**3. Processes specification: Workflow and Petri nets**

A business process can be defined as the set of activities and performances to be carried out in an integrated manner to achieve a more general organizational goal. These processes are usually performed within an organizational structure in which there are various functional roles and hierarchical relationships. A process can be developed entirely in a single business unit or can be applied to various and even to different organizations (inter-organizational processes) as the processes with customers and suppliers, or collaborative processes in which knowledge is shared between different organizations.

Assessment of Operational Experience as Strategy

**3.1 Petri nets** 

for Knowledge Acquisition and Learning in Organizations 135

In the literature on information systems have been proposed various techniques to model the process perspective. Some of these techniques are not formal in the sense that the diagrams used have not a defined formal semantics. This kind of techniques (Data Flow Diagram or Structured Analysis and Design Technique) but allow to design and discuss about processes, are not the most appropriate when it is trying to design complex workflow

However, Petri nets for its formal semantics are an effective tool for modeling and analyzing business processes of organizations, and its main features include: 1) provide a natural form of representation and a high expressiveness, 2) its graphical language allows to specify complex workflows according to the workflow primitives of Workflow Management Coalition1 (WfMC), 3) the theory of Petri nets provides a powerful analysis tool to verify the correctness of the definition of workflow processes and 4) Petri nets can represent the states of a process as elements of "first class", in contrast to other modeling techniques that focus

In this chapter the Petri nets are used to formally specify the model of knowledge acquisition and organizational learning that is presented, which has interest as case of study, because it allows to contrast some of the theories and ideas developed by the different

A Petri net is a directed graph that uses two different types of nodes, called places and transitions. The places are represented by circles () and the transitions as rectangles (). The nodes are connected to each other through directed arcs ( ) and are not allowed

The formal definitions that constitute the basic theoretical framework of this technique and

The arcs are considered of weight one because in the context of workflow processes the

Definition 2. (Input place). A place p is called an input place of a transition t iff there exists a

Definition 3. (Output place). The place p is called an output place of transition t iff there

1 Founded in May 1993 is a global organization of adopters, developers, consultants, analysts, as well as university and research groups engaged in workflow and Business Process Management (BPM). The WfMC creates and contributes to process related standards and interoperability of workflow

that we have considered convenient to collect are the following (Aalst, 1998):

processes, because they are incomplete and subjective (Aalst, 2002).

exclusively on the active parts of the process, on the activities.

authors to which we have referred above.

connections between two nodes of the same type.

Definition 1. (Petri net). A Petri net is a triple (P, T, F), where:


places are achieved when certain conditions are fulfilled.



**3.2 Formalization of Petri nets** 


directed arc from p to t.

exists a directed arc from t to p.

Be the place p and the transition t, then:

management systems. http://www.wfmc.org.

Business processes can involve formal or relatively informal interactions among participants. Some of the activities of the process can be automated, in which case it is the system the responsible for carry them out. Manual activities however, are outside the control of the system being made directly by people (WFMC-TC00-1011 Issue 3.0, 1999). The business processes of organizations, generally involve the development of complex workflows that are part of their daily activity.

A workflow refers to a process or work procedure (susceptible to be automated using informatics methods and systems) in which involves different knowledge (relational data, documents, experience) and the tasks performed by the participants according to a defined set of rules, to contribute to the achievement of a business objective (Hollingsworth, 1995).

The design of workflows requires develop specifications that describe an abstraction of the processes of the company. The workflow specification languages are used for this purpose to create workflow models that support the structure of the process activities (control flow) and the exchange of information (data flow) between such activities (Mentzas et al. 2001).

The classic model of representation allows to specify the workflow based on a set of activities (with a total or partial order) and its dependencies, and enables reference the objects are handled (resources, knowledge) and the actors involved in the process (people, according to their role, and the system that can automate some activities).

Fig. 1. Workflow model based on activities

This technique is suitable to model processes from a logical point of view, but is insufficient when you need to have a detailed specification when you want to automate such processes. The use of formal techniques that offer a greater capacity for analysis and verification of the process that is being designed help to overcome this limitation (Solana, 2006).

In general, systems are increasingly oriented towards the idea of process, which is leading to the development of collaborative environments that have ever greater capabilities to manage the knowledge of organizations.

### **3.1 Petri nets**

134 New Research on Knowledge Management Models and Methods

Business processes can involve formal or relatively informal interactions among participants. Some of the activities of the process can be automated, in which case it is the system the responsible for carry them out. Manual activities however, are outside the control of the system being made directly by people (WFMC-TC00-1011 Issue 3.0, 1999). The business processes of organizations, generally involve the development of complex

A workflow refers to a process or work procedure (susceptible to be automated using informatics methods and systems) in which involves different knowledge (relational data, documents, experience) and the tasks performed by the participants according to a defined set of rules, to contribute to the achievement of a business objective

The design of workflows requires develop specifications that describe an abstraction of the processes of the company. The workflow specification languages are used for this purpose to create workflow models that support the structure of the process activities (control flow) and the exchange of information (data flow) between such activities

The classic model of representation allows to specify the workflow based on a set of activities (with a total or partial order) and its dependencies, and enables reference the objects are handled (resources, knowledge) and the actors involved in the process (people,

This technique is suitable to model processes from a logical point of view, but is insufficient when you need to have a detailed specification when you want to automate such processes. The use of formal techniques that offer a greater capacity for analysis and verification of the

Ai Aj Manual activity

A5 A6 A7 A0

In general, systems are increasingly oriented towards the idea of process, which is leading to the development of collaborative environments that have ever greater capabilities to

process that is being designed help to overcome this limitation (Solana, 2006).

A4

A3

A2

Automation activity

according to their role, and the system that can automate some activities).

workflows that are part of their daily activity.

Fig. 1. Workflow model based on activities

A1

manage the knowledge of organizations.

(Hollingsworth, 1995).

(Mentzas et al. 2001).

In the literature on information systems have been proposed various techniques to model the process perspective. Some of these techniques are not formal in the sense that the diagrams used have not a defined formal semantics. This kind of techniques (Data Flow Diagram or Structured Analysis and Design Technique) but allow to design and discuss about processes, are not the most appropriate when it is trying to design complex workflow processes, because they are incomplete and subjective (Aalst, 2002).

However, Petri nets for its formal semantics are an effective tool for modeling and analyzing business processes of organizations, and its main features include: 1) provide a natural form of representation and a high expressiveness, 2) its graphical language allows to specify complex workflows according to the workflow primitives of Workflow Management Coalition1 (WfMC), 3) the theory of Petri nets provides a powerful analysis tool to verify the correctness of the definition of workflow processes and 4) Petri nets can represent the states of a process as elements of "first class", in contrast to other modeling techniques that focus exclusively on the active parts of the process, on the activities.

In this chapter the Petri nets are used to formally specify the model of knowledge acquisition and organizational learning that is presented, which has interest as case of study, because it allows to contrast some of the theories and ideas developed by the different authors to which we have referred above.

#### **3.2 Formalization of Petri nets**

A Petri net is a directed graph that uses two different types of nodes, called places and transitions. The places are represented by circles () and the transitions as rectangles (). The nodes are connected to each other through directed arcs ( ) and are not allowed connections between two nodes of the same type.

The formal definitions that constitute the basic theoretical framework of this technique and that we have considered convenient to collect are the following (Aalst, 1998):

Definition 1. (Petri net). A Petri net is a triple (P, T, F), where:



The arcs are considered of weight one because in the context of workflow processes the places are achieved when certain conditions are fulfilled.

Definition 2. (Input place). A place p is called an input place of a transition t iff there exists a directed arc from p to t.

Definition 3. (Output place). The place p is called an output place of transition t iff there exists a directed arc from t to p.

Be the place p and the transition t, then:


<sup>1</sup> Founded in May 1993 is a global organization of adopters, developers, consultants, analysts, as well as university and research groups engaged in workflow and Business Process Management (BPM). The WfMC creates and contributes to process related standards and interoperability of workflow management systems. http://www.wfmc.org.

Assessment of Operational Experience as Strategy

Implicit OR-split

Explicit OR-split

Source: Elaborated by the authors from Aalst (1998)

**4.1 Background and organizational context** 

basis for other industries.

Fig. 2. Symbolic notation of constructors for workflow nets

**4. Organizational knowledge and learning model** 

for Knowledge Acquisition and Learning in Organizations 137

**Constructor Graphic notation Constructor Graphic notation** 

Implicit OR-join

Explicit OR-join

The next paragraph contains the empirical development of research, which has designed a model of knowledge acquisition based on analysis and evaluation of operational experience and has developed an organizational learning system that has been brought to practice successfully in Nuclenor, company that manages the operation of the nuclear power plant of Santa María de Garoña in Spain. Following are commented the background and organizational context in which the work has been developed, are characterized the organizational roles involved in the model and are described the activities of the process of analysis of operational experience and acquisition of knowledge, to finalize presenting

Currently the development of policies of continuous improvement in the organizations can be achieved through the implementation of organizational learning processes and the evaluation of operational experience, according to the phases of detection, communication, evaluation and, correction and improvement of findings, events or incidents reported. Following is presented a methodology for the acquisition of knowledge and learning based on experience, this methodology is been used to improve the efficiency in Nuclear Power Plants, which for the strict and rigorous control that they are subjected, can provide the

The Nuclear Security Council in Spain carried out periodically a review of the security of Nuclear Power Plants, through the evaluation of the operational experience required to the holders of energy production plants. It must demonstrate that the power plants maintain a

learning model developed using workflow nets as a technique of specification.

AND-split AND-join


Definition 4. (State of a Petri net). The state M of a Petri net is a distribution of tokens () over places. We will denote n1p1 + n2p2 + n3p3 + … + nkpk, the state of a Petri net such that there are n1 tokens in the place p1, n2 tokens in p2, n3 tokens in p3 and so on nk tokens in the place pk. The number of tokens can change during the execution.

Transitions are the active components in a Petri net, change the state of the net according to the following firing rules:


Given a Petri net (P, T, F) and a state M1, we establish the following notations:


#### **3.3 Workflow nets**

The Petri nets that are applied to modeling of workflow processes are called workflow nets (WF-net), are based in network theory proposed by Carl Adam Petri at the beginning of sixties (Petri, 1962) and whose basic theoretical principles have just presented.

The workflow nets allow to define the workflow processes, modeling using graphical elements the process activities (transitions), the states (places) that can be found as a result of the development of such activities and the dynamics of the process. In a workflow net the transitions are the active parts of process, the places are the passive parts and the arcs between transitions and places represent relations of causality (Aalst, 2002).

The workflow nets let model different cases or instances of the process, as layers in the WFnet which are represented by tokens () on the network places. The overlapping of the different cases allow to view the execution status of the process and perform an analysis of its situation in a certain moment (eg. when a place has a large number of tokens, indicates that there is a bottleneck in that place).

The workflow nets can represent the routing or dynamic of transitions between the process places, for which are used routing constructors. The constructors AND-split, AND-join, ORsplit and OR-join, allow to represent sequential models, conditional, parallel and iterative. It is also possible to represent under what conditions are triggered the execution of the activities, for which are used attributes or control variables.

The AND-split constructor allows to model that multiple activities are executed simultaneously. The AND-join constructor models the synchronization of two or more activities whose execution can be developed in parallel, so that the next activity does not start until they are completed. The OR-split constructor (it is a exclusive OR) models the case that for a given activity, as a result of compliance with certain conditions, begins a single subsequent activity. The OR-join constructor models the case in which a certain activity begins after the completion of one of the alternatives precedents activities, not being required the synchronization of these (WfMC-TC00-1011 Issue 3.0, 1999).

The difference between explicit OR and implicit OR is that in the latter case, the time of the election, which is normally done in terms of fulfillment of some condition, is made as late as possible.

Definition 4. (State of a Petri net). The state M of a Petri net is a distribution of tokens () over places. We will denote n1p1 + n2p2 + n3p3 + … + nkpk, the state of a Petri net such that there are n1 tokens in the place p1, n2 tokens in p2, n3 tokens in p3 and so on nk tokens in the

Transitions are the active components in a Petri net, change the state of the net according to

1. A transition t is said to be enabled iff each input place p of t contains at least one token. 2. An enabled transition may fire. If a transition t fires, then t consumes one token from

M2: transition t is enabled in state M1 and firing t in M1 results in state M2.


The Petri nets that are applied to modeling of workflow processes are called workflow nets (WF-net), are based in network theory proposed by Carl Adam Petri at the beginning of

The workflow nets allow to define the workflow processes, modeling using graphical elements the process activities (transitions), the states (places) that can be found as a result of the development of such activities and the dynamics of the process. In a workflow net the transitions are the active parts of process, the places are the passive parts and the arcs

The workflow nets let model different cases or instances of the process, as layers in the WFnet which are represented by tokens () on the network places. The overlapping of the different cases allow to view the execution status of the process and perform an analysis of its situation in a certain moment (eg. when a place has a large number of tokens, indicates

The workflow nets can represent the routing or dynamic of transitions between the process places, for which are used routing constructors. The constructors AND-split, AND-join, ORsplit and OR-join, allow to represent sequential models, conditional, parallel and iterative. It is also possible to represent under what conditions are triggered the execution of the

The AND-split constructor allows to model that multiple activities are executed simultaneously. The AND-join constructor models the synchronization of two or more activities whose execution can be developed in parallel, so that the next activity does not start until they are completed. The OR-split constructor (it is a exclusive OR) models the case that for a given activity, as a result of compliance with certain conditions, begins a single subsequent activity. The OR-join constructor models the case in which a certain activity begins after the completion of one of the alternatives precedents activities, not being

The difference between explicit OR and implicit OR is that in the latter case, the time of the election, which is normally done in terms of fulfillment of some condition, is made as late as

each input place p of t and produces one token in each output place p of t. Given a Petri net (P, T, F) and a state M1, we establish the following notations:

sixties (Petri, 1962) and whose basic theoretical principles have just presented.

between transitions and places represent relations of causality (Aalst, 2002).



the following firing rules:

t1 M2 t2 … tn-1 Mn.

that there is a bottleneck in that place).

activities, for which are used attributes or control variables.

required the synchronization of these (WfMC-TC00-1011 Issue 3.0, 1999).


possible.

**3.3 Workflow nets** 

place pk. The number of tokens can change during the execution.


Source: Elaborated by the authors from Aalst (1998)

Fig. 2. Symbolic notation of constructors for workflow nets
