Investigation of the Effectiveness of an Augmented Reality and a Dynamic Simulation System Collaboration in Oil Pump Maintenance

Abstract

The maintenance of oil pumps is a complex task for any operating organization, and for an industrial enterprise in the oil and gas sector of the economy, this issue has a high degree of urgency. One of the reasons for this is a wide spread of pumping equipment in all areas of oil and gas enterprises. At the same time, an aggressive environment, uneven load, remote facilities, and harsh climatic zones (especially in the areas of the Arctic region or production platforms) are factors that make it relevant to develop special systems that help or simplify the maintenance of pumping equipment. Dynamic modeling is one of the modern technologies which allows for solving the urgent issue of assessing the technical condition of equipment. It is the basis of systems that carry out diagnostics and prognostic calculations and allow for assessing the dynamic state of objects under various conditions of their operation, among other functions. Augmented reality technology is a technology that allows for reducing the time for equipment maintenance by reducing the time for searching and processing various information required in the maintenance process. This paper presents an investigation of the effectiveness of an augmented reality and a dynamic simulation system collaboration in oil pump maintenance. Since there is insufficient research on the joint application of these two technologies, the urgent issue is to prove the effectiveness of such collaboration. For this purpose, this paper provides a description of the system structure, gives a description of the development process of the augmented reality system application and tests the application using Microsoft HoloLens 2.

1. Introduction

There are several maintenance principles for pumping equipment [1,2]. First, principles based on preventive measures (the so-called condition-based maintenance principles). When the condition of the pumping equipment is monitored, diagnostics and maintenance are performed as needed. Second, there are the schedule-based maintenance principles (principles based on statistics of typical equipment failures). Their analysis predicts what and at what time it is necessary to perform service and technological operation to maintain the pump in operating condition. Third, there are the principles based on reactive measures, when maintenance is performed on the fait accompli of equipment failure or breakdown. Another pump maintenance strategy is to keep some equipment in reserve and replace it reactively as necessary. These measures impose significant costs on the operating company. Not all plants can have the necessary equipment reserve at all process units. There are many developments with recommendations for the introduction and implementation of systems that provide technological maintenance for equipment based on preventive methods [3,4].

For example, [5] describes a methodology for planning the maintenance of pumps at existing thermal power plants. A Weibull distribution was used to predict future failures based on the current values of the pumping equipment parameters. As a result of the analysis, the optimal maintenance interval for each individual pump type in the pumping system was determined, taking into account system availability to reduce the cost of unnecessary scheduled maintenance.

The authors of [6] discussed maintenance planning for pumps at thermal power plants through reliability analysis based on a small amount of data.

An approach to developing a preventive maintenance system for rotodynamic pumps that focuses on diagnosing abnormal events related to hydrodynamic operating conditions is presented in [7]. This method is based on the experimental characteristics of the dynamic response of the pump under different loads and operating anomalies. The technique is implemented on a medium-sized centrifugal pump. The results show that a simple spectral analysis of pressure signals captured at the pump inlet or outlet can provide sufficient decision criteria to form the basis for a diagnostic system.

The authors of [8] proposed a complete productive maintenance approach based on intelligent machine learning to achieve zero downtime in industrial equipment. Maintenance includes functional checks, maintenance, and the repair and replacement of necessary parts in, for example, equipment, machinery, and buildings.

A data-driven approach to minimize the cost of pump operation and maintenance is discussed in [9]. A neural network algorithm is used to model pump performance using data collected from the process.

The authors of [10] presented the concept of intelligent forecasting using neural network algorithms and electronic maintenance of centrifugal pumps.

However, despite the rather massive number of works with methods for the preventive maintenance of pumps, practice shows that companies still face difficulties in implementing such maintenance. Based on a survey of 98 Swedish companies, it was found in [11] that maintenance personnel still mostly perform repairs when equipment fails rather than preventive maintenance. Similar results were found in the U.S. [12] and Italy [13]. In addition, [14] notes that among the companies that have created preventive maintenance implementation projects, many do not follow systematic processes. Thus, it can be pointed out that there is a gap between the challenges faced by companies that want to implement preventive maintenance and the advanced solutions presented in the literature.

Nevertheless, the work related to the implementation of the system of preventive maintenance of pumps is based on the recommendations of collection and processing of big data and the implementation of special models and calculation complexes to assess the service life and technical condition of pumping equipment [15]. All information is collected and transmitted to the users or specialists who operate this equipment for processing. The potential problems in implementing predictive analytics systems are associated with the lack of a tool that can clearly recognize a possible dangerous situation for the equipment at the right time or indicate the cause of a situation [16,17]. For example, [18] states that maintenance technicians were largely not using field maintenance data because of problems with data availability and reliability. It is said that technicians spend 12+% of their total maintenance time on each piece of equipment requesting and searching for information. However, these activities can be alleviated by providing proper computing support and implementing a tool that will provide the right information support for oil pump maintenance. The authors of the present work propose using an augmented reality system as such a tool.

Today, the world lives in the age of digitalization [19,20]. Digital technologies are being introduced everywhere and affect all sectors and areas of human activity [21,22]. According to the authors, one of the digital technologies most suitable for pump maintenance tasks is augmented reality technology. By matching real objects with virtual ones, the augmented reality system can become the main support for technicians when making decisions [23,24]. The use of an augmented reality system in equipment maintenance is not a new thing. Such systems are used, developed, and presented in various studies. For instance, Ref. [25] provides an overview of the current state of AR in maintenance and the most important technical limitations based on a systematic literature review. The results of the study show the areas where there are still shortcomings. As an example, traceability methods are indicated as the main problem in the way of implementing AR technologies in the industry. Because of their imperfection, it is difficult to compare real and virtual objects. The authors of [26] considered the application of augmented reality for creating an integrated object management system in the oil industry.

The authors suggest combining two technologies in the present paper: augmented reality [27] and dynamic simulation [28,29]. This will make it possible to evaluate the effectiveness of their collaborative use in the maintenance of pumping equipment and to form recommendations for the development, implementation, and collaboration of these technologies.

In a way, the dynamic modeling system works on the principle of a digital twin [30,31]. However, it is essential to point out that the digital twin is a rather complex concept as it comprises different functionalities. For example, a numerical simulation system [32] is used to solve local target tasks (e.g., determination of the pump shaft impeller load and blade movement). For this purpose, this system uses a pump/pump motor numerical simulation system. However, if it is necessary to extend the functionality and determine changes in the technological parameters over time (e.g., changes in the flow rate of raw materials flowing through the pipeline, temperature, and other regime parameters that characterize a particular technological process), then the numerical simulation system is not appropriate for solving this problem. In this case, dynamic modeling [33] is a more preferable tool.

To achieve this goal, the Methods and Materials section will present ways of connecting the dynamic model and the augmented reality system, and the methodology for conducting experiments and testing and evaluating the effectiveness of the system will be given. The Experiments section presents the way of creating and describing the application that allows for combining the augmented reality and dynamic simulation. The Results and Discussion section presents an evaluation of the system’s functioning, an evaluation of the decisions made in the creation of the collaborative application, and the results of testing. The Conclusions section describes the recommendations for industrial enterprises to develop and implement augmented reality technology in the maintenance of oil pumps.

2. Materials and Methods

2.1. Main Functions of the Augmented Reality System

The main function of an augmented reality system for oil pump maintenance is to create a so-called one-stop shop for equipment operation. A number of tasks arise during equipment maintenance which can be accomplished in a few ways. These include the use of paper or electronic equipment manuals, searching for information on instruments (e.g., setting process values), consulting the dispatcher, and searching for information via third-party software. The equipment maintenance process is kept logged, often manually or using electronic systems with little or no reference to the manufacturing site. In this paper, an augmented reality system is proposed to implement the above-mentioned functionality. The key features of the developed system are presented in

Table 1. Main functions of the augmented reality system.

Function Name	Without AR System	With AR System
Troubleshooting or Preparation for Maintenance
Search for the right piece of equipment	Comparison of flow charts or diagrams and actual equipment	Call up and use backlighting, colorization (e.g., arrows, sounds, when a watch or wristband vibration is used). The system automatically visualizes an object that requires the operator’s attention.
Pump disassembly	Electronic or paper instructions, personal or work experience of the operating staff	Calling up and using the assembled 3D model of the equipment, on-screen demonstration of the equipment required to service the pump (e.g., types of keys).
Involving additional information (e.g., technological parameters)	Search for instrument information, use portable devices, compare real and system objects, call the dispatcher	Receive information when this function is activated at the same time on any piece of equipment.
Eliminate the Cause or Perform a Maintenance Action
Testing of “what if…” actions	Personal or work experience of the operating staff	A dynamic model that allows any scenario to be simulated (e.g., switching equipment off, putting process areas on or off, increasing productivity).
Call for technical support	Using multiple devices to make calls, transfer information, etc.	Using a single screen, switching to the real-time view to display the object as seen by the staff, marking by the expert and visualization of these markers in front of the operating staff.
The Final Stage of Maintenance
Introducing service operations into the system	Use of electronic devices or a journal	Automatic service logging.
Quality control	Use of additional staff and teams	Automatic quality control by algorithms (photo and video capture and classification and recognition of user actions).
General System Functions
Voice control	-	Response to user commands - Search for the right piece of equipment - Start or stop equipment. - Call the dispatcher - Call the expert - Upload the model - Take a photo - Start or stop video

.

2.2. Development of the Structure and Connection of the Dynamic Model and the Augmented Reality System

In this study, the model of oil flow separation in the separator was chosen as a research model. The model was made in the Aveva Dynamic Simulation software package. The authors emphasize that the development and verification of the model was part of a separate study like [28] and was not essential in this study. An exterior view of the model connected to the augmented reality system is shown in the Figure 1.

This model was interfaced with a communication-level server and had a number of functional features. As a result, the model could be linked to an augmented reality system and used in conjunction to gain efficiency in oil pump maintenance:

1. The dynamic model was exemplary and contained a process model, a power system, and automation system models;

2. The dynamic model provided predictions of the changes in the technological parameters and parameters of the energy supply system as well as the automation system in dynamics, taking into account the inertia of the facilities;

3. The dynamic model had an acceptable solver speed for real-time simulation. This means that its use in an augmented reality system would provide information about the object with an acceptable response time;

4. The dynamic model was linked to the object communication layer and used data from sensors and actuators of real objects for calculations;

5. The data of the dynamic model changed at intervals defined by a time step. The time step size depended on the operation of the network communication layer and the data transmission protocols used. Changing this time step had no influence on the operation of the dynamic model. However, it did affect the result at a particular level where the dynamic model was applied.

In this study, a framework has been developed to create a special real virtual environment for pumping equipment maintenance. The functionality of the developed framework is the same as that listed in Section 2.1. An essential feature is its modularity and ability to integrate easily with any digital solution and technology, such as digital twin and analytical modules based on artificial intelligence systems. The first step was to define the communication protocols between the dynamic model and the AR system. Aveva Dynamic Simulation as the main integration protocol supports OPC protocol (OPC UA). The communication channel between the dynamic model and the OPC server was configured by the developer of the dynamic simulation software environment and was not an object of this study. The subject of this study was essentially finding and testing a communication channel between the augmented reality system and the OPC server (OPC UA). Implementing and configuring the OPC UA in the augmented reality application development environment was challenging for the authors (it was possible to connect the values once without updating in real time). Since the augmented reality application functions like the WEB application, obtaining values from the OPC server required the development of a special polling block (i.e., the program code) which pulled the main code upon obtaining values from the server. The authors found this method unacceptable because of the potentially large number of errors due to delays in obtaining the signal. Therefore, a structure was proposed which is shown in Figure 2.

This structure was based on the application of the modern MQTT protocol for message transfer. In order to use it, the MQTT broker (Mosquitto MQTT) was added to the framework to coordinate message transmission between the dynamic model and the augmented reality system. This protocol is a basic protocol in IoT systems, and the effectiveness of its use was repeatedly proven [34,35].

Thus, according to the authors, the optimal structure for the development of a software application for the collaborative use of dynamic simulation and an augmented reality system should correspond to the scheme shown in the figure and include the following functionality and technological links:

1. The system contains three levels: communication, application, and physical;

2. On the physical level, there is an object (laboratory equipment was used in this work), a physical object, and a dynamic model. The power system and the control system are distinguished separately on the physical level. Allocating these systems into a separate class is necessary to expand the system maintenance functionality. In the separate version, it is possible to transfer these parts to the real or virtual execution environment at any time as the production need arises. For instance, in the normal mode of operation, all systems are real. When the task of introducing new control algorithms arrives, a dynamic model is used instead of the object, and the control system remains real or is virtualized as needed. This approach will, among other benefits, significantly reduce costs during commissioning new systems and the maintenance of old systems;

3. The main protocols of the communication module are MQTT and OPC (OPC UA). OPC allows communication between the dynamic model, automation system, and power supply system. Then, the data from the OPC server are published on the MQTT broker, which interacts with the augmented reality application on HoloLens 2 glasses. The SQL server in this structure is a tool to keep a history of all changes at all levels. All values are recorded with corresponding timestamps. For the system to work correctly, all components have to be time-synchronized;

4. The physical and communication levels in the system are arranged in such a way that any system using the digital technologies described in [28] can be connected to them. It is sufficient only to support the MQTT or OPC (OPC UA) protocol for data exchange;

5. At the application level, the structure highlights the augmented reality application running on HoloLens 2. It should be noted that a smartphone or tablet can be used as a device for interaction with the user. The main functionality of the developed solution will not be broken, but it is necessary to emphasize the advanced capabilities in the case of HoloLens 2 application, related primarily to the 3D approach to visualization and other factors.

2.3. The Methodology of the Experiments and Testing and Evaluation of the System’s Efficiency

Testing and evaluating the system effectiveness is quite a complex and ambiguous issue. Undoubtedly, this topic and process requires more in-depth study. The authors emphasize that this work is one of the first on the method of complex application of augmented reality for the maintenance of oil pumps, and accordingly, testing will be greatly simplified. When the work continues and the final product is developed, the approach to testing should be more complicated, a special application for testing should be written, and algorithms at all levels of the system structure should be run (Figure 1).

To test the suitability of the developed structure, the following methodology was used:

1. An application for installation with multiple units of equipment (up to 10 parameters and 10 3D models of units of equipment) was developed;

2. The performance of the main channels was assessed, and the result of test signal transmission was recorded. The following main characteristics were measured: CPU, used memory, peak memory, and power;

3. An application for 3 settings (each with up to 20 parameters and 10 3D models) was developed to identify the location of installations in 3-dimensional space with the possibility of changing the position by the user’s action;

4. The main characteristics were measured by the protocol;

5. An application for five installations under the conditions of paragraph 5 were developed;

6. The main characteristics were measured by the protocol;

7. Analysis of the test protocols with a conclusion on the feasibility of collaborative application of the augmented reality system and dynamic model was conducted.

Additionally, to evaluate the developed solution, the augmented NASA Task Load Index (NASA-TLX) test was used to evaluate the cognitive-psychological factors when working with the application [36,37].

To implement the test, 5 scenarios were simulated for the same task: “Vibration sensor detected pump failure”. The task was to put the pump out of operation:

1. The testers had only a smart shield and the option of calling a person with access to the equipment;

2. The testers had a smart shield and a Supervisory Control and Data Acquisition (SCADA) system where the values of all system promoters could be viewed;

3. The testers had a smart shield and documentation with information on how to connect the system;

4. All interaction was with the augmented reality system only, and commands were only entered manually;

5. All interaction was with the augmented reality system only, and commands were entered by voice and manually.

To conduct the test, we selected a group of test takers and asked them to complete the test. The test subjects were asked to answer the following questions:

• Assess whether you felt threatened, discouraged, or irritated.
• How difficult was it for you to do your job?
• How successful were you in your task?
• How quickly did you complete the task?
• How physically demanding was the task?
• How difficult was the task as a whole?

3. Experiments

A vertical centrifugal pump with a Grundfos electric drive (CR1S-4 A-FGJ-A-E-HQQE), which is a part of the laboratory installation shown in Figure 3, was used as a service object.

The following is a list of equipment from the installation: 2 Grundfos pumps, 2 tanks, heat exchanger, pipeline section, valves, sensors, and electric actuators. The sensors displayed the status of such indicators as the flow, temperature, pressure, and pressure drop.

The figure shows a hub for the physical simulation of the power supply system (Figure 4), as well as a hub of the “smart shield” complex for the physical simulation of the power supply system’s control system (Figure 5).

The main distinctive feature of the applied equipment was the collaborative use of the real and virtual parts of the equipment. In this case, the control system as well as the power supply system could be real and virtual. The user would be able to pass from one type of these systems to the other, solving various tasks.

The experiment in this paper was as follows:

1. An augmented reality application for pump maintenance was created, including an element of dynamic simulation according to the method described in paragraph 2.2;

2. The augmented reality application was tested using the methodology described in Section 2.3.

Microsoft HoloLens 2 augmented reality glasses were used as a device for user interaction with the augmented reality system.

The characteristics of the PC for the experiments are presented in the

Table 2. Characteristics of the PC for the experiments.

Personal Computer Number	Hardware List	Software List
PC 1	Processor: Intel(R) Core(TM) i5-7500 [email protected] GHz 3.41 GHz Memory: 8 GB	OS Windows 10 Pro 64 bit Aveva Dynamic simulation KEPServer EX 6 Mosquito broker MQTT Explorer SQL Server Management Studio (SSMS)
PC 2	Processor: Intel(R) Core(TM) i7-4710HQ [email protected] GHz 2.49 GHz Memory: 16 Gb Video Card: NVIDIA GeForce 840 M	OS Windows 10 Pro 64 bit Unity SDK Vuforia Microsoft Mixed Reality Toolkit …. Mosquito broker MQTT Explorer KEPServer EX 6 Microsoft Hololens App for Windows Device Portal (Hololens WEB portal)

.

Figure 6 shows the testing process of the developed software.

4. Results and Discussion

To test the effectiveness of the solutions proposed in the paper, the following application was created in Unity software. Figure 7 shows a detailed view of the developed application.

The 3D installation model was exactly the same as the dynamic model. The variable names were identical at all levels of use (dynamic model, data channels, communication server, etc.). The use of consistent tag names was essential, as it reduced the development time and potential errors.

Figure 8 shows a list of the dynamic model variables in the OPC Quick Client to create application 1.

Figure 9 shows the MQTT broker operation. The Iot_aveva topic was subscribed to the KEPServer agent, which collected values from the Aveva OPC server and forwarded them to the Mosquitto broker. The iotgateway topic was subscribed to the MQTT agent, which sent messages from the user through the MQTT augmented reality system to the broker. Additionally, the MQTT client KEPServer was subscribed to this topic, which sent values from the MQTT Broker back to the OPC server.

Three applications were developed to investigate the effectiveness of the collaborative use of the dynamic model and the augmented reality system:

1. One installation with 15 parameters shown on the 3D model, with 2 parameters being controllable. They could be changed at all levels of the system, with the main functional feature of the system being their change via an augmented reality application;

2. Three units similar to the installation described in point 1;

3. Five units similar to the installation described in point 1.

Figure 10 and Figure 11 show the appearance of the system as tested, with the view obtained via live streaming of the Microsoft Hololens application.

Therefore, we had three parts of laboratory units. The first was a technological process (Figure 3) with pumps, the second was a power system simulation unit containing part of the energy-mechanical equipment: motors for pumps (Figure 4) with the control system and smart shield (Figure 5), which allowed us to connect the pump to the power supply system. On the laboratory unit (Figure 3), the main technological parameters—pressure, flow rate, and temperature at different points—were taken. The dynamic model (Figure 1) represents the co-modeling of the power supply system and the technological process. Some of the equipment was represented in its real form on the laboratory unit. For instance, this included a pump motor with current, voltage, and power sensors and the switches of the smart shield. They allowed for controlling the power supply system. This equipment represented real objects. Partially the same objects and additional other objects were presented in the form of a dynamic model by their 3D models (the images of some parts of the installation as presented in the industry). The full 3D model of the installation is shown in Figure 11. Apart from the 3D view of this installation, the application showed the change of the parameters of the dynamic model in real time. Thus, according to Figure 2, we had a dynamic model of the process and physical models of the process parts. Figure 11 (left) shows the detailed part of the installation in its industrial 3D form. Aside from that, Figure 11 (right) shows the interaction between the user, the augmented reality system (control buttons), the real part of the equipment (laboratory stands that received control signals and changed the physical state of the equipment or the dynamic model), and the virtual part of the equipment (the dynamic model and its detailed 3D image). It should be emphasized that there was a difference between the variables in Figure 1 (dynamic model), Figure 8 (data collection server), and Figure 11. This was due to the fact that Figure 8 shows a list of possible variables of the dynamic model. The names of the tags in it were derived automatically and were defined by the Aveva Dynamic Simulation software manufacturer. Figure 10 and Figure 11 show the variables actually captured in the unit, and their nomenclature corresponds to the nomenclature of the industrial object. There was a difference between the parameters of the dynamic model, the augmented reality application, the parameters of the automation system, the parameters of the energy system, and the parameters accepted as standard in an industrial enterprise. This is a big concern. It is required to use a single standard for forming the variable name. However, currently such standards are not used, and the correlations between the variables are established by means of special linking tables. In this work, the following variables, shown as an example in Figure 8, were removed at the level of the dynamic model: special automation system operation variables (control loop starts (Master 1_B_out)), set point feed for temperature changes, flow (LVLSP) in streams and equipment, pressure (P and PB), temperature (T), flow rate (W) in the raw streams (S2, S5, S7, S9, and S11), in the source (SRC1 and SRC2), in the drum (V2), and in the final reservoirs (SNK1, SNK2). Additionally, the positions of valves XV1, 2, and 3, the status of motors M1, 2, 3, and 4, and other parameters of the power system (e.g., current and power voltage) were taken. Figure 8 shows a sample list of the variables. If necessary, one could display any parameter in the system (for one flow or type of equipment, there were about 100 parameters). The variables shown in Figure 11 for the engine are presented with the name used in production. These names must be clear to the users of the system. Therefore, I321 for the current, U321 for the voltage, and P321 for the power were used. It should be emphasized that these parameters were taken from the laboratory model of the pump shown in Figure 4 with the current, power, and voltage sensors mounted on them. Linking (linking the same and other variable names) was performed in the KEPServer (i.e., in the communication layer) of Figure 2.

Table 3. Communication channel test report (example test report).

Action	Expected Result	Test Result (Passed, Failed, or Blocked)
Sending a test message over a channel OPC UA: Aveva Dynamic Simulation	Read/write value	Passed
Sending a test message over a channel OPC UA: AR app	Read value	Passed
Sending a test message over a channel OPC UA: AR app	Write value	Blocked
Sending a test message over a channel OPC UA: MQTT broker directly	Read/write value	Blocked
Sending a test message over a channel OPC UA: MQTT broker through an MQTT agent	Read/write value	Passed
Sending a test message over a channel MQTT broker: AR app	Read/write value	Passed
Sending test control command (physical device: smart shield) over a channel AR app: Aveva Dynamic Simulation	Read/write, deliver, and understanding command	Passed
Sending test measured signal (physical device: smart shield) over a physical channel: AR app	Read/write, deliver, and understanding signal	Passed
Testing Channel OPC UA: SQL	Recording and storage of all events in the system with a time stamp (5 times for 2 min)	Passed

shows the results of the channel test.

The test protocol (

Table 4. Assessment of the application.

Test Number	Tested App	CPU (fps)	Used Memory (Mb)	Peak Memory (Mb)	Power (% per Minute)
1	1st app (1 unit with 3D objects, 15 tags)	19–62	837	955	1
2		42–68	838	841	1
3		34–62	828	830	1
4	2nd app (3 units with 3D objects, 45 tags)	17–31	843	845	1
5		27–45	850	858	1
6		24–40	841	843	1
7	3rd app (5 units with 3D objects, 75 tags)	26–40	847	849	1
8		32–42	850	853	1
9		27–45	875	877	1

) shows that data transfer via the augmented reality application to the physical devices and the dynamic model was possible, with MQTT being the primary protocol. However, the physical devices and the dynamic model at greater numbers did not support the MQTT protocol. For this purpose, the OPCUA protocol was required to be included in the system structure.

Table 4 shows the evaluation protocol for the application.

The system test protocol (Table 4) showed the performance of the proposed solutions. The values shown in the table are acceptable for the implementation of such applications in industry. The CPU parameter did not fall below 20 fps, and the memory parameter did not rise above 960 Mb. Moreover, the “power” parameter showed that when the application was launched and tested, the Microsoft HoloLens2 augmented reality glasses worked in normal mode, with battery discharge in accordance with the technical documentation (one percent per minute during active operation). Thus, the solutions presented in this paper are relevant for the development of a system for the collaborative use of dynamic models and an augmented reality system.

Figure 12 shows the NASA-TLX test for assessing cognitive-psychological factors when using the application.

The figure shows the average NASA-TLX test results for all the participants in the experiment. A total of 93 people took part in the experiment. The participants ranged in age from 21 to 28 years, with 81 males and 17 females.

To implement the test, five scenarios were simulated for the same task: “vibration sensor-detected pump failure”. The task was to put the pump out of operation. The division into groups was performed as follows:

1. The testers had only a smart shield and the option of calling a person with access to the equipment;

2. The testers had a smart shield and SCADA system where the values of all system promoters could be viewed;

3. The testers had a smart shield and documentation with information on how to connect the system;

4. All interaction was with the augmented reality system only, and commands could only be entered manually;

5. All interaction was with the augmented reality system only, and commands could only be entered by voice or manually.

The test results showed that the testers felt less threatened when using the augmented reality system. The participants noted the difficulty of the task and felt a sense of heaviness in completing the task. However, they did note the physical difficulty of completing the task. This was probably largely due to the need to interact with the augmented reality glasses. It should be emphasized that the age of the participants did not exceed 27 years, as the experiments were carried out in student groups. In an enterprise, this indicator may be higher, as the average age of people is over 27 years old. Therefore, when evaluating the system’s performance, an additional indicator should be introduced for the level of proficiency in interacting with augmented reality glasses. In this paper, the skills of the testers were minimal; most of them were working with the glasses for the first time.

5. Conclusions

The main goal of this paper was to check the operability of the oil pump maintenance system functionality when the augmented reality system and the dynamic modeling system were used together. The conducted research has shown the possibility and expediency of using this approach. At the same time, it is necessary to mark a number of moments which are worth paying attention to using this approach:

1. The structure of the system should have a modular principle of construction and be able to scale and expand its functionality. For this purpose, it is required to allocate the central communication level. This level should include software and hardware that allows one to transfer data using the most advanced protocols such as OPC UA, MQTT, and others. In this case, the use of OPC UA is mandatory, as is the main protocol of interaction with physical models and the real object;

2. The modular principle of the structure of the developed system should support the possibility of allocation of various subsystems involved in common processes (e.g., the control system, power supply system, and other similar systems). This structure will provide an option to replace the real parts of the system with virtual ones and vice versa for the implementation of additional functionality. The virtual models are exact 3D models of each part and a dynamic model that replicates the dynamic properties, including the inertia of each object;

3. When evaluating the performance of a particular augmented reality solution, it is preferable to use at least two types of testing: testing the application performance and testing the cognitive-psychological impact of the system on a person;

4. Performance testing of the system has shown that it is possible to use dynamic simulation systems in augmented reality together with real process data. Experiments to evaluate the CPU and memory consumption for the application’s performance showed its performance both with minimal and increased load in the form of different numbers of connected dynamic models (the maximum load tested was five dynamic models running simultaneously). Since this test showed no significant performance degradation, the authors considered the connection of five models to be sufficient;

5. Testing of the cognitive-psychological impact of the system on humans has shown that the emergency shutdown task when using the augmented reality system in collaboration with the dynamic model yielded improvement in the following ways: mental demand, physical demand, temporal demand, performance, and frustration.

The authors emphasize that this work is a continuation of the study described in [28]. The continuation of the works presented in the article is the works on extension of the connected dynamic models and connection of the dynamic models and diagnostic systems to determine the life cycle and residual life of oil pumps. It also includes methods of the collaborative use of real and virtual parts of the system and the development of an extended method of testing augmented reality applications. Works showing the connection of other digital technologies, such as the digital twin of electrical equipment, and its use in industrial applications and a number of other works are also a continuation of the present work.