Development of an Industrial
Boiler Virtual-Lab for Control
Education Using Modelica
CARLA MARTIN-VILLALBA, ALFONSO URQUIA, SEBASTIAN DORMIDO
Departamento de Inform ´ atica y Autom ´ atica, UNED, Juan del Rosal 16, 28040 Madrid, Spain
Received 19 October 2009; accepted 14 March 2010
ABSTRACT: The use of the VirtualLabBuilder Modelica library and the Dymola modeling environment facilitates
the implementation of virtual-labs with elaborated user interfaces, and based on large and complex Modelica
models. This implementation methodology is applied to develop an industrial boiler virtual-lab for process control
education. VirtualLabBuilder is freely available at www.euclides.dia.uned.es. © 2010 Wiley Periodicals, Inc. Comput
Appl Eng Educ; Published online in Wiley InterScience (www.interscience.wiley.com); DOI 10.1002/cae.20449
Keywords: software for engineering education; virtual laboratory; Modelica; interactive simulation; object-
oriented modeling
INTRODUCTION
Virtual-labs that support interactive simulations are effective ped-
agogical resources in engineering education [1,2]. The aim of a
virtual-lab is to explore simulations of a mathematical model and
to do it through visualization aids such as graphics and anima-
tions in a user-friendly way. Designing and performing simulation
experiments make virtual-lab users active players in the learning
process, which increases motivation to further learning.
Virtual-labs are typically composed of three interrelated
parts: (1) the simulation of the mathematical model that describes
the relevant properties of the system under study; (2) the interactive
user-to-model interface; and (3) a narrative providing information
about the system under study and the virtual-lab use.
The user-to-model interface, referred to as the “virtual-lab
view,” is intended to provide a visual representation of the model
dynamic behavior and to facilitate the user’s interactive actions
on the model. The model behavior can be represented in dif-
ferent ways, for example, plotting the model variables against
each other and by means of animated schematic diagrams of the
system. User’s actions on the model can be performed by interact-
ing with different elements of the view, such as buttons, sliders,
check-boxes, and certain graphic elements of the model schematic
diagram. The graphical properties of the view elements are linked
to the model variables, producing a bidirectional flow of informa-
tion between the view and the model. Changes in a model variable
Contract grant sponsor: Spanish CICYT; Contract grant numbers:
DPI2001-1012, DPI2004-01804, DPI-2007-61068.
Correspondence to C. Martin-Villalba (carla@dia.uned.es).
© 2010 Wiley Periodicals, Inc.
value are automatically displayed by the view. Reciprocally, any
user interaction with the view automatically modifies the value of
the corresponding model variable.
Virtual-labs which emulate a system’s real-time behavior
support runtime interactivity, that is, these virtual-labs allow the
user to perform interactive actions on the model at any time dur-
ing the simulation run. Changes in input variables, parameters,
or state variables will trigger an immediate response which will
show the user how these changes affect the model dynamic. An
arbitrary number of actions can be made on the model during a
given simulation run.
Domain-specific software packages for virtual-lab imple-
mentation in certain engineering fields have been developed. There
also exist general-purpose software tools specifically intended for
virtual-lab implementation. These tools provide their own pro-
cedure to define the narrative, the model, and the view of the
virtual-lab, and they automatically generate the virtual-lab exe-
cutable code. Another approach proposed in Ref. [3] consists
in using only the object-oriented modeling language Modelica
and the Dymola modeling environment. To this end, the Virtual-
LabBuilder Modelica library was developed. The advantages and
disadvantages of these approaches are discussed in the next section.
In this article, we discuss the application of the methodol-
ogy proposed in Ref. [3] to the implementation of a virtual-lab
for control education. This virtual-lab is intended to illustrate the
dynamic behavior of an industrial boiler, which operates under
two different control strategies: manual and decentralized PID.
Other virtual-labs for chemical process control, which have been
developed using this methodology, are described in Ref. [3].
The structure of this article is as follows. Firstly, different
approaches to virtual-lab implementation are discussed. Then, the
1