Empirical Studies on the Development of Executable Business Processes

Perhaps one of the reasons BPM research concentrates on analytical modeling of business processes is that BPMN is standardized fully in this regard and modeling tools support the notation very well. In this book, we focus instead on empirical research in executable process models. This requires a complete and precise specification of process models, which graduate from “PowerPoint slide” into an executable artifact running inside a workflow engine in the Cloud. In this chapter, we introduce fundamental background concepts defining executable business processes, discussing empirical research methods suitable for business process management, and presenting different architectural options for process execution and close with a brief history leading toward executable BPMN.

Empirical research is becoming increasingly important for understanding the practical uses of and problems with business processes technology in the field. However, no standardization on how to report observations and findings exists. This sometimes leads to research outcomes which report partial or incomplete data and makes published results of replicated studies on different data sets hard to compare. In order to help the research community improve reporting on business process models and collections and their characteristics, this chapter defines a modular template with the aim of reports’ standardization, which could also facilitate the creation of shared business process repositories to foster further empirical research in the future. The template has been positively evaluated by representatives from both BPM research and industry. The survey feedback has been incorporated in the template. We have applied the template to describe a real-world executable WS-BPEL process collection, measured from a static and dynamic perspective.

The implementation of business processes has been neglected for many years in research. It seemed to be that only hard coding was the appropriate solution for business process implementations. As a consequence in classical literature about business process management (BPM), the focus was mainly on the management aspects of BPM, less on aspects regarding an effective and efficient implementation methodology. This has changed significantly since the advent of BPMN 2.0 (Business Process Model and Notation) in early 2011. BPMN is a graphical notation for modeling business processes in an easy to understand manner. Because the BPMN standard had the process execution in mind when it was designed, it allows for a new way of implementing business processes, on which the process-driven approach (PDA) is based. This approach has been applied in a huge project at SAP SE since 2015 comprising more than 200 business-critical processes. In order to get an impression about the power of the process-driven approach for really complex business process implementation scenarios, this chapter explains the basics about the process-driven approach and shares experiences made during the execution of the project.

Service orchestrations are frequently used to assemble software components along business processes. Despite much research and empirical studies into the use of control-flow structures of these specialized languages, like BPEL and BPMN2, no empirical evaluation of data-flow structures and languages, like XPath, XSLT, and XQuery, has been made yet. This paper presents a case study on the use of data transformation languages in industry projects in different companies and across different domains, thereby showing that data flow is an important and complex property of such orchestrations. The results also show that proprietary extensions are used frequently and that the design favors the use of modules, which allows for reusing and testing code. This case study is a starting point for further research into the data-flow dimension of service orchestrations and gives insights into practical problems that future standards and theories can rely on.

The Business Process Model and Notation (BPMN) has become the de facto standard for process modeling. Currently, BPMN models can be (a) analyzed or simulated using specialized tools, (b) executed using business process management systems (BPMSs), or (c) used for requirements elicitation. Although there are many studies comparing BPMN to other modeling techniques for analyzing and executing processes, there are few showing the suitability of BPMN models as a source for requirements comprehension in projects where process-aware software is built without using BPMSs. This chapter presents a study aimed at comparing the comprehension of software requirements regarding a business process using either BPMN or traditional techniques, such as use cases. In our study, we analyzed responses of 120 undergraduate and graduate students regarding the requirements comprehension achieved when using only BPMN models, only use cases, or both. The results do not show significant impact of the artifacts on the comprehension level. However, when the understanding of the requirement involves sequence of activities, using the BPMN shows better results on the comprehension time.

This chapter describes the development of an information system to control the execution of high-tech manufacturing processes from the business process level, based on executable process models. The development is described from process analysis to requirements elicitation to the definition of executable business process, for three pilot cases in our recent HORSE project. The HORSE project aims to develop technologies for smart factories, making end-to-end high-tech manufacturing processes, in which robots and humans collaborate, more flexible, more efficient, and more effective to produce small batches of customized products. This is done through the use of Internet of Things (IoT), Industry 4.0, collaborative robot technology, dynamic manufacturing process management, and flexible task allocation between robots and humans. The result is a manufacturing process management system (MPMS) that orchestrates the manufacturing process across work cells and production lines and operates based on executable business process models defined in BPMN.

Hospitals are facing high pressure to be profitable with decreasing funds in a stressed healthcare sector. This situation calls for methods to enable process management and intelligent methods in their daily work. However, traditional process intelligence systems work with logs of execution data that is generated by workflow engines controlling the execution of a process. But the nature of the treatment processes requires the doctors to work with a high freedom of action, rendering workflow engines unusable in this context. In this chapter, we describe a process intelligence approach to develop a platform for clinical pathways for hospitals without using workflow engines. Our approach is explained using a case in liver transplantation, but is generalizable on other clinical pathways as well.

Workflow management systems (WfMSs) are broadly used in enterprise to design, deploy, execute, monitor, and analyze automated business processes. Current state-of-the-art WfMSs evolved into platforms delivering complex service-oriented applications that need to satisfy enterprise-grade performance requirements. With the ever growing number of WfMSs that are available in the market, companies are called to choose which product is optimal for their requirements and business models. Factors that WfMS vendors use to differentiate their products are mainly related to functionality and integration with other systems and frameworks. They usually do not differentiate their systems in terms of performance in handling the workload they are subject to or in terms of hardware resource consumption. Recent trend saw WfMSs deployed on environments where performance in handling the workload really matters, because they are subject to handling millions of workflow instances per day, as does the efficiency in terms of resource consumption, e.g., if they are deployed in the Cloud. Benchmarking is an established practice to compare alternative products, which helps to drive the continuous improvement of technology by setting a clear target in measuring and assessing its performance. In particular for WfMSs, there is not yet a standard accepted benchmark, even if standard workflow modeling and execution languages such as BPMN 2.0 have recently appeared. In this chapter, we present the challenges of establishing the first standard benchmark for assessing and comparing the performance of WfMSs in a way that is compliant to the main requirements of a benchmark: portability, scalability, simplicity, vendor neutrality, repeatability, efficiency, representativeness, relevance, accessibility, and affordability. A possible solution is also discussed, together with a use case of micro-benchmarking of open-source production WfMSs. The use case demonstrates the relevance of benchmarking the performance of WfMSs by showing relevant differences in terms of performance and resource consumption among the benchmarked WfMSs.

Executable business processes contain complex business rules, control flow, and data transformations, which makes designing good tests difficult and, in current practice, requires extensive expert knowledge. In order to reduce the time and errors in manual test design, we investigated using automatic combinatorial test design (CTD) instead. CTD is a test selection method that aims at covering all interactions of a few input parameters. For this investigation, we integrated CTD algorithms with an existing framework that combines equivalence class partitioning with automatic BPELUnit test generation. Based on several industrial cases, we evaluated the effectiveness and efficiency of test suites selected via CTD algorithms against those selected by experts and random tests. The experiments show that CTD tests are not more efficient than tests designed by experts, but that they are a sufficiently effective automatic alternative.