A Mixed Reality application for studying the improvement of HVAC systems in learning factories

(1)

ScienceDirect

Available online at www.sciencedirect.com

Procedia Manufacturing 45 (2020) 373–378

This is an open access article under the CC BY-NC-ND license (https://creativecommons.org/licenses/by-nc-nd/4.0/) Peer-review under responsibility of the scientific committee of the 10th Conference on Learning Factories 2020. 10.1016/j.promfg.2020.04.039

10.1016/j.promfg.2020.04.039 2351-9789

This is an open access article under the CC BY-NC-ND license (https://creativecommons.org/licenses/by-nc-nd/4.0/) Peer-review under responsibility of the scientific committee of the 10th Conference on Learning Factories 2020.

ScienceDirect

www.elsevier.com/locate/procedia

This is an open access article under the CC BY-NC-ND license https://creativecommons.org/licenses/by-nc-nd/4.0/)

Peer-review statement: Peer-review under responsibility of the scientific committee of the 10th Conference on Learning Factories 2020

10th Conference on Learning Factories, CLF2020

A Mixed Reality application for studying the improvement of

HVAC systems in learning factories

Marvin Czarski

a,

_{*, Yen Ting Ng}

b

_{, Marcus Vogt}

a

_{, Max Juraschek}

a

_{, Bastian Thiede}

a

_{, Puay}

Siew Tan

b

_{, Sebastian Thiede}

a

_{, Christoph Herrmann}

a

a _{Technische Universität Braunschweig, Institute of Machine Tools and Production Technology, Sustainable Manufacturing & Life Cycle}

Engineering, Langer Kamp 19 b, Braunschweig, 38106, Germany

b _{Singapore Institute of Manufacturing Technology, 2 Fusionopolis Way,#08-04 Innovis, Singapore 138634}

Abstract

Heating, ventilation and air conditioning (HVAC) systems in factories provide controlled conditions for workers and production equipment. At the same time, these systems are responsible for a significant share of industrial energy consumption. Commonly, HVAC systems are treated separately from production systems. However, numerous interactions and cross-influences occur affecting the overall energy efficiency and air quality. With analyzing and understanding these indoor air conditions the goal is to enable future engineers and experts to design and set them up in a way that improves human comfort, while reducing energy consumption. To achieve this, a cyber-physical system approach in a learning factory is presented. Based on data provided by the learning factory infrastructure, a building performance simulation with an integrated computational fluid dynamics simulation is composed. With the implementation in the learning factory, different ventilation and operation scenarios can be examined in learning scenarios and trainings to convey competencies about cyber-physical production systems in general and influences on the connection to HVAC systems. A mixed reality application provides three-dimensional visualization of the cyber-model and computed results for the learners.

Keywords: HVAC; Learning Factory; Die Lernfabrik; Manufacturing Control Tower; HoloLens; BPS; CFD;

* Corresponding author. Tel.: +49 531 391-7658; fax: +49 531 391-5842.

E-mail address: m.czarski@tu-braunschweig.de

Available online at www.sciencedirect.com

ScienceDirect

www.elsevier.com/locate/procedia

10th Conference on Learning Factories, CLF2020

A Mixed Reality application for studying the improvement of

HVAC systems in learning factories

Marvin Czarski

a,

_{*, Yen Ting Ng}

b

_{, Marcus Vogt}

a

_{, Max Juraschek}

a

_{, Bastian Thiede}

a

_{, Puay}

Siew Tan

b

_{, Sebastian Thiede}

a

_{, Christoph Herrmann}

a

a _{Technische Universität Braunschweig, Institute of Machine Tools and Production Technology, Sustainable Manufacturing & Life Cycle}

Engineering, Langer Kamp 19 b, Braunschweig, 38106, Germany

b _{Singapore Institute of Manufacturing Technology, 2 Fusionopolis Way,#08-04 Innovis, Singapore 138634}

Abstract

Heating, ventilation and air conditioning (HVAC) systems in factories provide controlled conditions for workers and production equipment. At the same time, these systems are responsible for a significant share of industrial energy consumption. Commonly, HVAC systems are treated separately from production systems. However, numerous interactions and cross-influences occur affecting the overall energy efficiency and air quality. With analyzing and understanding these indoor air conditions the goal is to enable future engineers and experts to design and set them up in a way that improves human comfort, while reducing energy consumption. To achieve this, a cyber-physical system approach in a learning factory is presented. Based on data provided by the learning factory infrastructure, a building performance simulation with an integrated computational fluid dynamics simulation is composed. With the implementation in the learning factory, different ventilation and operation scenarios can be examined in learning scenarios and trainings to convey competencies about cyber-physical production systems in general and influences on the connection to HVAC systems. A mixed reality application provides three-dimensional visualization of the cyber-model and computed results for the learners.

Keywords: HVAC; Learning Factory; Die Lernfabrik; Manufacturing Control Tower; HoloLens; BPS; CFD;

* Corresponding author. Tel.: +49 531 391-7658; fax: +49 531 391-5842.

(2)

1. Introduction

Industrial value creation provides products for public consumption and is the backbone of numerous economic systems. At the same time, the energy demand of the industrial sector is responsible for negative environmental impacts connected to the extraction and combustion of fossil fuels. In Germany for instance, 30 % of the total electricity consumption is connected to industrial activities [1]. A significant share of approximately 20 % of this energy demand is caused by heating, ventilation and air conditioning (HVAC) [2]. This illustrates the importance and the leverage of HVAC systems towards raising the overall energy efficiency of production systems and factories. Learning factories can play an important role in utilizing this efficiency potential by conducting research, education and training. A learning factory provides a physical, factory-like environment that can effectively foster the learning success by lowering the transition barrier between theoretical knowledge and practical application [3]. Furthermore, it can be a suitable environment for research, development and testing of technical systems within a realistic yet safe environment. The implementation of learning modules on HVAC systems in learning factories is still scarce. For this purpose, a concept has been developed for integrating learning content on HVAC systems in learning factories based on a cyber-physical systems approach with a mixed reality interface. The developed approach was implemented in two learning factories in Singapore and Germany and tested for feasibility. Students can learn about HVAC operation and control, the implementation of cyber-physical systems and mixed reality visualizations of the underlying models as well as the effects of input and output parameters.

2. State of the Art

2.1. HVAC Systems

The actual performance of HVAC systems depends on their design, the environmental conditions and implemented operational behavior. An inefficiently operating system can lead to increased energy demand of a factory system while providing inadequate performance, observable for instance in insufficient air quality or temperature [4]. To improve the performance of HVAC systems in industrial applications, constant adjustment to the actual conditions within the production system are necessary [5]. Decreasing costs for sensors and computational power have led to an increased penetration of information and communication technology (ICT) in industrial environments [6]. For the case of HVAC systems, this allows a shift from simple traditional control loops towards integrated cyber-physical systems.

From a simplified perspective, HVAC systems consist of the combination of fans, heating and cooling registers, air filters, sensors, air distribution system and, if necessary, additional components such as heat recovery and humidification or dehumidification devices. Modelling and simulation is commonly used in dimensioning and planning of HVAC systems. Integrated planning approaches considering the dynamic behavior of the production system and outdoor conditions can result in more efficient system designs [7]. In order to simulate a HVAC system, a combination of two modeling approaches can be used. Building performance simulations (BPS) describe the physical building system based on the material and functional combination of its constituting elements, while computational fluid dynamics (CFD) focus on the modelling of air flows and its parameters in detail [8]. BPS is generally used to provide the boundary conditions for CFD simulations. Successful implementation of this combined approach can be found for ventilation opening location planning [9] or cooling of an airport terminal [10].

2.2. Implementing cyber-physical production systems in learning factories

In cyber-physical production systems (CPPS), the physical world of production equipment, materials and products are combined with a cyber-world containing digital models, data processing and simulation. Both worlds are connected through data acquisition from the physical to the cyber world and feedback or control actions in the opposite direction. A human-centered approach recognizes and utilizes the unique abilities of human beings such as creativity, flexibility and experience [11]. A framework for the design and analysis CPPS is illustrated in Figure 1. Cyber-physical systems are a viable approach to enable adaption to changing influencing conditions and for the case of HVAC systems to control the delivery of heat or cold air adequately [12]. However, an energy efficient operation of HVAC systems enabled by CPPS is rarely taught in learning factory environments [13]. A generic approach for

(3)

analysis and implementation of appropriate production processes and systems based on a simplified rating system presented by Thiede et al., can be used to identify suitable implementation scenarios for CPPS in these teaching and learning spaces [11]. This approach can support the transformation and coherence of industrial scale processes and systems and their respective skills required for efficient operation and development between industry and learning factories.

Fig. 1. Structure for analysis and development of CPPS in learning factories [11].

2.3. Mixed Reality

As CPPS connect the analog, physical world of the production system with the digital, cyber world of data processing and simulation integration of human stakeholders is required through appropriate visualization and interaction techniques. An approach to meet these requirements is the concept of mixed reality (MR). In MR, digital objects and real objects are combined in one user experience to create a mixture of both perspectives. Real objects are characterized by having an actual objective existence such as machines, products and tools. Contrary, virtual objects only exist in essence or effect as information or concept and not formally or in material shape [14]. MR can be a powerful tool for supporting teaching and learning processes [15], for instance by improving the efficiency of tasks and adopting to the user’s specific experience level. An overview of fields of application of MR in learning factories by Juraschek et al. shows the potential support in imparting knowledge and skills [16]. The targeted combination and mixing of digital content with the real world application environments in learning factories can support the understanding of processes, data, methods and systems. Specific learning scenarios, that would not be feasible in a fully physical environment due to high cost, risk or complexity, can be experienced by different users. Thus, the spatial, temporal and functional scope of a physical learning factory can be extended by utilizing capabilities and elements of MR. One of the main challenges for productive implementation of MR applications in learning factories as well as in industrial practice is the process for selecting suitable concepts and hardware and software environments. 3. Conceptual implementation of HVAC systems

The concept development for HVAC systems in learning factories is set around the goals of creating transparency for learners to foster understanding of energy demand of production systems, enabling interaction with the systems and providing a safe environment for testing new approaches. Generally, the environmental conditions in a given space are influenced by three factors: the environment, the ventilation system and internal heat loads in the room. The CFD domain is composed of the physical setup, the surrounding walls, and the properties of the inside air. The cyber world is an abstraction of the real world based on the model in form of the combined CFD and BPS simulation as shown in Figure 2. Each cell of the CFD domain can be defined to be solid or gaseous. For the solid cells, like the physical equipment, different materials can be assigned to properly represent heat absorption, storage and release. The CFD domain is connected to the BPS by flow ports and surface adapters. Flow ports are used to add external and

(4)

internal airflow sources and sinks, like machines and equipment. The induced chilled or warmed air mixes with the existing air in the room. To accurately model the behavior of the FCU a controller and a temperature probe is added to the CFD domain. Environmental influences, for instance the local weather, are connected to the CFD by surface adapters, which transfer a heat flux through the surrounding shell of the room using a diffusion model. On the inside, in the CFD domain, the heat gain is transferred through natural convection and radiation. To model the contribution of occupants inside the room a combination of flow ports and surface adapters is required to cover both, radiation and respiration.

Fig. 2. Abstraction of BPS and CFD model for HVAC system modeling.

4. Exemplary Implementation in the Model Factory@SIMTech

4.1. Model Factory@SIMTech

The concept is exemplary implemented at the Model Factory@SIMTech, a learning factory setup for the Manufacturing Control Tower (MCT) in Singapore. It is composed of real-scale machines connect in a process chain for a functional consumer electronics demo product. The main purpose is the training of industry-relevant skill in the fields of digitization in production engineering, CPPS and flexible automation across the shop floor, enterprise level and the supply chain. With a real-scale production system, the factory has the same technical building equipment, including HVAC, as it would be found in a real-world producing enterprise.

The specific HVAC system is composed of a fan coil unit (FCU) that receives chilled water and pre-cooled air from a central system and supplies cold air. Inside the learning factory, various influencing factors for the indoor climate can be recognized: a) the building shell with heat diffusion through the surrounding walls, b) local (outside) climatic conditions, c) value-creating and non-value-creating production equipment (machines, IT, autonomous guided vehicles) with a specific location inside the learning factory and production state (heat emissions), and d) present workers imparting radiation heat and respiration. Control parameters of the HVAC systems for influencing the indoor air conditions are a set temperature goal and the fan speed of the fan coil unit (FCU).

4.2. Technical Implementation

To apply the outlined concept to the Model Factory@SIMTech, an extensive network of sensors to monitor the air quality parameters of temperature and humidity is installed. The sensors gather data from different spatial locations inside the production environment, providing a detailed distribution map of the production environment. Moreover, airflow sensors inside the FCU and diffusers extend the gathered data with the cool air supply. For heat emission estimation, the present power metering system for the machines is utilized. The data is acquired using Node-RED [17] and persisted in a relational database for the subsequent modeling process. As modeling language, widely used

Modelica is implemented to describe the environment of the Model Factory@SIMTech. The identified key

influencing factors (a-d) described in section 4.1 are integrated using components provided by the commercial Human

Production Activity Cooling Unit Room (CFD) Occupants Environment Surface adapter Flow adapter

Controller Settings & system _parameters

Internal and external parameters Surface adapter Flow adapter

(5)

internal airflow sources and sinks, like machines and equipment. The induced chilled or warmed air mixes with the existing air in the room. To accurately model the behavior of the FCU a controller and a temperature probe is added to the CFD domain. Environmental influences, for instance the local weather, are connected to the CFD by surface adapters, which transfer a heat flux through the surrounding shell of the room using a diffusion model. On the inside, in the CFD domain, the heat gain is transferred through natural convection and radiation. To model the contribution of occupants inside the room a combination of flow ports and surface adapters is required to cover both, radiation and respiration.

Fig. 2. Abstraction of BPS and CFD model for HVAC system modeling.

4. Exemplary Implementation in the Model Factory@SIMTech

4.1. Model Factory@SIMTech

The concept is exemplary implemented at the Model Factory@SIMTech, a learning factory setup for the Manufacturing Control Tower (MCT) in Singapore. It is composed of real-scale machines connect in a process chain for a functional consumer electronics demo product. The main purpose is the training of industry-relevant skill in the fields of digitization in production engineering, CPPS and flexible automation across the shop floor, enterprise level and the supply chain. With a real-scale production system, the factory has the same technical building equipment, including HVAC, as it would be found in a real-world producing enterprise.

The specific HVAC system is composed of a fan coil unit (FCU) that receives chilled water and pre-cooled air from a central system and supplies cold air. Inside the learning factory, various influencing factors for the indoor climate can be recognized: a) the building shell with heat diffusion through the surrounding walls, b) local (outside) climatic conditions, c) value-creating and non-value-creating production equipment (machines, IT, autonomous guided vehicles) with a specific location inside the learning factory and production state (heat emissions), and d) present workers imparting radiation heat and respiration. Control parameters of the HVAC systems for influencing the indoor air conditions are a set temperature goal and the fan speed of the fan coil unit (FCU).

4.2. Technical Implementation

To apply the outlined concept to the Model Factory@SIMTech, an extensive network of sensors to monitor the air quality parameters of temperature and humidity is installed. The sensors gather data from different spatial locations inside the production environment, providing a detailed distribution map of the production environment. Moreover, airflow sensors inside the FCU and diffusers extend the gathered data with the cool air supply. For heat emission estimation, the present power metering system for the machines is utilized. The data is acquired using Node-RED [17] and persisted in a relational database for the subsequent modeling process. As modeling language, widely used

Modelica is implemented to describe the environment of the Model Factory@SIMTech. The identified key

influencing factors (a-d) described in section 4.1 are integrated using components provided by the commercial Human

Production Activity Cooling Unit Room (CFD) Occupants Environment Surface adapter Flow adapter

Controller Settings & system _parameters

Internal and external parameters Surface adapter Flow adapter

Comfort Library (HCL) from XRG Simulation GmbH. The HCL offers CFD calculations inside the Modelica model,

providing timely and spatial information about the air conditions [18]. The domain for the CFD simulation is discretized into Cartesian cubes with a side length of 70 cm resulting in 780 total cells in the simulation. While the room is built with its three inside walls that are assumed to be adiabatic and one outside door and wall through which heat flow occurs, the induced heat by workers is estimated through HCL taking their activity and the clothing level into consideration. For the other internal source, production machines, the majority of heat is emitted via forced convection through integrated fan systems. The amount of airflow and heat released is related to the status of the machine.

As previously introduced in 2.2, one element of CPPS is the recommendation for action, which can be derived from the cyber part. For this purpose, the simulation is capable to process different scenarios, e.g. process flows or the number of workers on-site. This enables new insights to be gained into the behavior of HVAC and the starting point for an iterative improvement process. To visualize the results, an MR application is developed. With the Microsoft HoloLens [19], a powerful device for augmenting the simulated environmental conditions directly in the real environment is available. As development environment Unity [20] is used, providing the required functions of editing three-dimensional assets as well as instantiating and manipulating these during application runtime. The user can interact with the visualization through a natural user interface using gestures. Since the simulation provides several properties with a special resolution, different representations are provided for the user. For instance, an application of in different ventilation scenarios provides a visualization of areas with over- or undersupply of cold air. In the case of the Model Factory@SIMTech, direct control of the HVAC is yet not possible. However, the CPPS modeling already provides the foundation for further implementation of automatically adopting HVAC settings to improved settings.

4.3. Didactical Integration

With the introduction of the proposed approach into learning factories, learners can work on technical systems and get experience in application of the CPPS paradigm in a close-to-reality environment. With respect to the HVAC system, a valuable addition to the existing learning setup has been created, as each step can be reproduced and followed through by the learners. The developed workflow reduces the effort for installation of sensors, long-term data gathering and auxiliary steps. This leads to a more time efficient learning. The outline of the established learning setup in connection with existing theoretical modules is given in Table 1.

5. Conclusion and Outlook

This paper presents the application of the cyber-physical production system paradigm for indoor air conditioning in a learning factory environment for the practical extension of the existing theoretical curricula. The exemplary implementation of such an CPPS for the is visualized in Figure 3.

HVAC systems are often hidden from view yet responsible for a significant share of energy demand of a production system. With the developed MR application, a detailed insight into the system and its behavior is enabled. Learners can experience the implementation steps to acquire and process data to modeling the cyber counterpart of the physical

Table 1: Outlined learning objectives of the HVAC-CPPS.

Task Addressed topics Learning outcome

Analyzing operation of HVAC

system System engineering, measurement Understanding different modes of operation of the system and the interaction between occupancy, production activity and air conditioning Processing 3D data using mixed

reality Data treatment & analytics, programming Development and processing of spatial data describing the ambient air and thermal human comfort Deriving hot spots for improvement

(6)

environment. Simultaneously, the constructed modelling and simulation environment based on the installed real equipment provides a suitable learning case on CPPS implementation in a learning factory environment.

Fig. 3. Implemented CPPS framework at the Model Factory@SIMTech.

Next steps for this work is lie in enhancing the user experience by integrating a direct connection of the simulation with the physical system beyond visualization of the effects of different parameters, e.g. by changing the physical control variables or trigger new simulation runs. Future work should also consider usability studies to integrate the feedback of the users and learners.

Acknowledgements

We gratefully acknowledge the financial support for this project by BMWi (German Federal Ministry of Economics and Energy) under reference 03ET1660A and the support for the German-Singapore cooperation in the project ISURF-Hub by BMBF (German Federal Ministry of Education and Research) with reference FKZ 01DP17049.

References

[1] Arbeitsgemeinschaft Energiebilanzen. Auswertungstabellen zur Energiebilanz für die Bundesrepublik Deutschland 1990 bis 2018. [2] Kemmler, A., Straßburg, S., Seefeldt, F., Anders, N. et al., 2016. Datenbasis zur Bewertung von Energieeffizienzmaßnahmen in der

Zeitreihe 2005-2014.

[3] Abele, E., Metternich, J., Tisch, M., Chryssolouris, G. et al., 2015. Learning Factories for Research, Education, and Training 32, p. 1. [4] Guo, W., Zhou, M. Technologies toward thermal comfort-based and energy-efficient HVAC systems: A review, in 2009 IEEE

International Conference on Systems, Man and Cybernetics, IEEE, 3883.

[5] Hesselbach, J., 2012. Energie-und klimaeffiziente Produktion: Grundlagen, Leitlinien und Praxisbeispiele. Springer-Verlag.

[6] Colombo, A.W., Karnouskos, S., Kaynak, O., Shi, Y. et al., 2017. Industrial cyberphysical systems: A backbone of the fourth industrial revolution 11, 6.

[7] Thiede, S., 2012. Energy Efficiency in Manufacturing Systems. Springer Berlin Heidelberg, Berlin, Heidelberg. [8] Zhai, Z.J., Chen, Q.Y., 2005. Performance of coupled building energy and CFD simulations 37, 333.

[9] Fan, Y., Ito, K., 2012. Energy consumption analysis intended for real office space with energy recovery ventilator by integrating BES and CFD approaches 52, 57.

[10] Gowreesunker, B.L., Tassou, S.A., Kolokotroni, M., 2013. Coupled TRNSYS-CFD simulations evaluating the performance of PCM plate heat exchangers in an airport terminal building displacement conditioning system 65, 132.

[11] Thiede, S., Juraschek, M., Herrmann, C., 2016. Implementing Cyber-physical Production Systems in Learning Factories 54, 7.

[12] Gunes, V., Peter, S., Givargis, T. Improving energy efficiency and thermal comfort of smart buildings with HVAC systems in the presence of sensor faults, in 2015 IEEE 17th International Conference on High Performance Computing and Communications, 2015 IEEE 7th International Symposium on Cyberspace Safety and Security, and 2015 IEEE 12th International Conference on Embedded Software and Systems, IEEE, 945.

[13] Büth, L., Blume, S., Posselt, G., Herrmann, C., 2018. Training concept for and with digitalization in learning factories: An energy efficiency training case 23, 171.

[14] Milgram, P., Kishino, F., 1994. A taxonomy of mixed reality visual displays 77, 1321.

[15] Bacca, J., Baldiris, S., Fabregat, R., Kinshuk et al., 2015. Mobile Augmented Reality in Vocational Education and Training 75, 49. [16] Juraschek, M., Büth, L., Posselt, G., Herrmann, C., 2018. Mixed Reality in Learning Factories 23, 153.

[17] Node-RED. Node-RED - Low-code programming for event-driven applications. https://nodered.org/.

[18] XRG Simulation GmbH. FLUIDDYNAMICS LIBRARY. https://www.xrg-simulation.de/en/products/xrg-library/fluiddynamics. [19] Microsoft. HoloLens (1st gen) hardware. https://docs.microsoft.com/en-us/hololens/hololens1-hardware.