SIM-CIP: concept of a spatial information model for complex industrial plants

1 Introduction and motivation

The provision of tailored information at the appropriate time and place is becoming increasingly important in relation to human-centricity in a modern industrial environment [1]. In this context, this means an adaptive adjustment of the technological solutions used to the individual needs of a worker [2]. The provision of targeted location-based information for humans is part of such a need.

Complex large-scale industrial plants extend over an area of several thousand square meters, often distributed over different levels. The information occurring in such plants is manifold and numerous. These include comprehensive assembly manuals and operating instructions, detailed plans and schemes, real-time process data and warning signals, safety areas and documenting control reports. Also, the increasing complexity of processes such as increased automation or advancing system integration contributes to this. This amount and variety of information is therefore not efficiently accessible to humans without advanced technological support. For example, a maintenance technician responding to alarm signals or error messages faces the challenge of efficiently and accurately obtaining relevant information from extensive documentation and database systems in this complex environment to solve a task. This applies to both complex, multi-layered operations and routine activities. Without further improvements to the interface technology used, it will become increasingly difficult for the technician to obtain necessary information or to consider relevant information. When it comes to minimizing costs, direct access to useful information is one of the most important factors.

In addition to task-specific information filtering, a filter based on local position is a desirable approach. For example, a technician in the field can be provided with necessary plans or data for supplier parts filtered directly when in close proximity to a component or other relevant maintenance and servicing locations. This allows immediate position-related access to information without the technician having to perform a time-consuming manual selection. This type of information provision can also be beneficial for mobile plant components, such as mobile robots. Furthermore, informative safety mechanisms can be established for safety-critical areas to protect a technician or the integrity of a plant. Other user groups can also benefit from location-based information filtering. For example, such a filter can already be used when providing plans for the construction of a plant in order to support construction workers or a site manager with the distribution of information. Also, regarding state-of-the-art assistance systems using extended reality (XR), a location-based filter is an attractive way to coordinate these applications.

For an appropriate implementation, an underlying model should fulfill various requirements. In this article, four general categories were identified for this purpose: Location, Data Schema, Context and Transmission. Additionally, two specific categories were identified: Localization Adaptability and Hybrid Information Map. Thus, the requirements for a model within these categories are as follows:

1. Location (Req-A): Provision of information based on the position of a user

2. Data Schema (Req-B): Structure that enables information to be represented, supplemented, integrated and updated

3. Context (Req-C): Information is available to a user based on relevance to their role or task and is adaptive and personalizable according to their individual needs

4. Transmission (Req-D): Information is provided to assist a user in solving a task

5. Localization Adaptability (Req-E): Adaptable, flexible and scalable localization strategy that is aligned with the various conditions and structures of a complex industrial plant

6. Hybrid Information Map (Req-F): Information within a plant can be mapped either dependent or independent of components

Considering these requirements, this paper presents a concept of a spatial information model in the context of complex industrial plants (Spatial Information Model for Complex Industrial Plants, SIM-CIP) with selected presentations of conceptual modeling approaches for context-aware Location-based Services.

2 Related work for context-aware location-based services

Location-based Services (LBSs) are applications that use a user’s location to provide digital services based on this location [3]. These applications are typically used on portable devices such as tablets, cell phones and XR headsets. The fields of application range from marketing or entertainment to navigation or inventory management [4]. In the manufacturing industry, LBSs are increasingly being integrated into Industry 4.0 concepts and technologies, such as the Internet of Things (IoT). By combining location data with information on e.g. assets, material and personnel within a production line, a service can be provided which increases efficiency and reduces production costs. One example of this is the optimization of search times and picking routes within a warehouse [5]. LBSs can also be used to increase worker safety [6]. For example, alarms can be triggered when a worker enters safety-relevant areas. LBSs rely on spatial information models that represent the physical environment through digital geometric and semantic data structures. Therefore, with increasing digitalization, LBSs and their underlying spatial information models have developed into an emerging field of research. Research works addressing corresponding architectures describe various layer models. While some proposeintegrating location information as part of a comprehensive context layer [7], others introduce multi-layered approaches specifically designed for the processing and semantic interpretation of location information [8]. Within such architectures and for the design of spatial information models for LBSs, aspects of positioning, mapping and context awareness are of central importance. These will be covered in the following. Data storage in databases for LBSs respective information models also represents an important aspect and therefore a wide research field. However, since the focus of this paper is on data linkage rather than data storage, this will not be examined in detail.

3 Concept of a spatial information model illustrated by the example of a heat pump plant

This paper proposes a concept of a spatial information model for the introduction of context-aware LBSs in complex industrial plants. The conceptual model is intended to describe a structure for linking spatial information with other attributive information in order to extend information by the real factor of location where it occurs or is required. The following section first describes a use case of a complex heat pump plant for which context-aware LBSs are intended to be utilized. Subsequently, the components of the model are described with regard to the use case.

3.1 Servicing and maintenance procedures on a heat pump plant

In the use case of a complex heat pump plant considered in this paper, the technical documentation consists of several thousand pages. The overall documentation is separated into various partial documentations, whereby a major part is dedicated to documentations for supplier parts, which themselves also contain several sections. For maintenance or servicing tasks, however, only specific parts are relevant, according to the task to be executed. In the case of an alarm signal from the plant, for example in the form of an error code, a plant supervisor has to search through the detailed technical documentation and, if necessary, consult with other plant experts to identify potential causes of the error. These are then forwarded to a technician who searches for the root cause of the failure in the plant area. The technician then needs to find the necessary documents within the technical documentation in order to resolve the problem according to the instructions.

In this context, the power plant identification system (Kraftwerk Kennzeichensystem, KKS) [19], [20] takes a crucial role (Figure 1). Within the standardized system, all components are uniquely marked. This serves as an overall reference system for the entire plant documentation. In addition, it provides a central organizational structure for the technical plant documentation. All trades and suppliers are contractually bound to structure their documentation accordingly. This means, for example, that construction drawings, maintenance manuals and supplier documents are consistently identified and referenced via the KKS. Therefore, internal drawing numbers or document identifiers of the suppliers are linked to the corresponding KKS identification code via cross-reference tables. Despite this beneficial aspect of the KKS, it is still challenging for a technician to navigate through the extensive documentation to obtain necessary information.

Figure 1:

Example of a KKS-oriented tagging number of a pressure transmitter on a P&ID diagram within a system train for the heat pump plant covered here.

In order to improve the efficiency of the process chain described above, a technician in the field needs support in accessing relevant information quickly and in a targeted manner. Context-aware LBSs, i.e. location-based information provision with useful filter options, can provide support here. The organizational structure of the KKS already provides a spatial distribution of technical information that can be used for context-aware LBSs, as illustrated in Figure 2. With regard to the use case described, for example, maintenance documents can be filtered and provided directly based on the position and role of the technician at a particular maintenance location. A technician should therefore receive the information relevant to him without time-consuming manual selection, receive additional safety-relevant warnings and, if appropriate, be assisted by technology such as XR.

Figure 2:

Spatial distribution of technical information as a result of the use of the KKS organizational structure shown by the example of a complex heat pump plant (Image source: MAN ES SE).

3.2 Description of the spatial information model

The concept of the spatial model contains the components shown in Figure 3 in order to realize context-aware LBSs for the heat pump plant discussed here. It includes a database layer, a location layer, a context management layer and a user interaction layer. It therefore describes the path from an information unit to the user (in the context of this paper, an information unit is a single, enclosed piece of information that represents a fundamental modular component of higher-level information structures, such as knowledge graphs). The specific characteristics of the individual components with corresponding selected presentations of conceptual modeling approaches for context-aware LBSs are explained in the following.

Figure 3:

Concept of the spatial information model separated into basic layers.

3.2.1 Database layer

The database layer contains all information required for various tasks within the plant (Req-B). In the case of the heat pump plant discussed in this paper, this includes, for example, installation manuals, operating instructions, plans, real-time process data, warning signals, safety areas and control reports. The information contained within the database layer is divided into information units, which can vary in their extent. They can range from comprehensive PDF documents to individual chapters, text sections or graphics or even to single values. These information units are organized, for example, using metadata, tags, semantic descriptions or spatial coordinates. In the concept described here, the database layer delivers requested information units to higher-level layers.

3.2.2 Location layer – positioning

The positioning component of the location layer determines the position of a user device within the plant by evaluating the sensor data described in the following (Req-A). For the heat pump plant considered here, various conditions have to be taken into account when determining the position. The size of the plant plays a significant role, as does its illumination and the reflections from metallic surfaces that occur in it. Interferences with other devices are also possible. Since it is a static plant, changes in the plant environment are barely to expect. Additionally, there are areas of varying relevance within the plant. For example, information is increasingly required at the compressor and a precise embedding of information can be useful, whereas in intermediate passages a zone-based provision of information is normally sufficient. The following levels of localization (Req-E) are therefore considered in this use case:

1. Level I – Proximity to radio signal transmitter (single beacons): Zone positioning with rough accuracy (several meters). Primarily for warning and safety instructions or status messages when signals are received, as well as for mobile information provision (mobile beacons)

2. Level II – Lateration of radio signals (multiple beacons): Point positioning with medium accuracy (few meters to centimeters). Provides the basis for location-based information provision in order to designate for example document provision zones and provides the coordination basis for the CV used in Level III

3. Level III – CV with SLAM (camera and IMU usage): Point positioning with high accuracy (few centimeters to millimeters). For precise spatial embedding of information via XR visualizations

Within this use case, the application of Level II is primarily intended. However, the principles from Level I can optionally be used, or additional fine positioning can be achieved using Level III positioning. The resulting location data is then forwarded to the context management layer for a coordinated provision of information.

3.2.3 Location layer – mapping

Within the mapping component of the location layer, a 3D map of the heat pump plant is created using CAD model data combined with references to information units from the database layer (Req-A & Req-B). For this use case, an approach based on grid coordinates is used for the map. The coordinate system of the map and the positioning component are identical so that the user can be correctly positioned on the resulting map. Inside the coordinate system, descriptive attributes (e.g. metadata, tags) are attached to reference corresponding information units.

The deposit respectively referencing of the information units, which is the main aspect of the mapping component of the location layer, is based on two principles (Req-F) in the context of this use case. The first principle is a model-centered approach. Therefore, the BIM modeling process with a high LoD is used. Component CAD data is positioned in a defined coordinate system and supplemented with descriptive attributes that serve as a reference to information units. The most important attribute here is the KKS number. The second principle is a zone-centered approach. Freely defined zones are supplemented with attributes that can be placed independently of model data. This allows information units to be referenced independently of a component, enabling, for example, a zone-related designation of safety information. It is therefore assumed that a reference to an information unit is most likely required in a defined sector. With this principle, zones can be created according to various criteria. One could be based on physical conditions, resulting from structural factors such as room boundaries through walls or different levels. Further area divisions are possible in accordance with functions, processes or safety aspects. However, a clear local identification is a prerequisite for this. In the context of this use case, the information referencing is mainly considered using the model-centered approach. Since the technical documentation of the plant is structured according to the KKS, a model-centered approach is therefore particularly suitable. The zone-centered approach is envisaged for information that has to be stored independently of CAD model data.

The information referencing using the BIM process is illustrated in Figure 4. Therefore, CAD models of the plant are used and supplemented with attributes. The model data resulting from the process is exported to a server in Industry Foundation Classes (IFC) [21] format, which represents a standardized data exchange format within BIM processes. The server contains the BIM map with the context management layer components. The contextual attributes linked with the models are used to create requests to the database layer (e.g. JSON exchange). The attributive referencing of an “AR-Workcard” is also shown in Figure 4. This illustrates that the map can also be used to reference or coordinate applications for augmented reality. In addition, by bypassing the positioning attributes, the use as an interactive 3D map is also possible.

Figure 4:

Map creation process using the BIM process with attributes for information referencing.

3.2.4 Context management layer

Within the context management layer, various roles and other potentially useful filters (e.g. task, time, etc.) are defined to control access to information units of the heat pump plant (Req-C). This allows them to be filtered in a targeted manner and, for example, provided to a technician via a graphical user interface. For this purpose, all attributes to which information units are linked within the model scope are coordinated within this layer so that the adequate information units can be provided in the defined areas of the map. The context management layer also coordinates the position based on the map created and the location values received from the user.

Due to the variety of industrial plants, there are no universally valid roles or filters, which is why a systematic definition of these is required. One suitable approach is to first identify the necessary user groups. For the heat pump plant discussed here, this includes the plant supervisors, mechanics and electricians. Afterwards, tasks and permissions are analyzed for each of these user groups. Based on this, individual roles with associated specific permissions can be created. Here, for example, this includes “plant supervision”, “inspection mechanic” and “maintenance mechanic”. With this structured approach, it is possible to assign information and access rights systematically and precisely to user groups according to the specific roles. On the other hand, such an access system requires a high initial analysis effort, which also increases with the number of user groups and roles. Therefore, numerous interviews must be conducted with the respective user groups and associated workflows must be analyzed. In principle, such an analysis must be carried out individually for each plant due to the specific conditions, but it is possible that similar roles can be transferred to similar plants. However, a final implementation is supported by the availability of the BIM process and the existence of the KKS organizational structure, which enables a straightforward and up-to-date assignment of information.

3.2.5 User interaction layer

The concept is completed with the user interaction layer, which is designed for devices such as tablets, cell phones or XR glasses. All relevant information are made accessible to a user and specific configurations are also provided (Req-C & Req-D). Therefore, the various user roles and other filters from the context management layer are utilized. This and other information organized by the context management layer are provided for the LBS regardless of the end device. Only the client application, which includes the graphical user interface, has to be adapted to the end device. The addition of information units to the database layer via the graphical user interface respectively client application, such as test reports, is also being considered.

4 Evaluation and outlook

The proposed concept of the spatial information model with its described layers has several advantages for an industrial application, as well as in comparison to described related works discussed in the second Chapter. In general, the separation into layers enables a modular approach in which each layer serves specific and independent purposes. In contrast to other approaches where the location information is handled as part of an overall context layer in the architectural concept, this approach provides a central separated layer. Therefore, the resulting focus on a central information reference entity for location information is especially valuable for information access in LBSs. A flexible adjustment of the localization technology to the specific requirements of the plant environment is made possible by the different localization levels. This also allows for a simple scaling to different plant sizes. Other approaches in a similar field of research also describe the combined use of different technological approaches for positioning, such as the dual use of GPS for outdoor positioning and BLE for indoor positioning, but do not address the simultaneous use of positioning variants, such as zone positioning and point positioning, within a chosen technological concept. The combined use of a model-centered and a zone-centered mapping enables both a precise linking of information to specific components and a comprehensive linking of information to areas. This increases the possibilities and flexibility of information provision. Other approaches typically rely on singular mapping strategies. The integration of BIM processes in combination with the KKS structure creates a standardized basis for the organization of information and the spatial mapping of plant data. This is particularly valuable when a systematic and unique component identification is used within a process.

However, the proposed concept also has limitations. A basic prerequisite is the existence of digital technical documentation. Therefore, the model assumes a high level of digitalization. Suitable information units must be defined, a BIM process must be available respectively a central model for the deposit of plant data, as well as an organizational structure such as the KKS for the technical documentation of a plant. Only on this basis can a model-centered and zone-centered allocation of information can be carried out. This is also made more difficult as information cannot always be clearly separated to components or zones, which may lead to redundancies in a final implementation. Furthermore, it should be noted that in the case of significant plant changes, updating is also associated with a greater effort. The definition of suitable interfaces between and within the different layers of the model also poses a challenge but is essential for a seamless communication between the different layers. Also, when using different positioning levels, the coordination of the levels is an extra effort that needs to be considered.

Additionally, the concept of the spatial information model regarding the described usecase was discussed with experts from the fields of technical documentation, multimedia applications and plant engineering. Individual discussions were held with nine experts, who had been working in their respective fields for more than 10 years. All of the experts generally agreed with the concept and emphasized the importance of a holistic approach in order to provide information in a structured and targeted manner. The technical documentation experts (four people) particularly pointed out the difficulty of defining and extracting information units. Especially with regard to role definition in the context management, they see this as a challenge. In addition, they advised that the structure given by the KKS should be an important aspect when developing a system. The experts for multimedia applications (three people) underlined the importance of a user adding information units to the database layer. The plant engineering experts (two people) consider it possible to use the model implementation as early as the construction of a plant in order to continue to use respectively expand it for maintenance processes. All experts mentioned the interface problems caused by the large number of systems with different data formats within a plant. However, the majority see the concept as a basic structure for a holistic information-related plant overview.

In a future study, the presented model shall be evaluated through a prototypical realization. In addition to the clarification of technical implementation details, the provision of information between traditional technical documentation and technical documentation as context-aware LBS shall be compared in the context of maintenance and servicing activities at a research plant. Therefore, aspects such as user acceptance, as well as the influence on the information flow and the decision-making process of a user shall be examined. An increased integration of XR in maintenance and servicing activities with context-aware LBS as a central coordination platform shall also be investigated.

5 Conclusion and discussion

In this paper, a concept of a spatial information model for complex industrial plants was presented. Therefore, maintenance and servicing processes for a complex heat pump plant were considered. In this context, possible requirements and components of a model were identified and explained. The resulting model approach consists of a database layer, a location layer, a context management layer and a user interaction layer. These layers interact and are intended to form a basis for context-aware LBSs within industrial plants. Due to the focus of the paper on the conceptualization of the model, it was assumed that the required information units are already predefined for certain plant areas. However, this is a demanding process that is also highly dependent on a plant and should therefore be considered separately. In addition, the concept is based on a static map that can be extended by a dynamic position detection. In the future, dynamic mapping should be considered, to ensure an updated map. Due to the massive progress in the field of artificial intelligence, its use in the interaction between layers or within a particular layer should also be considered.