The paper presents the latest development in modeling tools that support the development and usage of Fractal Enterprise Models (FEMs). It describes the new features in the latest version of FEM toolkit - a software tool that facilitates the development of FEMs, which is built using the ADOxx environment, and presents a new tool - FEM viewer. The latter is a web-based tool that allows non-modeling stakeholders to study and navigate through a package of interconnected FEMs using a simple, but powerful user interface.
As FEM is not widely used, we provide some necessary (but not all) details on what it entails. To do that, we will use an example of a hospital that performs surgical operations. The FEMs for this case are presented in Fig. 1 and Fig. 2, where Fig. 1 gives a general picture, and Fig. 2 adds some details to it.
FEM was developed as a modeling language and notation for representing operational activities of an organization. The basic version of FEM has two main concepts - process and asset - and two types of relations between them – used in and managed by.
A used in relation between a process and asset means that the asset plays a certain role in the process. The relation is visually represented by an arrow with a solid line that goes from the asset to the process (see Fig. 1, 2). A managed by relation between an asset and a process means that the process changes the set, i.e. adds or removes elements or changes their properties (see Fig. 1,2). The relation is visually represented by an arrow with a dashed line that goes from the process to the asset. A label is added to the relation to identify which role the asset plays in the process, or how the process changes the asset. The set of labels is standardized; more exactly, there are 8 labels that can be added to a used in relation and three labels that can be added to a managed by relation, namely Acquire, Maintain and Retire.
Assets and processes can be presented on different levels of granularity. If needed, a connection between different granularity levels can be expressed with the help of decomposition. As an asset is a set of things, its decomposition means splitting the set into subsets. For the process, which represents behavior, decomposition can be of two types: (1) specialization – elements of the decomposition represent concrete versions of generic behavior; (2) part of – elements of decomposition represent parts of the decomposed behavior. Visually, decomposition is presented via placing elements of the decomposition inside the border of the decomposed element. In this case, the shape of the decomposed element changes to a rectangle with rounded corners; the border also changes from a solid line to a dashed or dotted line. The dashed border is used for decomposing assets or decomposing processes – behavior – using a specialization decomposition. The dotted border is used for decomposing a process using a part of decomposition. An example of asset decomposition and part of decomposition for a process are presented in Fig. 2.
At the root of Fig. 1 is a primary process called Improving health through surgery. To run it repeatedly, the process needs some assets, three types of which are presented in the model: Patients, Staff, and Hospital. The roles they play in the process are on the respective arrow; they are chosen from a fixed set. As we can see, an asset can play more than one role in the process. For example, Patients play two roles, Beneficiary, and Stock. Stock is a special role that means that each time an instance of behavior is activated, one or several elements from the Stock asset are consumed by the behavior. The latter means that the asset needs constant filling. This is not true for other types of assets, e. g. Workforce, as it is not being depleted each time an instance of the behavior is activated.
To constantly fill the asset Patients, the model in Fig. 1 has an additional managerial process (behavior) called Admission that adds new elements to the asset. Other processes also need managerial processes to have the assets in “working order”, but we will not go into this level of detail in the paper.
FEM toolkit allows to connect models in a package using a special Ghost feature. A ghost represents the second, third, etc., occurrence of a shape that has already been introduced in some model of the package. The ghost has a thick arrow in the top right corner of the shape, clicking on which will move the focus to the ghost’s original. The ghost feature was used to connect the model in Fig. 2 to the model of Fig. 1.
The model in Fig. 2 decomposes the primary process into two subprocesses: Surgery and Rehabilitation. There is also a decomposition of the asset Staff to identify categories of employees that are needed for a more detailed model: Doctors and Lab staff. The subprocesses Surgery and Rehabilitation are connected via the asset Patients after operation. Surgery functions as an Acquire process for the asset Patient after operation, adding a new element to it each time a surgery operation is completed. The asset Patients after operation is also connected by a used-in relation to Rehabilitation. The arrow that represents the relation has two labels: (1) Stock, which means that the element of the asset disappears after rehabilitation has been completed, and (2) Partner, which means that this asset actively participates in the process, engaging in care activities while not being part of the hospital organization.
Both Surgery and Rehabilitation use Doctors as Workforce, which means that a Doctor is an active agent that belong to the organization. Rehabilitation uses one more asset Patient assessments, that contains data on the health state of the rehabilitating patients. The label is marked as Tech & Info Infrastructure, which means that it is used as information for decision making, and Stock – the information is consumed by the process and new information needs to be added. To add elements to Patient assessments, an additional process - Assessing patient’s state (analysis) – is added. It uses Lab staff as Workforce, and it monitors the state of the patient (special relation described in Section 3.1); patients are (relatively) passive during the assessment.
Readers who are interested to know more about FEM, and why it has the word fractal in its name
are referred to paper [
In this section, we describe the most important features that have been introduced in the new versions
of FEM toolkit based on our experience of using the toolkit in practice and in the educational context.
These features are not presented in the articles devoted to FEM and FEM toolkit, such as [
Two new relations were introduced in a new version of FEM and FEM toolkit: Inspects/Monitors and
Association, see Fig. 3. The first relation connects a process with any other elements of the model, i.e.
process, asset, pool, or external actor. The process, in this case, exhibits an observing behavior, i.e.
gathering information from the observed with no or minimum intervention with it. Visually, this
relation is represented by an arrow with a dash-dotted blue line; it also has a small rectangular tail.
The tail points to the observing process, the arrow points to what it observes. A label in the form of
free text can be added to specify the nature of the relation. This relation is used in the model of Fig.
2. The rationale for adding this relation has been to be able to represent activities of system 3* of the
viable system model (VSM) [
Monitors Process +
Monitoring/ inspection process +
Investigates Asset +
Tacit asset
+ Resides within Construction workers +
Asset 1
+ Intersect Asset 2 +
The relation Association does not have any definite meaning; it is used whenever there is a need to express something that is not possible to express with other relations. Visually, it is represented by an arrow with a blue dotted line. This relation can be symmetrical – an arrowhead on both ends, or asymmetrical – an arrowhead only on one end. The meaning of the relation is explained using a free text label. The rationale for adding this relation has been to enable representing additional connections between the model elements that are not covered by specific relations.
As can be seen in Fig. 1 and 2, some of the process and asset shapes have icons that visualize the content of the asset, or the meaning of the process. For example, assets Patients and Doctors have icons that show that the assets consist of people. FEM toolkit includes a set of icons that can be chosen based on the needs. The choice is done when setting attributes of a shape, as shown in Figure 4, which illustrates the icon that has been chosen for the asset Patients assessments in Fig. 2. The rationale for adding icons has been to make it easier to grasp the model without reading text labels inside the shapes.
Text is the standard way of showing the label on a managed by relation. To make the model more visual, and less cluttered, the new version of the FEM toolkit makes it possible to show a symbol instead of a textual label for Acquire, Retire or Maintain. This is done by a special attribute of the relation, as shown in Figure 5. Choosing Symbol will substitute the label Acquire to the “+” sign, as shown in Fig. 2 for the relation between the process Assessing patients state (analysis) and the asset Patients assessments. For the Retire relation, the symbol would be “–“, and for Maintain, it would be “~”. Note that if there is more than one label on a relation, choosing “Symbol” would not have any effect. The rationale for the symbolic representation of some labels on the relations has been to make the models less cluttered with text labels.
In the new FEM toolkit, it is possible to define subclasses of FEM concepts that will be shown in a model via border coloring. This is illustrated in Fig. 1, where one of the pools has a yellow border color, and the other – a green border color. The meaning of this coloring is defined in a special sheet - a model of the type Border Subclassing. An example of such a model is presented in Fig. 6, left side. This sheet is then used for defining the border color of FEM components, as shown in Fig. 6, right side. The latter shows how the pool Healthy population in Fig. 1 gets the green border.
The new subclassing is, in addition to the already existing way of defining subclasses by assigning
a background color. The border subclassing is independent from the background subclassing. The
latter means that an individual concept can be classified from two independent points of view. The
need for subclassing is discussed in [
In the latest version of FEM toolkit, the model can be developed in one of two modes: Basic and Advanced. By default, a new FEM model is created in a basic mode. The latter includes only two concepts – process and asset – and two types of relations – used in and managed by. An advanced mode includes two additional concepts – external pool and external agent – and three additional relations – Drawing/Adding, Inspect/Monitors and Association. It is easy to switch from one to another. Switching from the basic to advanced mode results in new elements being added to the modeling panel window. Switching from the advanced mode back to basic will remove these elements. If no advanced elements exist in the model, the model remains the same. If the advanced elements have been added to the model, they are greyed and become non-accessible (until switching back to the advanced mode). The rationale for adding this feature has been to make it easy for a newcomer to learn FEM.
The latest version of FEM introduces some constraints that ensure the semantic correctness of the
model under construction. An example of such constraints is presented in Fig. 7. If an asset is used in
one process as Workforce, it cannot be used as Tech & Info infrastructure in another or the same
process. If a modeler tries to do this, he will get an error as shown in Fig. 7. Most of the constraints
are related to a combination of labels on the same relation or on relations related to the same modeling
element, see full list of such constraints in [
Labels on the used in and managed by relations are standardized, and they have English names. In Version 1.1 of toolkit it is possible to supply a translation file, where some, or all labels are changed, e.g. presented in another natural language. A translation file can be uploaded and activated or deactivated. Activation/deactivation allows the developers to work with the standard English labels but show the models to the stakeholders with the translated labels on the relations.
FEM toolkit is a tool aimed at modelers. It is not always easy to install FEM toolkit, especially on nonWindows machines, and using it requires some skills, even just for browsing through the set of already created models. Thus, giving this tool to the non-modeling stakeholders does not work for everyone.
An alternative method of providing stakeholders with a package of models as printed PDF sheets has its own drawbacks. While you can easily investigate an individual model, the possibility of going from one model to another using the ghost feature is lost. FEM viewer is designed to solve the problem of providing non-modeling stakeholders with a convenient method of studying the models on their own.
FEM viewer is a web-based tool that gives a possibility for its users to browse through a set of interconnected FEMs in an easy manner. The package of models can be exported from the FEM toolkit, and imported in the FEM viewer. After that, a user or group of users can be given access to the package. The user can easily browse through the set of models following the links connecting the models. An example of a package and the way a user can browse through it is presented in Fig. 8.
FEM viewer has three different kinds of users: administrator, developer, and viewer. An administrator can create new users and give them credentials. A developer can upload new models and create new users of the viewer type, and give them access to the uploaded models. Viewers can only view the model that they have been assigned to.
More information on FEM and FEM toolkit is available in [
There are several recorded materials about FEM/FEM toolkit on the website [
There are no immediate future plans for the FEM toolkit. Regarding FEM viewer, there are plans to extend its functionality, so that the user can provide feedback when browsing through the package of models. The feedback could be brought back to the FEM toolkit.
The work of the first and second authors was partly supported by the Estonian Research Council (grant PRG1226). The authors are grateful to Siim Langel - a former bachelor student of University of Tartu - who created the first version of FEM viewer (described in this paper) as part of his BS thesis. The authors are also grateful to ADOxx team that provided us with the tool that made it possible to create the FEM toolkit, and especially to Wilfrid Utz (OMiLAB NPO) who supports it. We are thankful to Dominik Bork, who created the first version of the FEM toolkit, and supported our efforts to further develop it. The authors are also grateful to the anonymous reviewers whose comments have helped to improve the text.