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