Variables

As we saw in the previous section, a model definition typically contains variable declarations. The basic syntax for a variable declaration is simply the “type” of the variable (which will be discussed shortly in the section on Built-In Types) followed by the name of the variable, e.g.,

Real x;

Variables sharing the same type can be grouped together using the following syntax:

Real x, y;

A declaration can also be followed by a description, e.g.:

Real alpha "angular acceleration";

Variability

Parameters

By default, variables declared inside a model are assumed to be continuous variables (variables whose solution is generally smooth, but which may also include discontinuities). However, as we first saw in the section titled Getting Physical, it is also possible to add the parameter qualifier in front of a variable declaration and to indicate that the variable is known a priori. You can think of a parameter as “input data” to the model that is constant with respect to time.

Constants

Closely related to the parameter qualifier is the constant qualifier. When placed in front of a variable declaration, the constant qualifier also implies that the value of the variable is known a priori and is constant with respect to time. The distinction between the two lies in the fact that a parameter value can be changed from one simulation to the next whereas the value of a constant cannot be changed once the model is compiled. The use of constant by a model developer ensures that end users are not given the option to make changes to the constant. A constant is frequently used to represent physical quantities like \pi or the Earth’s gravitational acceleration, which can be assumed constant for most engineering simulations.

Discrete Variables

Another qualifier that can be placed in front of a variable declaration is the discrete qualifier. We have not yet shown any example where the discrete qualifier would be relevant. However, it is included now for completeness since it is the last remaining variability qualifier.

Built-In Types

Many of the examples so far referenced the Real type when declaring variables. As the name suggests, Real is used to represent real valued variables (which will generally be translated into floating point representations by a Modelica compiler). However, Real is just one of the four built-in types in Modelica.

Another of the built-in types is the Integer type. This type is used to represent integer values. Integer variables have many uses including representing the size of arrays (this use case will be discussed shortly in an upcoming section on Vectors and Arrays).

The remaining built-in types are Boolean (used to represent values that can be either true or false) and String (used for representing character strings).

Each of the built-in types restricts the possible values that a variable can have. Obviously, an Integer variable cannot have the value 2.5, a Boolean or String cannot be 7 and a Real variable cannot have the value "Hello".

Derived Types

As we saw in the previous examples that introduced Physical Types, it is possible to “specialize” the built-in types. This feature is used mainly to modify the values associated with Attributes like unit. The general syntax for creating derived types is:

type NewTypeName = BaseTypeName(/* attributes to be modified */);

Frequently, the BaseTypeName will be one of the built-in types (e.g., Real). But it can also be another derived type. This means that multiple levels of specialization can be supported, e.g.,

type Temperature = Real(unit="K"); // Could be a temperature difference
type AbsoluteTemperature = Temperature(min=0); // Must be positive

Enumerations

An enumeration type is very similar to the Integer type. An enumeration is typically used to define a type that can take on only a limited set of specific values. In fact, enumerations are not strictly necessary in the language. Their values can always be represented by integers. However, the enumeration type is safer and more readable than an Integer.

There are two built-in enumeration types. The first of these is AssertionLevel and it is defined as follows:

type AssertionLevel = enumeration(warning, error);

The significance of these values will be discussed in a forthcoming section on assert.

The other built-in enumeration is StateSelect and it is defined as follows:

type StateSelect = enumeration(never, avoid, default, prefer, always);

Attributes

So far in this chapter we have mentioned attributes (e.g., unit), but we haven’t discussed them in detail. For example, which attributes are present on a given variable? This depends on the type of the variable (and which built-in and derived types it is based on). The following table introduces all the possible attributes indicating their types (i.e., what type of value can be given for that attribute), which types they can be associated with and finally a brief description of the attribute:

Attributes of Real

quantity

A textual description of what the variable represents

Default: ""

Type: String

start

The start attribute has many uses. The main purpose of the start attribute (as discussed extensively in the section on Initialization) is to provide “fallback” initial conditions for state variables (see fixed attribute for more details).

The start attribute may also be used as an initial guess if the variable has been chosen as an iteration variable.

Finally, if a parameter doesn’t have an explicit value specified, the value of the start attribute will be used as the default value for the parameter.

Default: 0.0

Type: Real

fixed

The fixed attribute changes the way the start attribute is used when the start attribute is used as an initial condition. Normally, the start attribute is considered a “fallback” initial condition and only used if there are insufficient initial conditions explicitly specified in the initial equation sections. However, if the fixed attribute is set to true, then the start attribute is treated as if it was used as an explicit initial equation (i.e., it is no longer used as a fallback, but instead treated as a strict initial condition).

Another, more obscure, use of the fixed attribute is for “computed parameters”. In rare cases where a parameter cannot be initialized explicitly, it is possible to provide a general equation for the parameter in an initial equation section. But in cases where the parameter is initialized in this way, the fixed attribute for the parameter variable must be set to false.

Default: false (except for parameter variables, where it is true by default)

Type: Boolean

min

The min attribute is used to specify the minimum allowed value for a variable. This attribute can be used by editors and compilers in various ways to inform users or developers about potentially invalid input data or solutions.

Default: -DBL_MAX where DBL_MAX is the largest floating point value that can be represented for the given platform.

Type: Real

max

The max attribute is used to specify the maximum allowed value for a variable. This attribute can be used by editors and compilers in various ways to inform users or developers about potentially invalid input data or solutions.

Default: DBL_MAX where DBL_MAX is the largest floating point value that can be represented for the given platform.

Type: Real

unit

As discussed extensively in this chapter, variables can have physical units associated with them. There are rules about how these units are expressed, but the net result is that by using the unit attribute it is possible check models to make sure that equations are physically consistent. A value of "1" indicates the value has no physical units. On the other hand, a value of "" (the default value if no value is given) indicates that the physical units are simply unspecified. The difference between "1" and "" is that the former is an explicit statement that the quantity is dimensionless (has not units) while the latter indicates that the quantity may have physical units but they are left unspecified.

Default: "" (i.e., no physical units specified)

Type: String

displayUnit

While the unit attribute describes what physical units should be associated with the value of a variable, the displayUnit expresses a preference for what units should be used when displaying the value of a variable. For example, the SI unit for pressure is Pascals. However, standard atmospheric pressure is 101,325 Pascals. When entering, displaying or plotting pressures it may be more convenient to use bars.

The displayUnit attribute doesn’t affect the value of a variable or the equations used to simulate a model. It only affects the rendering of those values by potentially transforming them into more convenient units for display.

Default: ""

Type: String

nominal

The nominal attribute is used to specify a nominal value for a variable. This nominal value is generally used in numerical calculations to perform various types of scaling used to avoid round-off or truncation error.

Default:

Type: Real

stateSelect

The stateSelect attribute is used as a hint to Modelica compilers about whether a given variable should be chosen as a state (in cases where there is a choice to be made). As discussed previously in the section on Enumerations, the possible values for this attribute are never, avoid, default, prefer and always.

Default: default

Type: StateSelect (enumeration, see Enumerations)

Attributes of Integer

quantity

A textual description of what the variable represents

Default: ""

Type: String

start

It is worth noting that an Integer variable can be chosen as a state variable or as an iteration variable. Under these circumstances, the start attribute may be used by a compiler in the same was as it is for Real variables (see previous discussion of Attributes of Real)

In the case of a parameter, the start attribute will (as usual) be used as the default value for the parameter.

Default: 0.0

Type: Integer

fixed

see previous discussion of Attributes of Real

Default: false (except for parameter variables, where it is true by default)

Type: Boolean

min

The min attribute is used to specify the minimum allowed value for a variable. This attribute can be used by editors and compilers in various ways to inform users or developers about potentially invalid input data or solutions.

Default: - \infty

Type: Integer

max

The max attribute is used to specify the maximum allowed value for a variable. This attribute can be used by editors and compilers in various ways to inform users or developers about potentially invalid input data or solutions.

Default: \infty

Type: Integer

Attributes of Boolean

quantity

A textual description of what the variable represents

Default: ""

Type: String

start

It is worth noting that an Boolean variable can be chosen as a state variable or as an iteration variable. Under these circumstances, the start attribute may be used by a compiler in the same was as it is for Real variables (see previous discussion of Attributes of Real)

In the case of a parameter, the start attribute will (as usual) be used as the default value for the parameter.

Default: 0.0

Type: Boolean

fixed

see previous discussion of Attributes of Real

Default: false (except for parameter variables, where it is true by default)

Type: Boolean

Attributes of String

quantity

A textual description of what the variable represents

Default: ""

Type: String

start

Technically, a String could be chosen as a state variable (or even an iteration variable), but in practice this never happens. So for a String variable the only practical use of the start attribute is to define the value of a parameter (that happens to have the type of String) if no explicit value for the parameter is given.

Default: ""

Type: String

It is worth noting that Derived Types retain the attributes of the built-in type that they are ultimately derived from. Also, although the type of, for example, the min attribute on a Real variable is listed having the type Real it should be pointed out explicitly that attributes cannot themselves have attributes. In other words, the start attribute doesn’t have a start attribute.

Modifications

So far, we’ve seen two types of modifications. The first is when we change the value of an attribute, e.g.,

Real x(start=10);

In this case, we are creating a variable x of type Real. But rather than leaving it “as is”, we then apply a modification to x. Specifically, we “reach inside” of x and change the start attribute value. In this example, we are only going one level into x to make our modification. But as we will see in our next example, it is possible to make modifications at arbitrary depths.

The other case where we have seen modifications was in the section on Avoiding Repetition. There we saw modification used in conjunction with extends clauses, e.g.,

extends QuiescentModelWithInheritance(gamma=0.3, delta=0.01);

Here, the modification is applied to elements that were inherited from the QuiescentModelWithInheritance model. As with modifications to attributes, the element being modified (a model in this case) is followed by parentheses and inside those parentheses we specify the modifications we wish to make.

It is worth noting that modifications can be nested arbitrarily deep. For example, imagine we wanted to modify the start attribute for the variable x inherited from the QuiescentModelWithInheritance model. In Modelica, such a modification would be made as follows:

extends QuiescentModelWithInheritance(x(start=5));

Here we first “reach inside” the QuiescentModelWithInheritance model to modify the contents that we “inherit” from it (x in this case) and then we “reach inside” x to modify the value of the start attribute.

One of the central themes of Modelica is support for reuse and avoiding the need to “copy and paste” code. Modifications are one of the essential features in Modelica that support reuse. We’ll learn about others in future sections.