6  Projects

NotePurpose of this chapter

The BPA Lab is not only an artefact — it is also a vehicle for research and teaching projects. This chapter collects projects conducted with, or around, the BPA Lab. Each entry summarises the project’s motivation, approach, and main outcomes. IT is not a complete list of all projects, but rather a curated selection that illustrates the lab’s potential as a research and teaching platform.

6.1 Overview

The following table gives a quick overview of the projects documented in this chapter. They are listed in the order in which they appear below.

Project Authors Type Year Contribution to the BPA Lab
Teaching BPA in a Learning Factory Zapp, Gonzalez & Gonzalez SoTL publication (peer-reviewed) 2025 Teaching concept and pilot evaluation of using the BPA Lab as a teaching environment for project- and problem-based learning.
System Architecture for a Modular BPA Learning Factory J. P. Iyebi Itomb Master’s thesis 2023 Modular system architecture for the BPA Lab based on Domain-Driven Design and Camunda 8
Data architecture of the BPA Lab D. Gonzalez Master’s thesis 2025 Data-virtualisation layer (Trino), IoT pipeline (Filebeat), and two Grafana dashboards for end-to-end and production-process analysis.
Integrating IoT Data into Process Mining B. Kücük Master’s thesis 2026 Integration of workflow event data with IoT data in event logs
User form design in Camunda 8 J. Will Bachelor’s thesis 2026 BPM-form usability guidelines, a consistent form layout, and redesigned or newly created user-task forms, with usability evaluation.

6.2 Teaching Business Process Automation in a Learning Factory (SoTL)

Authors: Matthias Zapp, Saphira Gonzalez & Domenic Gonzalez (TH Köln) Publication: Forschung und Innovation in der Hochschulbildung, Nr. 25, 2025, Research Paper Publisher: Cologne Open Science (COS), TH Köln DOI: 10.57684/COS-1326 Type: Scholarship of Teaching and Learning (SoTL) publication (Coleman et al. 2023)

6.2.1 Background and motivation

Business Process Automation (BPA) is a socio-technical challenge that requires a diverse set of competences — technical (process modelling in BPMN, implementing features in a BPMS, etc.), methodological (planning and running automation projects), as well as social and personal competences for collaborating with stakeholders from technical and business backgrounds (Nerdinger et al. 2008; Schaper 2012).

To teach this combination of skills, the bachelor’s programme in Business Information Systems at TH Köln uses Project-Based and Problem-Based Learning (PBL) (Barrows 1996; Chow 2021; Santos et al. 2023). Students work in teams of around five to six people on automation projects using a BPMS — an approach that creates active learning and practical problem-solving, similar to PBL formats described in the literature (Bandara et al. 2010).

However, the authors observed a persistent gap: in regular bachelor’s courses with around 100 students, project tasks must be simplified so they remain manageable. This unavoidable simplification means students often miss the experience of integrating automation into a realistic enterprise IT landscape. A telling classroom observation: on a final-presentation day, one student team was surprised by another team that demonstrated triggering a BPMS from a simulated company web portal — a feature mentioned in lectures, but absent from most project tasks because of the limits of a simplified setting.

6.2.2 Research question

The paper asks:

“To what extent can the use of a learning factory with software and hardware components improve project-oriented and problem-oriented teaching in the field of business process automation?”

The authors draw on the well-established concept of learning factories from engineering education — small, realistic production environments that bridge theoretical knowledge and industrial practice (Abele et al. 2024). While most learning factories are physically large and centred on manufacturing, the BPA Lab is deliberately small-scale, with a stronger focus on software systems and components than on hardware.

6.2.3 The teaching concept

The proposed teaching concept extends the existing PBL format. Instead of working on isolated, simplified scenarios, student teams work on project tasks set in the fictitious enterprise represented by the BPA Lab — a customer-specific bicycle manufacturer with end-to-end processes for order processing, production, and shipping. The lab’s process, decision, and data models together form a “digital twin” of this business scenario.

An illustrative example given in the paper: a project task to design and implement a shipment process for special goods, which requires integrating a software component of the BPA Lab dealing with shipment planning (using an external API for route planning) and integrating a software-hardware module — the Fischertechnik warehouse robot — to realise the connection with a warehouse management system.

Organisationally, milestone meetings with a “fictitious client” who asks critical questions and gives constructive feedback are incorporated into the existing PBL structure.

6.2.4 Preliminary evaluation

The evaluation was performed in two phases.

Phase 1 — Interviews with bachelor’s students. Students from two previous course cycles were presented with the proposed teaching concept and asked to imagine integrating the learning factory into their project. They reacted very positively, using terms like “very important”, “inspiring”, “exciting” and “very interesting”. They emphasised that the lab could make BPA “less abstract and more tangible”, deepen understanding through hands-on experience, and foster transfer of learning and motivation. They also identified several requirements and concerns: clear instructions and materials, sufficient teaching support, concerns about hardware reliability, and the importance of integrating the lab thoughtfully into existing modules to avoid overload.

Phase 2 — Pilot project with master’s students. A project simulation was carried out with eight master’s students from the Digital Sciences programme. Compared to bachelor’s students, the participants had similar BPM knowledge but stronger technical skills and software-development experience. They were asked to solve a project task and assess — from the perspective of bachelor’s students — whether the concept could realistically be implemented at that level.

Overall, students viewed the concept’s opportunities as positive: the practical experience of working with real BPM tools, the learning-by-doing approach, and collaborative work on specific challenges were all valued. At the same time, the evaluation revealed substantial challenges: complexity of installing and configuring the lab’s components, individual problems with hardware/software/network setup, and difficulties in dealing with certain BPMS concepts. Students requested both lab-specific documentation and basic background material on the underlying technologies.

6.2.5 Conclusions and consequences for teaching

The authors arrive at a nuanced conclusion: the teaching concept as evaluated is not directly applicable to the bachelor’s course at TH Köln with ~100 participants. The complexity of using the full BPA Lab — combining software development, network and hardware setup — exceeds what can reasonably be supported in a large bachelor’s course.

Two practical consequences follow:

  1. For the existing bachelor’s course: the BPA Lab can serve as a realistic project background that is demonstrated at the start of the project. The technical project task itself, however, should only be extended carefully — in particular, integration of hardware components requires more prior knowledge and support than is feasible. Pure software components of the BPA Lab are a more realistic extension.

  2. For future courses: the lab has potential as the basis of new elective course formats, which by design can cover a wider range of learning objectives — from BPM and BPA through to IoT and Cyber-Physical Systems — and allow for a higher student workload.

For BPM educators in general, the paper provides early evidence on the opportunities, challenges, and requirements of using a small-scale learning factory in a BPA course — an approach that is interesting but demanding, both for students and for teaching staff.

6.2.6 Funding

The underlying teaching project “Enhancing teaching in the field of business process automation through a learning factory” was supported by the Foundation for Innovation in Higher Education as part of the REDiEE project at TH Köln.

6.2.7 Further reading

6.3 System Architecture for a Modular BPA Learning Factory

Author: J. P. Iyebi Itomb · Supervisor: Prof. Dr. Matthias Zapp · Second supervisor: Prof. Dr. Christian Kohls Type: Master’s thesis, Digital Sciences – Software Architecture, TH Köln · Submitted: October 2023

6.3.1 Contribution to the BPA Lab

This thesis developed and implemented a modular system architecture for the BPA Lab, enabling multiple student groups to work independently on different aspects of process automation and analysis without interfering with each other. Concretely, the work contributes:

  • A Domain-Driven Design (DDD)-based decomposition of the BPA Lab scenario into seven business domains — Order Entry, Order Management, Logistics, Shipment, Production Execution, Purchasing, and Shopfloor — each modelled as an independent Bounded Context with its own BPMN process model.
  • A three-layer system architecture (documented as C4 context, container, and component diagrams) comprising a control layer for IoT/hardware control, a communication layer (MQTT broker and distribution application), and a business process layer with domain-specific Camunda 8 applications.
  • Migration from Camunda 7 to Camunda 8 Self-Managed, including a Docker Compose-based deployment and a rewrite of the Job Worker implementation to communicate with the Zeebe engine via gRPC.
  • Event-driven inter-domain communication using BPMN message events and Camunda Webhook connectors, enabling process domains to exchange control signals (e.g., triggering production planning from order management) while remaining loosely coupled.
  • Containerised Job Workers grouped by domain, reducing the setup effort for students to Docker Desktop only.

The evaluation confirms that the architecture effectively supports modular, independent development in teaching projects. Process analytics was identified as a direction for further development, with Camunda 8’s Optimize component providing REST access to process data for future process mining integration.

6.4 Designing and Implementing a Data Architecture for the BPA Lab

Author: D. Gonzalez · Supervisor: Prof. Dr. Matthias Zapp · Second supervisor: Prof. Dr. Dietlind Zühlke Type: Master’s thesis, Digital Sciences, TH Köln · Submitted: February 2025

6.4.1 Contribution to the BPA Lab

This thesis adds the data architecture to the BPA Lab that was previously missing: an architecture and running implementation that makes the heterogeneous data of the lab usable for end-to-end process analysis and monitoring. Concretely, the work contributes:

  • A data architecture based on data virtualisation using Trino as the central distributed query engine — heterogeneous sources (MySQL with order data, Elasticsearch with BPMS event data, FT-factory IoT data) are integrated virtually, allowing combined real-time and historical SQL queries without a separate physical data warehouse.
  • An IoT-data pipeline that forwards MQTT messages from the Fischertechnik factory (environmental sensors, machine status, warehouse occupancy) to Elasticsearch via Filebeat, making them queryable alongside process data.
  • Grafana dashboards integrated into the lab:
    • a tactical end-to-end dashboard with process KPIs and visualisations across the full bicycle ordering process (intended for taking process-improvement decisions),
    • an operational production dashboard based on the FT-factory’s IoT data, visualised in real time (intended for production-floor decisions).
  • A systematic requirements catalogue for data architectures in Industry 4.0 contexts, mapped to the BPA Lab’s constraints (small footprint, single host, no high-throughput or strict security requirements). This makes the rationale for design choices traceable and adaptable for follow-up projects.

6.5 Integrating IoT Data into Process Mining in an Industry 4.0 Environment

Author: B. Kücük · Supervisor: Prof. Dr. Matthias Zapp · Second supervisor: Prof. Dr. Dietlind Zühlke Type: Master’s thesis, Digital Sciences, TH Köln · Submitted: May 2026

6.5.1 Contribution to the BPA Lab

This thesis adds a proof-of-concept integration pipeline for IoT-enhanced process mining to the BPA Lab. It connects the workflow event data recorded by Camunda/Zeebe with the MQTT-based IoT observations produced by the Fischertechnik factory layer, making physical factory evidence accessible to process mining tools. Concretely, the work contributes:

  • A consolidated requirements catalogue for IoT-enhanced process mining in Industry 4.0 settings, derived from literature and the BPA Lab context, covering event log readiness, semantic interpretation, timestamp validation, case correlation, and traceability.
  • A staged hybrid integration approach that preserves a clean workflow-only event log as a baseline while selectively incorporating IoT observations — either as contextual attributes enriching existing workflow events, or as IoT-derived event rows added to the process timeline.
  • Five complementary event log artefacts: a cleaned baseline workflow event log, a structured IoT event log, an IoT-enriched workflow event log, a derived IoT event log, and a final combined event log containing both workflow and IoT-derived events with explicit source and correlation metadata.
  • A Python-based prototype pipeline (implemented in Jupyter notebooks) that prepares, maps, and integrates workflow and MQTT data; documented in the thesis appendix for reuse in follow-up projects.
  • An evaluation in Celonis comparing the workflow-only baseline with the combined event log, demonstrating that IoT data can make production-floor behavior visible and reveal mismatches between workflow execution and physical factory behavior — while also quantifying the limitations introduced by timestamp uncertainty, missing shared case identifiers, and rule-based case correlation.

The evaluation confirms that IoT-enhanced process mining is feasible within the BPA Lab setting and analytically useful, but requires careful semantic interpretation and transparent handling of correlation uncertainty. The thesis builds directly on the data architecture established in the Gonzalez 2025 thesis and provides a foundation for future live-monitoring extensions of the lab.

6.6 Human-Centred Design of Form-Based User Interfaces in Camunda 8

Author: J. Will · Supervisor: Prof. Dr. Matthias Zapp · Second supervisor: Prof. Dr. Raphaela Groten Type: Bachelor’s thesis, Business Information Systems, TH Köln · Submitted: January 2026

6.6.1 Contribution to the BPA Lab

This thesis delivered a redesigned set of user-task forms for the BPA Lab demonstration factory, replacing the previously rudimentary Camunda forms with usability-tested versions. Concretely, the work contributes:

  • A consistent layout template — TH Köln header with form title and job role, plus colour-coded message blocks (blue for guidance, red/orange/green for error/warning/confirmation) implemented as reusable HTML views.
  • BPM-specific usability guidelines distilled from Nielsen’s heuristics, Norman’s design principles, and ISO 9241-11/210 — directly applicable to future student projects using Camunda forms.
  • Redesigned or newly created forms covering the full bicycle-manufacturer scenario, including four entirely new forms (Confirm Supply, Quality Control, Missing Workpiece, Factory Failure) that close previously empty user-task slots.
  • Small technical extensions, e.g. a connector now retrieves warehouse coordinates dynamically rather than hard-coding them in the form.
  • An empirical baseline from a usability test with five participants (8 forms, 40 runs, mixed-methods: think-aloud, completion time, success rate, Likert-scale questionnaire) that future revisions can be measured against.
TipGet involved

Companies and researchers: If you are interested in collaborating with the BPA Lab — whether as an industry partner contributing real-world scenarios, or as a researcher exploring joint projects in business process automation an analytics, or related fields — we welcome your enquiry. Please contact Prof. Dr. Matthias Zapp.

Students: If you would like to conduct your bachelor’s or master’s thesis, or a semester project, in the context of the BPA Lab, please reach out as well. Typical topics include extending the lab’s software or hardware components, applying BPA methods and technologies, or applying process analytics to the lab’s data.