help Context sensitive

help Tool tips

PresenterPresentation NotesShow them how to use the help.Show them how to use the 'what's this' helpHow to : good for beginnnersReference : good for everyoneContext help : with the Demonstrate (but think about what you want to demonstrate...)


Messages Window

Displays all messages Loading files, building/compiling types, checking specifications,

diagnostic messages from solvers during solution, ...

Right mouse button to change messages options View/Messages to hide/display window

PresenterPresentation NotesShow how to hide and get it back. Mention that it is cleared when you close it (I don't like it, but this is working as designed, as they say)Show the Windows menu with the list of windowsShow the flowsheet as wall paper option.


Workshop

User interface workshop


Specification


Equation Oriented Modeling

The ACM language is declarative, i.e. you say "what" you want, and leave it to ACM to find out "how" Different numerical algorithms are applied to solve all the

equations simultaneously


Equation Oriented Modeling

The simulation model consists of: Equations Variables Specifications

In a procedural language (like Basic, C or FORTRAN) you would have to provide the calculation algorithm, i.e. "how" to solve

PresenterPresentation NotesBased on Tom Epperly's comment. The whole section could be renamed "Equation Oriented Modeling Concepts". Ok, let's leave that for 10.2 or later. For people with a procedural programming background it is worth warning them about the non-procedural nature of ACM (hoping not to confuse those with no programming background at all). Again, I would *love* to have the time to talk about modeling techniques, but this is something people will have to learn by themselves as time is too limited in the 3 days ACM course.If all of them have a programming experience, then tell them that they can use "procedures" in ACM, which call some fortran or C routines they would have written and interface that with the rest of the resolution. If they are not lost, then tell them that the actual picture is a tiny bit more complicated as we have assignments and tasks, which are executed in yet another way.


Steady state example

Steady State model : 3 equations, 6 variables

h V

Fin

Fout Area k

hAVhkF

FF

out

outin

..

0

=

=

= Fin, Fout as flow_vol; k as realvariable; h as length; V as volume; A as area; 0 = Fin - Fout; Fout = k*sqrt(h); V = A*h;

ACM language

note that the equations need not to be variable = expression, they can be expression = another expression

Equations


Specifications

Degree of freedom Every specification decreases the number of degrees of

freedom of the system by one

Example: for the tank model, if you give 3 variables (fixed variables) out of the 6, ACM can resolve the equations to find the other 3 (free variables)

Degree of Freedom = Number of variables - Number of equations - Number of Specifications


Specifications

Simulation must be have zero degree of freedom We call this a "square" system (see next slide)

The equations and specifications must be independent (i.e. not leading to a structurally singular system) Not every combination of the 3 variables is allowed: for

example you can not fix both Fin and Fout (because Fin - Fout = 0)


Tank specifications for steady state

"*" indicates the incidence of a variable in one equation "Square" when we have 6 rows and 6 columns

variable 1 2 3 4 5 6

equation Fin Fout V A h k

1 0 = Fin-Fout * *2 Fout = k*sqrt(h) * * *3 V = A*h * * *4 spec 1 ?

5 spec 2 ?

6 spec 3 ?

PresenterPresentation NotesThe "*" indicates the incidence of a variable in the equation. So this is the incidence matrix. Usually this is a huge sparse matrix.One thing to find out the correct specification is to match every equation with a different variable. The obvious choice is the diagonal, so in this case this would be to specify A,h and k, which are the natural "design" variables. Don't spend too much time on this slide. I think this is a good mental picture to understand what ACM does, but ACM works even if you don't understand it (ok, sometimes it will do something else that you want -- but then, hey, do you really understand what you told it to do?).It could be a good slide to tell them that not every combination of spec is ok, like for example V,A,h leave only 2 equations to find Fin, Fout and k, so the resolution is not possible. Zero degree of freedom is a necessary condition, but not sufficient. The status button counts the dof, but during a run, the solvability is checked rigorously.


Specifications

Every variable has a "Spec" property which selects its specification mode

Valid values: Spec DescriptionFixed The variable value is specified by the user (A variable

whose value is not being solved for)

Initial A variable whose value is known at time zero for an initialization or dynamic run

RateInitial A state variable whose time derivative is known at time zero for an initialization or dynamic run

Free A variable whose value is being solved for (default)

PresenterPresentation Notesnote that objective and vary are gone... they were defined in 10.0 as the objective function and the freed specificaitons for the optimization resp.STRESS that Free means the solver/acm solves the system to get the value of the free variables, while Fixed means that the user gives the value.


Note

A variable is FIXED for the solver, but can be manipulated during a dynamic run manipulate the value in tables define and activate tasks to implement ramps

time

FIXED

time

FIXED

user edition in a table or task ramp in a task


Checking System Specification

Status button (depends on run mode) Incomplete Underspecified Complete Overspecified Initial state underspecified

- initialization and dynamic runs only Initial state overspecified (idem) Singular


Checking System Specification

Double click on Status check button

Shortcut to Tools Variable find with

preset options

Check legality of specifications

Analyze simulation

specifications

PresenterPresentation NotesFind shortcut : it launches find configured to find selected variables specs. Idea is "I must fix something, show me all the frees so I can fix one"


Analyze Tool

Button on the Specification Status dialog

Menu View, Specification Analysis...

PresenterPresentation NotesUnless you've done it already, this is a good time to demonstrate the feature. The index2 example is available in the demo directory.It finds the specs that make it ok wrt to specifications. Then it is up to you to make some sense... Instead of accepting blindly the changes, you can also make it more "smart".Encourage them to use the tool in the workshops, and don't spend too much time here. The logic is a mathematical matching algorithm, a bit like the one that groups the equations for the decomposition.Make them understand that they need to activate their intelligent deviceTM full gear, not have the damn machine tell them what to do and think! (ie the analyze picks a correct set of specs, not a physically meaningful one -- my recommendation is that you accept some specs from the analyze, then exchange them with physically sound ones -- that can be done in the analyze tool or after (ie to fix the typos in your model or connect streams are good examples of things the analyze will not do for you). YES -- failing to connect streams is the one mistake people do during the training course (and PFS odd behaviours [is is the stick to grid?] is probably not stranger to this...)


Alternative to the analyze tool

Give sensible default to the specification property of the variables in your models

Fin as flow_vol (fixed); Fout as flow_vol; k as realvariable (fixed); h as length; V as volume; A as area (fixed); 0 = Fin - Fout; Fout = k*sqrt(h); V = A*h;

default specifications that make the

model "square"


Dynamic example

Now we have 1 differential equation, 2 algebraic equations, 1 state variable, 5 algebraic variables

Fin, Fout as flow_vol; k as realvariable; h as length; V as volume; A as area; $V = Fin - Fout; Fout = k*sqrt(h); V = A*h;

$variable in ACM language represents the time derivative of the variable

Equations

hAVhkF

FFdtdV

out

outin

..

=

=

=

accumulation


Initial Conditions

Number of initial variables required = number of differential equations

Initial variables : Initial or RateInitial Natural choice : state variables Engineering choice : levels, temperatures


Initial Conditions

INITIAL and FIXED are different things!

Initial conditions : the starting point for the dynamic simulation

time

FIXED

time

INITIAL

0


Initialization run

Consider $V as a new variable variable 1 2 3 4 5 6 7

equation Fin Fout V A h k $V

1 $V = Fin-Fout * * *2 Fout = k*sqrt(h) * * *3 V = A*h * * *4 spec 1 ?

5 spec 2 ?

6 spec 3 ?

7 initial 1 ?


Specifications

One of the possible specifications set

Variable type specificationFin algebraic FIXED

Fout algebraic FREEV state FREEk algebraic FIXEDh algebraic INITIALA algebraic FIXED


Specification Status "Initial Changes"

Shortcut to specify all state variables

Defines all the derivatives in the simulation as initialed to zero

(RATEINITIAL)

Defines all the state variables as INITIAL to their current value


Steady State run with a dynamic model

Steady state is equivalent to time derivative equal to 0 In the example : dV/dt = 0

Specification types "Initial" and "RateInitial" are equivalent as "Free" for a steady state run


Initialization Run

Solve the system's equations at time 0 Fixed variables Initial variables

The value of variable time derivative gives an idea of the direction of change and how fast the system will move in the dynamic run


Dynamic run

Numerical integration proceeds step by step Fixed time step for Implicit Euler method Variable time step for Gear and VSIE Settings in Run/Solver Options

Results are available at every communication intervals Tables, plots, etc Settings in Run/Run Options


Simulation settings

In Run/Solver Options Selects the numerical resolution parameters Integration method Tolerance Diagnostics

In Run/Run options Communication interval Pause simulation at


Overview of the language


Syntax

ACM language is not CaSe SensiTIVE

Mark the end of each statement with ; Free format

You can write over several lines and insert spaces for clarity

Comments : // comment to the end of the line /* multi lines

comments */


Overview of ACM Types Flowsheet

Port Types

Port Types

Stream Types Stream Types

Model Types Model Types

Models submodels

... is declared by ... name AS type;

block

block

stream

stream

port

Parameter Types Variable Types

variable or parameter

PresenterPresentation NotesFor lack of a better picture. This shows the spaghetti bolw that ACM maze can handle. Up to you on how to present this. I think I would just mention this slide as a road map (a crude one). It is quite approximative, as for example a stream model can refer to a model, a script to another script, which could be in a model, etc.

The slide show is actually quite good to repeat the template model of the previous slide and the relations to other types.


Language

Declaration of an instance

Type definition Type indicator :

VARIABLE, PARAMETER, PORT, MODEL, STREAM, ...

instancename AS typename ;

TYPEINDICATOR typename // definition of type END


Names

Names can be up to 27 characters

First character must be a letter

Characters allowed : a-z, A-Z, 0-9, "_"

Keywords are reserved

Keywords are displayed in blue in the editor


Variable Types define defaults for the variables

Value, lower and upper bound

PhysicalQuantity : base units of measurement in the models Either just a unit string to be displayed Predefined names for unit conversion in GUI

VARIABLE MyVariableType Value: 123.4; Lower: 50.0; Upper: 500.0; PhysicalQuantity: "kg/hr (dry)" ; END

PresenterPresentation NotesSee the on-line help for a full description. Of course variable type defines more... But these are the fundamental things, ok?The attributes are:value : default valueupper : upper boundlower: lower boundscale: scaling factorisconnected:true if a control signal is connected on the variabledescription: variable descrptionbaseuomname: base unit of measurement namespec: specification statusrecord:true means the value is saved in historyulagrange:???optimllagrange:???optimtag:???dmcplus and SAX


Variables properties inherits from RealVariable type Property Type DefinitionValue Real current variable valueUpper Real upper bound for variable valueLower Real lower bound for variable value

Scale Real scaling for the variable (the solver sees the value divided by scale)Derivative Real time derivative (for state variables only)Description String Variable descriptionPhysicalQuantity String defines the category of physical unitsUnits String physical units currently usedSpec SpecType free, fixed, initial or rateinitial

Tag String variable tagname (ie for external interface like SAX)

Record Boolean controls whether the history of the variable is recorded


Creating a Variable Type

Simulation/Custom Modeling/Variable types

1: Click here to create a new variable type

2: right mouse button (RMB)

PresenterPresentation NotesSee next slide notes about rebuild vs. compile.


Compiling Types

Compile = a lot of things Analyze the code for syntax Equations and variables Check port connectivity Etc...

Recommendation: Compile as you

create new types

This type has not been

compiled successfully

PresenterPresentation NotesCB.: What is the difference between Compile and Rebuild?Rebuild compiles everything in the order of creation. Compile compiles only the selected object. Because of the order the rebuild is performed, this command is useless (ie if you create a model, then a port, the rebuild will try to compile first the model, so it will fail saying the port type does not exist). My recommendation is that you compile after each edition and never use the rebuild. Note that a rebuild is done when you load the file (it works because the save sorts the sections in the correct order).

Tell them that if they get a green cross, they should call the hotline immediatly. :-) If they get a red cross, have them clear the simulation messages then recompile again. (this is to avoid looking at old error messages... make sure the simulation messages window is large enough...)


Port Types

Port type: Define the variables that can be passed between models

Syntax:

PORT PortTypeName VariableName AS VariableTypeName ; ... END


Example

Pre-defined MainPort (in Modeler):

PORT MainPort F as flow_mol; z(componentlist) as molefraction; T as temperature; P as pressure; h as enth_mol; END

This is an array variable

(syntax to be presented later)


Basic elements of a Model MODEL ModelName // variable and parameter declarations VariableName AS VariableTypeName; ParameterName AS ParameterTypeName; // port declarations PortName1 AS INPUT PortTypeName; PortName2 AS OUTPUT PortTypeName; // Assignments VariableName.property : expression; // Equations EquationName: expression1 = expression2; END

PresenterPresentation NotesMention this slide as a template, just read the various items. Mention the assignements and the equations. Then go to the next slide.


Variable Declaration

Variables are declared on variable types

Examples Tmetal as Temperature (25); Tamb as Temperature (FIXED, 25); Ttank as Temperature (INITIAL); Ptank as Pressure (upper:13); C as Conc_Mole (description:"conc. of X"); a,b as RealVariable;

VariableName AS VariableTypeName; VariableName AS VariableTypeName (AssignmentList);


Assignments

Assignment symbol ':'

Used to assign a default value to an object types properties Variables Parameters

You can overwrite default values in tables for the instance (i.e. the block, the stream)


black if different from default valueblue if equal to default valuewill be updated if you have never edited it beforewill not be updated if you have edited before


Assignments

Examples T as temperature; T.value : 23; // changes the default value of T T.spec : FIXED; // changes default spec. type T.lower : -10; // changes the default lower bound


Defining a Port in a Model

Declaration of a port (in a model type or stream type): PortName : name of the port Direction : INPUT or OUTPUT Defines an anchor point for streams to be connected on a block

using this model

PortName AS Direction PortTypeName (description:"some description");


Accessing Port Variables in Models

Syntax : PortName.VariableName

MODEL tank In_F as input Fport; Out_P as output Fport; V as volume (initial); // material balance $V = In_F.F - Out_P.F; END

PORT Fport F as flow_vol; END


Equations

Equation syntax : Optional "equationlabel" to document the equation (optional but recommended) Useful for diagnostics if things go wrong

$variable : time derivative of the variable (variable.derivative)

equationlabel: expression = anotherexpression; expression = anotherexpression;

PresenterPresentation NotesYou can use the following mathematical operators in equations and assignments:

OperatorUsageDescription+x + yAddition-x - ySubtraction*x * yMultiplication/x / yDivision^x ^ 3Raised to the power ofSINSIN(x)SineCOSCOS(x)CosineTANTAN(x)TangentASINASINx)Arc SineACOSACOS(x)Arc CosineATANATAN(x)Arc TangentSQRTSQRT(x)Square rootSQRSQR(x)Square ofEXPEXP(x)ExponentLOGeLOGe(x)Natural LogarithmLOG10LOG10(x)Logarithm to base 10ABSABS(x)Absolute value ofSIGMASIGMA(x,y,...)Sum values of an expressionPRODUCTPRODUCT(x,y,)Product of a list of arrays and variablesMAXMAX(x,y,)Maximum value in a list of arrays and variablesMINMIN(x,y,)Minimum value in a list of arrays and variablesTIMETIMECurrent simulation timeDELAYDELAY x BY lagApply a lag to the value of a variable


Expressions

Operations: Addition + Subtraction - Multiplication * Division / Exponent ^

Functions LOGE( ), LOG10( ), ... See on line help for details


Icon Editor

To create your own icons for blocks user located

ports

port browser

universal ports

PresenterPresentation Notesyou can create a shape and not plane ports -> get universal ports for any port not placed graphicallyexplain difference in topology editor


Workshop


Parameters


Parameters

Use parameters to represent Structural values, e.g. Number of trays, number of elements in an array Alternative formulations in a model

Rigorous/simplified correlations

Values which are never solved, e.g. Pi, conversion factors, acceleration due to gravity


Parameter Types

Create your own parameter types

IntegerParameter

LogicalParameter

RealParameter

StringParameter Syntax presented later

PresenterPresentation NotesCB. Integerparameter,stringparameter are generic types for parameters as realvariable is for variables.


Integer Parameter Type

Syntax To change from the defaults of the integerparameter parameter

type

PARAMETER ParameterName USES INTEGERPARAMETER Value : DefaultParameterValue ; Description : "description"; Upper : UpperParameterValue; Lower : LowerParameterValue; END


Real Parameter Type

Syntax To change from the defaults of the realparameter parameter

type

PARAMETER ParameterName USES REALPARAMETER Value : DefaultValue; Description : "description"; Upper : UpperParameterValue; Lower : LowerParameterValue; PhysicalQuantity: "physicalquantityname"; END

PresenterPresentation NotesCB.: Parameter types can be defined within the parameter object or directly in a model type, stream type, or port type?Where you want... However I think it is more useful to define them in the parameter types folder (to maintain some consistency). Note that I typically define a parameter type in a model when doing some debugging (ie to switch on/off some equations).

SPEEDUP users will appreciate the fact it is possible to define the units for the realparameter.

MAKE SURE EVERYONE UNDERSTANDS the difference between a parameter (fixed by nature) and a variable (up to the user). A variable is more flexible so I would recommend that in doubt to use a variable. (the only cost is memory usage -- but this should not be an issue when you start using ACM... leave this sort of code optimization for later -- ie be smart : make it run then make it run better (or make it run on larger problems)).


Conditions


Structural and Run-Time conditions

If the condition only uses parameters, then it is evaluated before simulation This is called a "structural IF"

Otherwise the condition is evaluated while running the simulation This is called a "Run-Time IF" Can only change the equations forms

The syntax is very similar


Structural condition

Use to change model structure (equations) depending on e.g. Connections Parameter values

This is called "Structural IF" because: The parts of the model which are active can be determined

before running the simulation These parts will remain the same during the whole simulation


Structural condition syntax

Simple:

Conditions can be nested e.g.

IF condition THEN statements1; ENDIF

IF condition THEN statements1; ELSE statements2; ENDIF

IF condition1 THEN statements1; IF conditions2 THEN statements12; ELSEIF conditions3 THEN statements13; ENDIF ELSE statements2; ENDIF


Logical expressions in condition

Operators (in decreasing priority order) Use parenthesis to change priority order

Operator Meaning== (two equals signs) Equal to

> Greater than< Less than

>= Greater than or equal to


Structural condition

Conditions on parameters, connectivity

Fixed model structure at run time

Can have different number of equations ModeOfOperation as stringparameter; IF ModeOfOperation "PressureDriven" THEN // ModeOfOperation has the value FlowDriven // Find minimum inlet pressure Pin = MIN(In1.P, In2.p); // 1 eq ELSE // ModeOfOperation has the value PressureDriven // Equate inlet pressures In1.P = Pin; // 2 inlets = 2 eqs In2.P = Pin; ENDIF


Checking if a stream is connected on a Port

Built-in Port property "IsConnected" true if a stream is connected on the port false otherwise

Example Out_Vent as output MaterialPort; IF Out_Vent.IsConnected THEN // a stream is connected Out_Vent.F : free; ELSE // no stream connected - fix vent flow to zero Out_Vent.F : 0, fixed; ENDIF


Example of a run time condition: level of liquid changes during simulation

VAPOR IN

LIQ OUT

VAPOR IN

LIQ OUT

VAPOR IN

LIQ OUT

flooded

dry

between


Run Time IF

Model: Atot, Aliq as area; level as length; h_low, h_tube as length; if level > h_low+h_tube then Aliq = Atot; // flooded elseif level < h_low then Aliq = 0; // dry else Aliq = Atot*(level-h_low)/h_tube; // between endif


Run-time Condition Must Be Balanced

Incorrect:

How to correct? Need to have the same number of equations in all branches of

the IF

if flow > 0 then dp = k*flow^2; chock = flow/fmax; else chock = 0; endif

PresenterPresentation NotesHow to correct? No idea! This is a made-up example, without real modelling background. This example should be changed. Any suggestion?I want to find the rules behind why some conditional equations are problematic while others are perfectly fine. Obviously this depends on how the system is specified and not only the equations. Force the model to use the equations of one side of the IF and see how it solves, then the other side, as when the IF switches from one branch to the other, there is a need for continuous robust solution.Examine how the system will be resolvedConvergence difficultiesif a step is large, then variables may move far from convergence region: convergence failureborder caseTroubleshooting: make sure the resolution try different integration methods


Workshop


Stream Types


Stream Types

Built-in stream types: ControlSignal: connecting control variables

Connection: to connect blocks via ports

Define your own stream types to: Define heat streams, work streams, etc Handle heating/cooling, pressure drop along pipes Handle feed streams Control variables in streams

PresenterPresentation NotesCB.: can use online help to show definition of the 2 built-in stream type


Streams on the Flowsheet

Example

LItank lc

s_in

s_out

ControlSignal streams

Streams


Ports on the Flowsheet

Connect input port and output port with the selected stream type

LItank lc

s_in

s_out

Ports OUTPUT -INPUT Variables

OUTPUT - INPUT

ControlSignal stream


ControlSignal Stream Type

ControlSignal is a built-in stream type used to connect a variable marked for INPUT to a variable marked for OUTPUT

Note that both the INPUT, OUTPUT qualifiers together can be used only for variable declarations and not for port declarations

// Q can be the destination of a control signal Q as INPUT enthflow; // Level can be the source of a control signal Level as OUTPUT length; // F can be the source or the destination of a // control signal F as INPUT,OUTPUT flow_mol;


Connection Stream Type

Connection stream type is a built-in stream type which allows any input port to be connected to any output port

Port types can be different Variables are matched by name Unmatched variables reported in Simulation Messages window

processing connection between B1.out_p and B2.in_f, the following variables don't match B1.out_p.y

PresenterPresentation NotesNote in SPEEDUP it was the variable order that was used for matching the variables.


Stream Types

A stream type has the characteristics of a model, except; One input port and/or One output port

Like a model, a stream type can have: Equations and variables Tables, Plots

PresenterPresentation NotesLine style can be changedInheritance (STREAM ... USES ...)


Stream Equations

Stream variables

STREAM Fstream F as flow_vol; In_F as input Fport; Out_P as output Fport; Out_P.F = In_F.F; F = In_F.F; END


Stream Properties

You can change the line style for the streams

Select stream type, then click on right mouse button


Workshop


Additional details


Note about assignments with expressions

Example : parameter : expression

Assignments work in a similar way as a spreadsheet does. Defining dependencies between the parameters

Example: Ntotal : Ntop + Nbottom If you change the value of a parameter, then the list of

dependencies is examined and assignments executed

Do not use variables on the right hand side of assignments


What is Inheritance in ACM?

Add properties or override default values

VARIABLE Flowrate PhysicalQuantity : "kg/hr"; upper : 10; lower : 0; END

VARIABLE Flowrate_big USES Flowrate upper : 1000; END

VARIABLE RealVariable // content as on previous slide END

keyword for inheritance

inherits from the generic type

uses

uses


Variable Declaration Qualifiers

Additional flags HIDDEN : does not show the variable on Forms GLOBAL : makes the scope of variable "global"

Tamb as GLOBAL temperature; Pamb as GLOBAL pressure; coef as HIDDEN realvariable;


The Global Qualifier

GLOBAL makes the variable global (i.e. defined once) for the whole simulation

Global

Simulation


Compatibility of Ports

You can connect ports which are of the same type or if the type of one port inherits from the other port type

See example "PortCompatibility.acmf"


Delay Function

Implements a delay function It saves the past history and plays it back later, depending on

the dead time

time

x y td

dead time

x, y as fraction; deadtime as time_; y = DELAY x BY td;


Model Development Workshop


Model Development Workshop

Build a model for a non-isothermal CSTR reactor, cooled with a jacket

Suggested approach Decide how to implement the problem i.e. how many models Create your own variable types Keep it simple!


Modeling Language: Sets and Arrays


Sets

Set (group of elements) can be: INTEGERSET : integer elements STRINGSET : character string elements

Usage, e.g. Declaration for arrays Iteration index for equations Validation of entries (i.e. string parameter types)

Options for a model


Integer Sets

Declaration of an integerset called "MySet":

Assignment ":" of elements to the set: Enumeration

Range operator ":" increment is always 1

Declaration and assignment in one line:

MySet as integerset;

MySet : [3,4,5,6];

MySet : [3:6];

MySet as integerset ([3:6]);

This is the range operator


Integer Sets

Integer parameters can be used to specify the bounds of the set The parameters can be changed in tables, and the set will be

re-assigned automatically

First as integerparameter (4); Last as integerparameter (10); MySet as integerset ([First:Last]);


String Set

Declaration of a stringset called "YourSet":

Assignment ":" of elements to the set: Enumeration

There is no range operator for stringsets

Declaration and assignment in one line:

YourSet as stringset;

YourSet : ["alpha","beta"];

YourSet as stringset (["alpha","beta"]);


Arrays

One or more sets can be used to define the index values for an array Array of variables, array of parameters

Accessing elements of the array: array(index)

N as integerparameter (10); T([1:N]) as temperature; // array of vars w([1:N]) as realparameter;

T_in = T(1); // just a few examples! T_out = T(N); T = w;


Arrays

String sets can also be used to dimension the arrays:

pona as stringset (["paraffin", "olefin", "naphtene","aromatic"]); analysis(pona) as fraction; // example of accessing one element of the array rate = 11.04*exp(-1966/T)*analysis("aromatic")^0.34;


Multidimensional Arrays

You can declare multidimensional arrays

Example : 2D array M as integerparameter (10); N as integerparameter (3); Mset as integerset ([1:M]); Nset as integerset ([1:N]); T(Mset, Nset) as temperature; T(5,2) = T(1,3)+T(3,1); // just an example!


Operations between sets

Union : A + B or UNION(A,B)

- Example : 1,2,3,4,5,6

Intersection : A*B or INTERSECTION(A,B)

- Example : 3,4

Differences : A - B or DIFFERENCE(A,B)

- Example : 1,2 A \ B or SYMDIFF(A,B)

- Example : 1,2,5,6

A B 1

2

3 4

5

6


SIZE function

SIZE (set) returns the number of elements in the set

Example to initialize molefractions ComponentList is the set of components SIZE(ComponentList) returns the number of components "x : value" assigns the same value to all the elements of the

array x

x(ComponentList) as molefraction; x : 1/(SIZE(ComponentList));


SIGMA function

Sums up a list of variables: The SIGMA command is interpreted as:

More typically, SIGMA is used to sum elements of arrays: x("N2") + x("O2") + x("CO2") = 1.0

A1, A2, A3, B1 AS RealVariable; B1 = SIGMA (A1, A2, A3, 2.0 );

B1 = A1 + A2 + A3 + 2.0;

Components AS StringSet(["N2", "O2", "CO2"]); x(Components) AS Fraction; SIGMA(x(Components)) = 1.0;


SIGMA function

The following statement is equivalent to the previous SIGMA expression (index slice is implicit):

You can use SIGMA on the whole array, or a selected slice of an array:

SIGMA(x) = 1.0 ;

nsect AS IntegerParameter(10); Sections AS IntegerSet([1:nsect]); Tmean, Temp(Sections) AS Temperature; Tmean = SIGMA(Temp([2:nsect]))/(nsect-1);


SIGMA function

Operations between sets can be used to select the elements of the array Example:

Sum the fractions of all the components in a component list except for water

The minus sign "-" in the expression above is the set difference operator

OrgFrac AS Fraction; x(ComponentList) AS Fraction; OrgFrac = SIGMA(x(ComponentList - ["WATER"])) ;


FOREACH

Gives more control on the enumeration of the elements of a set in SIGMA function Example

Evaluate the sum of the diagonal elements of a matrix which is the sum of A(1,1), A(2,2), A(3,3), ..., A(n,n)

n as integerparameter (3); A([1:n],[1:n]) as realvariable (fixed); s as realvariable; s = sigma(foreach(i in [1:n]) A(i,i));


Functions of arrays

Functions working in a similar way to SIGMA and return a scalar:

PRODUCT(x,y,...) MIN(x, y, ...) MAX(x, y, ...)

Note : FOREACH can be used in all of these

The other functions apply element by element and return an array: EXP(x), SIN(x), LOGE(x), LOG10(x), etc...


FOR...DO...ENDFOR

Syntax:

"set" can also be an operation between sets of the same kind

No declaration required for the index variable It exists only within the FOR loop

FOR indexname IN set DO statements; ENDFOR


FOR...DO...ENDFOR

Example to repeat the assignments: n as integerparameter; position([1:n]) as realparameter; dx as realparameter; position(1) : 0; dx : 1/n; FOR i IN [2:n] DO position(i) : position(i-1) + dx; ENDFOR

position(1) position(i) dx position(n)

PresenterPresentation NotesMike comments: not sure it is a good example: assignement in loops may not do what is expected, ie you can't use a loop to accumlate a value.The example is correct. The point to make clear is that x : x + something is not allowed. Use a script!!! Use assignements for default values and simple things. Anything a bit involved should go in scripts, where you can control the order of execution of the assignements.

Note about the example : we could have said position(i) : position(1) + i/n. You should just say we used the form here on this slide to stress that position(i) : position(i-1); position(i) : position(i) + dx; is incorrect.


Arrays in equations

Example: This is repeated over the intersection of the sets that dimension the arrays, so this defines several equations (3 in this example)

F,L,V as flowrate; ns as integerset ([1:3]); z(ns), x(ns), y(ns) as fraction; F*z = L*x + V*y;

PresenterPresentation NotesThe loop :for i in ns do F*z(i) = L*x(i) + V*y(i);endfor

is in fact:for i in ns*ns*ns do F*z(i) = L*x(i) + V*y(i);endfor


Arrays in equations

Same equations with a FOR loop:

FOR i IN ns DO F*z(i) = L*x(i) + V*y(i); ENDFOR


Equations With Implicit Slices

Example with different dimensions: nsF as integerset ([1:3]); z(nsF) as fraction; ns as integerset ([1:2]) x(ns),y(ns) as fraction; F*z = L*x + V*y; F*z(1) = L*x(1) + V*y(1); F*z(2) = L*x(2) + V*y(2); // F*z(3) = ???


How to control of the repetition of equations? Write explicitly the index slices, unless all array variables

in the equation have the same index Example : x(ComponentList) instead of x

Expand your equations in FOR...ENDFOR loops Unless your code is simple, this makes your intentions more

clear


Examples

Evaluation of a polynomial: MODEL pol n as integerparameter (3); coef([0:n]) as realvariable (0, fixed); x as realvariable (fixed); y as realvariable; y = sigma(foreach (i in [0:n]) coef(i)*x^i); END


Two Dimensional Arrays - Matrix

Sum of two matrices with FOR loops: MODEL MatrixModel1 ms as integerset ([1:2]); ns as integerset ([1:3]); a(ms,ns) as realvariable (FIXED); b(ms,ns) as realvariable (FIXED); c(ms,ns) as realvariable; for i in ms do for j in ns do c(i,j) = a(i,j)+b(i,j); endfor endfor END

PresenterPresentation Notesthe two for loops are not needed -- you can really write this as c = a + b. This is a good example to show the matching of the index (index 1 of C with index 1 of A and index 1 of b --- the matching works on the dummy index so c(1,ns) = a(ns, 1) + b(ns,2) would be a disaster...


Two Dimensional Arrays - Matrix

Sum of two matrices with implicit repeat:

- Repeats over intersection of first index on left hand side with first index on right hand side, and repeats over intersection of second index of lhs with second index on rhs, etc...

MODEL MatrixModel2 ms as integerset ([1:2]); ns as integerset ([1:3]); a(ms,ns) as realvariable (FIXED); b(ms,ns) as realvariable (FIXED); c(ms,ns) as realvariable; c = a + b; END

PresenterPresentation Notesthe two for loops are not needed -- you can really write this as c = a + b. This is a good example to show the matching of the index (index 1 of C with index 1 of A and index 1 of b --- the matching works on the dummy index so c(1,ns) = a(ns, 1) + b(ns,2) would be a disaster...


Matrix Product

Definition: SIGMA lists the elements of arrays a and b using the set

"ns", like the dummy index k in the definition

kjk

ikij bac =MODEL MatrixModel3 ms as integerset ([1:2]); ns as integerset ([1:3]); ps as integerset ([1:4]); a(ms,ns) as realvariable (FIXED); b(ns,ps) as realvariable (FIXED); c(ms,ps) as realvariable; for i in ms do for j in ps do c(i,j) = SIGMA(a(i,ns)*b(ns,j)); endfor endfor END

PresenterPresentation NotesAlternative version (as in the help) for i in ms do for j in ps do c(i,j) = SIGMA(foreach(k in ns) a(i,k)*b(k,j)); endfor; endfor;

Explain the forms you prefer... the foreach is probably easier to explain


String Parameter Type Syntax

The declaration requires a stringset

See example on next slide

PARAMETER ParameterTypeName USES STRINGPARAMETER Valid AS STRINGSET( ValidList ); Value : DefaultParameterValue ; END


String Parameters and Sets

The property 'valid' defines the pull-down list in tables

PARAMETER AgitatorType USES STRINGPARAMETER valid as stringset (["paddle", "pitched blade turbine", "disk", "propeller", "anchor", "anchor-high", "helical-ribbon"]); value : "paddle";// default value END MODEL Mixer agitator as AgitatorType; ... END

PresenterPresentation Notesvalid has a special meaning for table. It is not a built-in property for the parameter type, but it is detected by tables to populate the pull-down menu.


Multiports


Multiports

To allow several streams to be connected at the same port

To make a port property a multiport use the "MULTIPORT OF" declaration qualifier

PortName AS INPUT MULTIPORT OF PortType; PortName AS OUTPUT MULTIPORT OF PortType;


Multiports

A multiport has 2 special properties ConnectionSet : a set, containing the names of streams

connected to the port Connection : an array of port type, one for each stream

connected to the multiport

Accessing the variables PortName.connection("streamname").VariableName

PresenterPresentation NotesThe example is in strawberry box, because althrough the code is correct, it is not a good idea to hard-wire the name of a stream in a model.note that this is very bad as people renaming streams in AD have discovered


Typical FOR loop with Multiport

Accessing the variables of all the streams connected

FOR i IN PortName.connectionset DO PortName.connection(i).VariableName = ... ENDFOR


Changes in a model to declare a multiport

Example to make In_F a multiport (before changes) PORT Fport F as flow_vol; END MODEL tank In_F as input Fport; Out_P as output Fport; V as volume (initial); // material balance $V = In_F.F - Out_P.F; END


Changes in a model to declare a multiport

Example to make In_F a multiport (after changes)

MODEL multitank In_F as input MULTIPORT OF Fport; Out_P as output Fport; V as volume (initial); // material balance $V = SIGMA(In_F.connection.F) - Out_P.F; END


Another Example

Steady state energy balances of a mixer: In_F as input multiport of MainPort; Out_P as output multiport of MainPort; hmix as enth_mol; // outlet enthalpy definition (mixing) SIGMA(In_F.connection.F)* hmix = SIGMA(In_F.connection.F*In_F.connection.h); for i in Out_P.connectionset do Out_P.connection(i).h = hmix; endfor

hmix


Workshop


Tables and Plots (Forms)


What is a Form?

Standard "Forms" Table Plots

Plot : time series Profile Plot (i.e. temperature profile in a reactor bed, etc)

History Table

Customized forms


Forms can be created at different levels

Forms can be created At the model level

All the model instances will get this form At the flowsheet level (from the Tools menu)

Global table in Simulation (top level) Shows global variables and parameters


Table Forms

Useful to show a selection of variables Forms in models types and streams types

AllVariables table shows all active variables and all parameters (automatic)

User defined Forms in flowsheet

LocalVariables table shows all active variables in flowsheet constraints User defined

PresenterPresentation NotesIt is not possible to create a table in the Simulation folder, only Global.


Time Series Plot Form

Plot the values of the variables vs. time

Must create the plot before running the simulation in order to record the variable history Unless you change the run options dialog (Run/Run Options) to

record the history for all variables Or set the record property to true for the required variables

Can be viewed as a History Table

Plot properties to change various display options


Profile Plots

Plots the values of an array t_profile

Index0 4 8 12 16 20 24 28 32 36 40 44 48 52

solid

tem

pera

ture

C T

ime:

81.

0000

00ga

s te

mpe

ratu

re C

Tim

e: 8

1.00

0000

gas

tem

pera

ture

540

550

560

570

580

590

600


Creating a Profile Plot

Go to Tools/New Profile Plot or the button

Selection of variables to plot - no drag&drop, sorry!

Definition of Profile Variables, using RMB


Creating a Profile Plot Name for the curve

Curve number - increase to create other curves on the same plot

Variable to plot * stands for the whole array

Values for the X axis * stands for the whole array

PresenterPresentation Notesthis dialog is not easy to use. Do the demo of the PDE bed, and tell them they will repeat the steps in the workshop.


Handling Components

PresenterPresentation NotesExplain how to handle the components:- component sets- component lists (ie with properties)


Components Lists

Define the components and properties With "external" physical property calculations

Using Properties Plus Or using your own package

Without physical property calculations Just to define the name of the components

PresenterPresentation NotesLaunch the section by saying why components are needed in process simulation. Examplesmaterial balancesreactions kineticsetc...In SPEEDUP 5, we had to map the components to numbers!!! That was very confusing! Now instead we can use meaningful (hum???) names.without phys proprs include very simple ones like constant densities, or things you write in a few lines in a submodels, not a full blown equation of state with Huron-Vidal mixing rules using UNIFAC group contribution method, if you see what I mean.


Components Lists

Have these components defined "globally" i.e. at the simulation level

So that any port, stream and model can use it To declare arrays To call procedures


Component Lists

Component Lists in Simulation "Default" component list

"ComponentList" property in Flowsheet Blocks Streams Ports


ComponentLists with physical properties reference "List" : with "external" physical properties

Set of the component names (component set)

Selection of thermodynamic options - E.g. Property Method specifications for Properties Plus

A reference to a properties definition file Only one file can be selected prop.appdf

010101010 101010101 010101010 101010101

"CH4" "H2O"

PresenterPresentation NotesThe words "Set" and "List" are misused consistently in ACM. The problem is that a ComponentList as the name says seems to be a list, but we have just talked about sets a few sections ago. So what? To me this naming is still complete non-sense, but as the CDs have been burned, there is no way back. So just try to make it clear. The most painful thing is the dialog with the mysteryous questyon "Is a ComponentSet". Note that a ComponentList of type List is not actually linked t o Aspen Plus directly but via the gpp.dll. If you want (and a few customers do that) you can create your own gpp.dll (via something called GPI = Generalized Properties Interface) : the standard gpp.dll actually calls the Aspen Plus routines. See the help for further details. Maybe when a Aspen Properties would be available, we'll need to remove the reference to Aspen Plus (which is percieved more as a flowsheeting program -- but PropPlus runtype is just what we use here -- this is still Aspen Plus -- minus the flowsheet)


Components list without reference to physical properties "Set" :

Set of component names

"SALT" "PEPPER"


ComponentList with physical properties

BKP file

APW file APPDF file

Aspen Plus

Aspen Custom Modeler Properties

ComponentList

Procedure

Parameter databanks

gpp.dll

010101010 101010101 111100001 010110110

"CH4" "H2O"


Creation of the properties definition file

Start Aspen Plus GUI

Select the Properties Plus run type

Select the components and physical property methods

Run the simulation

Save the simulation as an Aspen Document Extension apw This creates the properties definition file (extension appdf) at

the same time

A+

PresenterPresentation NotesI would recommend that at this stage you demonstrate the steps. If they all know Aspen Plus, have them start the workshop.


Component List Setup

Back in ACM, double click on the Default component list Its initial type is a 'List' (with properties)


Properties definition file selection

Select the APPDF file you want to use

PresenterPresentation NotesNote that the limitiation in 10.0 is removed -- you had to exit ACM to load another properties file.


Selection of the components

You can define as many components as you like in the Aspen Plus file and use just a few in ACM


Physical properties settings

Settings: See on-line help for definition and defaults


Property calculations using procedures

Procedures are in Modeler library See on-line help for details See procedure code comments

Example : molar enthalpy for liquid phase

// call to a procedure call (output list) = pProcedure_Name (input list);

call (hl) = pEnth_mol_liq(T, p, x);


Example of using the componentlist for a flash calculation

MODEL ExampleModel T as temperature (fixed); p as pressure (fixed); vf as vapfraction; hl as enth_mol_liq; hv as enth_mol_vap; h as enth_mol; z(Componentlist) as molefraction (fixed); x(Componentlist) as molefraction; y(Componentlist) as molefraction; sigma(z) = 1; h = vf*hv + (1-vf)*hl; call (y,x,vf,hv,hl) = pFlash(T,p,z); END


Example

Result of a steady state run

Liquid Vapor equilibrium calculated using Properties Plus, i.e. RK-SOAVE property method, for the given T, P and global composition z

PresenterPresentation NotesThe comment is just there to explain what the table shows. It shows the results, nothing more to explain.


Physical Property Calculation

If the componentlist is not explicitly specified in the procedure call, the built-in ComponentList is used Example

call (hl) = pEnth_mol_liq(T, p, x) Out_P.Componentlist;


Using Several Componentlists

Explicitly state the componentlist to use MODEL HtxModel Cl_p1 as componentlistname; Cl_p2 as componentlistname; p1_in as input MainPort (ComponentList : Cl_p1); p1_out as output MainPort (ComponentList : Cl_p1); p2_in as input MainPort (ComponentList : Cl_p2); p2_out as output MainPort (ComponentList : Cl_p2); // properties call (p1_out.h) = penth_mol (p1_out.T, p1_out.p, p1_out.z) p1_out.ComponentList; call (p2_out.h) = penth_mol (p2_out.T, p2_out.p, p2_out.z) p2_out.ComponentList; END


Component Sets

No physical properties

The Component set can be used as stringset, but defined globally (i.e. at the simulation level)

ComponentSet "SALT" "PEPPER"


Creating a Component Set

If you want to use the "Default" component list, you need to convert it to a "set" RMB

Double click on "Add Component list" Check "Is Component Set" to create a component set (without

properties)

PresenterPresentation NotesI know the dialogs are from 10.0 -- so what?


Component Sets

Enter the name of the components


Workshop


Modeling Language - Submodels


What is a sub model?

Declaration of an instance of a model within another model or stream type

The result is that all the equations and variables of the submodel are integrated in the model of stream type

Benefit : modular approach to modeling


Example for a distillation column model

Blocks and streams within the model

feed

distillate

residue

condenser

drum stage(1)

reboiler

stage(2)

stage(3)


Syntax for the declaration the blocks

block_name AS model_name

MODEL simpledist nstages as integerparameter (3); feedstage as integerparameter (2); cnd as condenser; drm as drum; reb as reboiler; stage([1:nstages]) as simpletray;

model


Syntax for the declaration of the port connectivity

MODEL simpledist ... feed as input processport; dist as output processport; rebo as output processport; link feed and stage(feedstage).feed; link dist and drm.out2; link rebo and reb.liq_ou;

link ports of

same direction

LINK port AND block_name.port


Syntax for the internal streams

CONNECT block.output AND block2.input CONNECT stage(1).vap_ou AND cnd.vap_in; CONNECT stage(1).liq_ou AND stage(2).liq_in; CONNECT stage(1).liq_in AND drm.out1; CONNECT stage(1).vap_in AND stage(2).vap_ou;

cnd

drm

stage(1)

stage(2)

vap_ou

vap_in

out1 vap_in liq_ou

liq_in

vap_ou liq_in


Connectivity of the stages

Complete code // internal streams connect stage(1).vap_ou and cnd.vap_in; connect cnd.liq_ou and drm.liq_in; connect stage(1).liq_in and drm.out1; for i in [2:nstages] do connect stage(i-1).liq_ou and stage(i).liq_in; connect stage(i-1).vap_in and stage(i).vap_ou; endfor connect stage(nstages).liq_ou and reb.liq_in; connect stage(nstages).vap_in and reb.vap_ou;

PresenterPresentation NotesNote that the feed port for the stage has to have some automatic specification like below to avoid hunting for missing specs.MODEL stage liq_in as input Fport; vap_in as input Fport; liq_ou as output Fport; vap_ou as output Fport; feed as input Fport;

liq_in.F + vap_in.F + feed.F = liq_ou.F + vap_ou.F; liq_ou.F = vap_ou.F;

if not feed.isconnected then feed.F : fixed; feed.F : 0; endifEND


Submodel Variables

Accessing variables in submodels This is equivalent to :

block as model (submodel variable = model variable submodel parameter : model parameter);

block as model; // block AS model block.variable1 = variable2; block.parameter3 : parameter4; // some equation $variable5 = exp(-block.variable6/Rgas/T);


Note

The equations required to access the variables in the submodels will be eliminated when using the equivalencing option Significantly decreases the size of the mathematical system to

be solved

PresenterPresentation NotesJust a comment if user is concerned about this lots of variables and equations. The equivalence will get compress these equations and variables, so at the end there is no penalty in using the submodel. Use submodel to make your models more modular.


Heat Exchanger example

Flow pattern (counter current)

1 2 3

3 2 1

Q Q Q

PresenterPresentation NotesAnother example -- see the file htx.amcf for all the details.


Heat Exchanger example MODEL htx h_in as input fthcpPort; h_out as output fthcpPort; c_in as input fthcpPort; c_out as output fthcpPort; // modeling a counter-current heat exchanger n as integerparameter (3); ns as integerset ([1:n]); h_cell (ns) as cell; c_cell (ns) as cell; // see next slides ...



MODEL htx ... // linking model ports with sub-model ports link h_in and h_cell(1).p_in; link h_out and h_cell(n).p_out; link c_in and c_cell(1).p_in; link c_out and c_cell(n).p_out; ...


MODEL htx ... // internal heat stream q_intern (ns) as QTAstream; for i in ns do connect h_cell(i).q_out and c_cell(n-i+1).q_in with q_intern(i); endfor ...


Internal heat streams using stream type "QTAstream"


... // internal material stream h_cell_out (ns - [n]) as fthcpStream; for i in ns - [n] do connect h_cell(i).p_out and h_cell(i+1).p_in with h_cell_out(i); endfor c_cell_out (ns - [n]) as fthcpStream; for i in ns - [n] do connect c_cell(i).p_out and c_cell(i+1).p_in with c_cell_out(i); endfor


Internal material streams using stream type "fthcpStream"


VIRTUAL and EXTERNAL

Model AverageValue VIRTUAL ValueType USES RealVariable; NValues as external IntegerParameter; Value([1:NValues]) as external RealVariable; Average as ValueType; Average = sigma(Value)/NValues; End

Model Pipe NSections as integerParameter; T([1:NSections]) as Temperature; // Use submodel to calculate average T_Av as AverageValue(NValues is NSections, Value is T, ValueType is Temperature); Q = U * WallArea * (T_Wall - T_Av.Average); End


Workshop


Snapshots and Results


Snapshots

What is in a snapshot? The name of the variables, their value and their specification Internal data (delay function workspace and procedure

workspace)

How to create a snapshot? Created automatically at specific events

Initialisation, file saving, ... At user specified intervals

Taken manually by user


Snapshots vs. results

Snapshots Relative to the current simulation structure

Results are "archived" snapshots Once you change the simulation structure, all snapshots

become results

Differences: Results contain only variable values, not delay nor procedures'

workspaces Results can not be used for rewind

PresenterPresentation NotesCB: Archived snapshots are binary files stored in a directory named with the name of the open ACM file. Can we control where these directories are stored or are they always stored in the working folder?No.


Snapshot vs. kept results

Tools/Settings/Snapshot for options on automatic handling

We can "mark" a snapshot so that it will be kept as a result (kept) results : saved in the acmf file Snapshots : archived in the simulation server working directory

(*.snp)


What's the use of snapshots?

To restart a simulation To time 0 for dynamic simulation (snapshot 'Dynamic

Initialization')

To rewind the simulation To any timed snapshot

To copy values from other results

snapshot management


Rewind and Restart

Rewind lets you select the snapshot to which you want to go

Restart returns a dynamic simulation to its solved initial state Uses the 'Dynamic Initialization' snapshot

Run Pause

Restart Rewind

Reset

Step


Copy Values and Advanced Copy

To copy variables from one result/snapshot to the simulation

Wildcards for selection of variables


Workshop


Tasks


Tasks

Implement actions during the simulation Change the value of some variables Write messages Suspend the simulation Create a snapshot

PresenterPresentation Noteslimitation: it is not the only limitation. Should we put this in instrruction notes because it will change (maybe, but when?) Or may be a list of known issues as appendix (this list is not available) or do the same as AP10 (what do we do for AP10, not document?)


Tasks

Types of tasks Event-driven tasks Callable tasks (parameterized)

Definition of tasks In the flowsheet In a model In the library task folder (callable task only)


Task manager

Overall picture Graphical User Interface

Simulation server Task manager

create/remove tasks activate/deactivate tasks

query variable

limit time step

change variable value

tasks task status action tk1 active wait tk2 active done ...

PresenterPresentation NotesCB. Can you write a few tips to comment this slideThis is to show that when we run ACM we have in fact 3 programs running: the gui, the simulation server (the number cruncher) and the special guest of this section, the task server or manager (not sure about its name). So we know that the GUI talks to the sim server and back. The tasks that we define in the GUI are being sent to the task manager. It then says in the simulation messages what is going on, talks to the sim server to step to the time events and locate the implicit events. It maintains some sort of table of tasks, which can be active or not, and executed.


Event driven task

Syntax:

Condition: time == value expression1 >, =,


Event driven task

You need to activate the event driven tasks before running the simulation Flowsheet tasks For tasks defined in the models, you need to activate them in

the blocks using these models


Tasks

Event-driven task status is checked at communication intervals

Task executed whenever the condition becomes true (i.e. changes from false to true)

task xmp runs when level > lev_alarm

time

level

lev_alarm


Tasks

"ONCE" qualifier

task bursting runs once when p > p_burst

time

p

p_burst


Task statements

Variable assignment (variable must be fixed!) Direct Ramping functions (RAMP and SRAMP)

TASK Task4 RUNS WHEN TIME == 4.0 // Flow changes to 5.0 linearly // over a period of 2 time units RAMP (Input1.flow, 5.0, 2.0, continuous); // Temperature changes with an S-shaped curve // to 15.0 over a period of 3 time units SRAMP (Input1.temp, 15.0, 3.0); Input2.FlowX("water") : 0; END


Ramp statements syntax

Linear ramp discrete

Linear ramp continuous

Sine ramp discrete

Sine ramp continuous

RAMP (variable, finalvalue, duration, discrete);

RAMP (variable, finalvalue, duration, continuous);

SRAMP (variable, finalvalue, duration, discrete);

SRAMP (variable, finalvalue, duration, continuous);


Task statements are executed in sequential order

4 time

15

5

6

0

9

first ramp

(continuous)

second ramp

(sramp) step change

(assignment)


Task statements

Create a snapshot

Print a message

Pause the simulation

TASK Task2 RUNS WHEN Time == 10.0 CREATE SNAPSHOT "Task-Created Snapshot #1"; END

TASK Test3 RUNS WHEN Time == 1.0 PRINT "Start Task Test3"; RAMP (Input1.Flow, 2.5, 5.0); PRINT "Task Test3 Finished"; PAUSE; END


Task execution : WAIT FOR and WAIT

WAIT FOR : To hold the task execution until condition is satisfied:

WAIT : To hold task execution for a given period

WAIT FOR condition;

// wait for 2 hours from now WAIT 2.0; // start feed S34.F : 400; // wait now for 1 hour WAIT 1.0; // stop the feed S34.F : 0;


Task execution : RESTART WHEN

RESTART WHEN condition;

Example: $y = x - y

x

Time Hours0 1 2 3 4 5 6 7 8 9 10

B1.

x B

1.y

0.5

11.

52

2.5

33.

54

Task res runs when time == 1 // step change of x B1.x : 4; // wait 2 units of time wait 2; B1.x : 0; restart when B1.y < 1; End

RESTART makes the task execution jump back to the first line


Callable task

Syntax

Call

TASK TaskName ( ParameterList ) TaskStatements end

CALL TaskName (ParameterList);

PresenterPresentation NotesTask language is another world... It uses a ";" to separate the arguments. It is so odd... I don't mention it because I secretly hope that the language will be improved before users go in these arcanes... (and I am not the only one to do so, the help does not mention more...)


Callable task example

Flowsheet tasks: Task dorecipe runs when time == 0.5 final_x as realparameter; final_x : 10; call recipe (final_x); End

pt

Time Hours0 1 2 3 4 5 6 7 8

B1.y

B1.x1

510

15

Task recipe (x as realparameter) B1.x : 3; wait for time == 1; B1.x : x; End


Task execution : PARALLEL

PARALLEL Each action is executed until it completes The parallel section completes when all actions in it are

complete.

TASK Task1 RUNS WHEN TIME == 1.25 PARALLEL CALL P1; // a ramp with duration=2 CALL P2; // another ramp with duration=1 ENDPARALLEL; Input4.Flow : 0.0; END


Task notes

Values must be expressed in their base units of measurement

Snapshots taken when a ramp is running will be flagged as "Modified"

Only fixed variables can be ramped

An event-driven task must be activated

Inactive Active

PresenterPresentation Notes"They" (you know who I mean) say this is because the snapshot is taken after the task does its job, but I say this is a flaw in the design of snapshots and tasks. Like it or not, the snapshot is flagged as non-converged. (I don't like it).


Task notes

Task state is checked at communication interval for all methods, except Gear and time-based conditions

communication interval

task action event

WAIT FOR x > 3

x

PresenterPresentation NotesGear locates the event. The difference is small. Note : strongly advise making the communication interval small. To me, this is a defiency, but the reason is that we have worked to improve the efficiency (at the expense of the user friendliness)


Workshop


Procedures

PresenterPresentation NotesThe procedures are used to interface ACM with external code, like a fortran subroutine. The presentation is split in two sections: first we show how to use an existing procedure. This case is important, as physical properties can be evaluated using the Properties Plus procedures. The next section covers how to create a new procedure.


What is a procedure?

An alternative way of performing calculations during the simulation resolution

External code is called by ACM Variables and Parameters as input External code is executed Results (output variables) returned to ACM

External code is in a Dynamic Link Library DLL is the file extension in Windows


What is required to create new procedures?

A compiler: Digital Visual Fortran Visual C++

Do I need to know one of these languages to create new procedures? Yes!


Do you require a compiler to call procedures?

No!

Modeler library Physical property and kinetics calculations utility calculations for control models

Procedures created by your colleagues To use (i.e. call) a procedure you just need the Dynamic Link

Library which contains the machine code (binary)


DLLs distribution and installation

Once it is built, ACM needs only the DLL, not the source code You can distribute the DLL without having to give out the

source code

Where to install the DLL Give full path in the PROCEDURE definition Or put the DLL in a directory that is listed in the command

PATH Or in the working directory

PresenterPresentation Notesworking directory = my simulations\problem_namethat is really for short term kludge....


Steps for creating a Procedure

Step 1 : Define the Procedure this is done in ACM

Step 2 : Write/adapt the source code use your favorite editor

Step 3 : Build the DLL library ACM generates the required commands to compile the source code to link the objects and libraries with the linker

PresenterPresentation NotesCB: What do we mean by wrapper?That's the stuff to get ACM call to a FORTRAN subroutine to work properly (ie ACM is in C++ so some pointer casting is required). Nothing worth worrying about.


Step 1 - PROCEDURE Definition

Name of the procedure

Name of the subroutine to call the path of the source file

List of input variable types

List of output variable types

Name of the DLL

Various options

PresenterPresentation NotesCB: Name of the DLL with the corresponding pathUnless the dll is in your path. Note that things are more involved in client-server???


Step 2 - Writing the Code

Arguments depend on the procedure definition use the template generation to have ACM give you the correct

list of arguments: input variables output variables additional arguments for communication with ACM

arrays: the dimensions are passed as integer af

acm

Documents

acm intro aspen custom

acm intro modeling

acm intro day

acm intro faq

custom models

yes2000 aspentech

acm intro buffer tank

custom modeling feature