A Real-Time Implementation of Air Audit System for Compressors towards Energy Conservation: An Industrial Case Study

Department of Electrical and Electronics Engineering, PSG College of Technology, Coimbatore, India Department of Electrical and Electronics Engineering, Kongu Engineering College, Perundurai, India Department of Electrical and Electronics Engineering, Karpagam Academy of Higher Education, Coimbatore, India CRISD, School of Engineering and Technology, University of Technology Sarawak, Sibu, Malaysia Department of Information Technology, Dambi Dollo University, Dembi Dolo, Ethiopia


Introduction
A compressor is basically a mechanical device that simply reduces the volume of air by increasing its pressure. Compressed air has a lot of applications such as refrigeration, air tools, energy storage, air brakes, and underwater breathing apparatus. Like any machine, a compressor requires electrical energy to work. Monitoring the power consumption and the pressure of the compressor is the main objective of this paper. An air audit is a review of an operation's use of compressed air. e audit evaluates both the generation and distribution of compressed air. It assesses a compressed air system and identi es areas of ine ciency with corresponding cost savings, if recti ed.
A sales engineer can always only predict the requirements of the customer, but he cannot estimate it. An air audit system should be designed for enabling sales engineers to capture customer's requirement of the compressed air system. e air audit system is required so that it can create quick audit solution to support customers with estimating air consumption. It also has the potential to educate the customers with identifying the improvement areas after understanding their existing demand pattern. ere are a few outcomes that should be met using the air audit system. e most important one is to understand the air demand pattern with respect to flow and pressure. e flow is calculated indirectly through power which is in kW. en, the existing power consumption details and pressure data must be captured. e customer should be able to select the optimum compressor size based on the demand. e process involved in data handling is given as follows: (1) Data logging process: (a) Measure current, voltage, and true power of each air compressor and log the data (b) Tank pressure to be logged in a separate logger mounted in an air receiver (c) Logging frequency (5 s-1 min) and logging period (24 hr-168 hr) are configurable (2) Data transfer: (a) End device to be communicated to cloud through a mobile or tab (tablet) (b) e mobile will receive the data from the logger through an app over WiFi (3) Report generation: (a) Mobile GUI to be developed for graph and report generation (b) All the received data to be stored in a cloud database (c) Web UI to be developed to generate report from the database and view the graph e first and foremost thing in developing the air audit system is the power measurement module. A detailed energy audit considering air conditioners was performed in [1]. In this paper, the authors assessed that 60-65% of the total energy consumption in commercial areas is accounted with air conditioners. Air flow rate and the difference in enthalpy across the coils of the evaporator are evaluated to show the performance of the air conditioners. An energy audit for the residential building is conducted in [2] involving majority of the electrical loads. It is observed that 24% of the total energy consumption of the building is from the air conditioners. e authors have also made an attempt to provide an energyefficient air conditioner without conciliation and comfort for humans. e selection of microprocessor for the air audit system was made by referring to [3]. A detailed study on the MSP432 processor with respect to power consumption was presented in the paper. Since the module is battery operated, the selection of optimized microcontroller is very essential. Brief details of the test-bed employed at Texas Instruments (TIs) as well as TI's energy trace technology are reviewed in paper [3]. A detailed procedure involved in the selection and sizing of batteries is provided in [4]. Lithium-ion batteries were selected due to their high charge density and minimum memory effect.
A mobile app-based harmonic distortion analyzer was developed in [5]. Transferring the logged data over WiFi to a mobile app is explained. e authors have also implemented the plotting and report generation features, which are some of the important features to be incorporated in an air audit system. e paper [6] provides a brief introduction to the programming of IDEs (integrated development environments) such as Code Composer Studio and Energia. e MSP432's operating parameters, as well as how to interface with peripherals, are also discussed. It also provided an insight on the memory, timing, power management, and communication systems of the microcontroller. A case study on energy audit in a building is given in [7] along with the best practices for various electrical loads. e extensive critiques given in the literature gives various aspects that are to be implemented into the air audit system. e only challenge that lies ahead is to incorporate all these features into a single module, so that it meets the desired conditions while using it as an air audit system. As the module is to be used in an industry, the design must meet all the industrial standards [8,9]. e paper is organized as follows: Section 2 presents the power measurement; Section 3 exhibits the component section; Section 4 illustrates the software design layout and experimental results; discussions and final conclusions are given in Sections 5 and 6, respectively.

Power Measurement in the Air Audit System
Power measurement involves sensing the voltage and current at every instant and feeding it to the microprocessor for computation [10,11]. e microprocessor returns the real power, power factor, and various other power parameters. However, before developing an energy meter, it must be understood that the primary objective is air auditing. A compressor is a three-phase device, but we cannot design a 3-phase energy meter for auditing because it increases the cost per unit. A compromise must be made between accuracy and cost. To estimate the accuracy, a comparative study of different wattmeter methods is made using simulation software Multisim. e three wattmeter method is the most suitable method to compute power in a compressor. As mentioned earlier, it increases the overall per unit cost of the air audit system. is leaves with 2 choices: a two wattmeter method and a single wattmeter method. e Multisim circuit schematic for the two wattmeter method is designed using three voltage sources connected in star and three RL (resistive-inductive) loads connected in delta and is shown in Figure 1. A slight imbalance in source and load is provided to take care of the practical imbalances. e wattmeter reading of the circuit can be viewed by opening the wattmeter, as shown in Figure 2. e sum of the two watt meters gives the actual power consumption. e total power consumed is 11.98 kW according to the simulation. e power factor (PF) can be calculated using the formula, PF � cos(arctan(( , only if the system is balanced. Since the compressor is mostly unbalanced, a separate power factor measurement circuit is required. e wattmeter readings of the three wattmeters are shown in Figure 3. e total power consumed is 11.97 kW using the three wattmeter method, which verifies the accuracy of real power in the two wattmeter method, but shows large error in power factor measurement. Even though the accuracy of real power is good, we cannot go for two wattmeter because the cost per unit is still high. Furthermore, a single wattmeter can be connected to any one of the phases. erefore, with a little compromise in accuracy, we can get fairly good results using the single wattmeter method. e Multisim circuit schematic with three individual wattmeters connected to every phase is shown in Figure 4. e actual rating of the compressor and that while using single wattmeter are analyzed for each phase. Even though this method also requires power factor measurement circuits, the fact that the voltage and current measurement circuits have reduced will cut out the cost significantly. Table 1 provides a concise comparison of the two methods analyzed. e results show that the single wattmeter method is most suitable for air audit application. e maximum error that occurs in real power measurement in simulations is around 4.06%, where the computation is shown in Table 2. e meter when developed will come under class 4 accuracy, which is shown in Table 3. For an auditing tool, class 5 gives fair results. erefore, it is decided to go with a single wattmeter method for the air audit system. Table 4 shows the accuracy when the wattmeter is connected in individual phases. e following equation helps in determining the total system error (εS) which is 4.06%: where ƐIT is the instrument transformer error, ƐVP is the voltage probe error, ƐCP is the current probe error, εS is the total system error, ƐM is the meter error, and ƐCE is the computation error.
A power quality analyzer (PQA) that was designed for this specific purpose was analyzed to validate the simulation results. e block diagram of the PQA is shown in Figure 5.
e voltage input to the power quality analyzer was given using a 3-phase supply. e derivative of the current probe is      coupled to a shunt resistor. e voltage drop across the shunt is taken as the current input. In order to practically test for different loads, the shunt resistor was soldered out and a signal generator was used. Different load conditions were emulated by varying the voltage of the signal generator.
e voltage and current signals from the PQA were tapped and fed into a TI CC3220 microcontroller. e power computation algorithm was implemented in the controller, and the results were verified. e microcontroller was capable of computing real power and power factor and logging   the data. e logged data were then transferred to a mobile application over WiFi through the inbuilt WiFi module. e received data were saved as a .CSV file in the internal memory of the mobile phone. e saved data were then used to plot voltage, current, and power factor graphs using Highcharts®.
As per the results, the two wattmeter method is most suitable for power measurement. However, the difference of accuracy between the single wattmeter method and the two wattmeter method is not quite small. Hence, the single wattmeter method is chosen as it was cost-effective. e accuracy of the system is 4.06%, and so the system falls under class 4.

Selection of Components for the
Proposed System e block diagram of the proposed air audit system shown in Figure 6 consists of single-phase measurement circuit using a voltage probe and a current transformer-based current probe. e voltage signal and current signal are then conditioned and fed into the microcontroller. Two more signals are fed through the zero crossing detection circuit to compute power factor. Since it is a battery-operated system, a rechargeable lithium-ion battery of 7.4 V powers the entire module. e air audit system also comprises an external realtime clock for obtaining time stamped data. A serial flash is used to log the power and pressure data. e communication is established using an external WiFi module.

Voltage and Current Probe.
A voltage probe with alligator clips is used to sense voltage. Since the maximum voltage that can occur even when connected across line-toline is less than 600 V, the rating of the voltage probe is selected to be 600 V. e voltage probe from Kyoritsu, KEW 7234, is selected for this application. A split core current transformer-based current probe with a maximum rating of 600 A is used to sense current. Since 600 A is not a standard rating, a 1000 A rated current probe is used. CCT50 from MECO instruments has a rated output of 5 A, and it also can be increased to 10 A, so MECO CCT50 is selected for the air audit system.

Differential Amplifier.
e voltage signal sensed using the probe is stepped down using a potential divider circuit. en, the minimized form of voltage is fed into an AMC1100 fully differential amplifier and the output is fed to the microcontroller. e key advantage of using AMC1100 is to enable the measurement of the signal with a good accuracy by eliminating the common mode noise. e differential amplifier must also be able to provide isolation in accordance with UL (Underwriters Laboratories) standards.
is process of blocking current flow such that there is no direct conduction path between circuits is often referred to as galvanic isolation. Isolation is necessary to prevent the flow of hazardous voltages between circuits. It also helps in avoiding potential ground loops. Considering these scenarios, AMC1100 (it  Figure 6: Block diagram of air audit system. 6 Mathematical Problems in Engineering comes under UL1577) is chosen to be the fully differential amplifier.

TI Microcontroller.
e simple link MSP432P401RIPZ microcontroller is an optimized wireless host MCU (microcontroller unit) with an integrated 16-bit precision ADC (analog-to-digital converter). e MSP (mixed signal processing) processor has a dedicated DSP (digital signal processing) processor which is capable of performing FFT (fast Fourier transform) analysis. e flash memory offered by the MSP432 processor is 256 kB. An external flash is also used to log the real-time data captured by the measurement system. Considering the fact that the microcontroller consumes very low power, it is selected for the air audit system.

Flash Memory.
e application requires logging of at most 7 days of data. e most trivial data format that could be used to log data is "HHMMSS,VVV,III.I,0.PF." e minimum logging interval that is required is 2 seconds. However, the flash is selected such that it can log at least every 2 second data for the data format mentioned above. e string occupies 21 bytes. To log every 2 second data for 7 days, memory space for 302400 strings should be available.
is requires 6.05 Mb of memory. Hence, a 64 Mbit flash (8 Mb) is selected. e flash memory is connected to the microcontroller using SPI protocol.

WiFi Communication.
e WiFi module is based on the standard 802.11 b/g/n and is mostly in sleep mode. During every loop of program, it checks for the connectivity for particular WiFi access points. When it gets connected to those particular access points, it starts to monitor client requests. If any client request matches the predefined code, the bulk of the logged data is transferred to the mobile app from where the request occurs.

Battery.
e microcontroller and almost every other component work at 3.3 V. e only component which requires a 5 V supply is the isolated differential amplifier. e high side of AMC1100 requires a 5 V supply. erefore, the battery must provide greater than 5 V. e battery must also be rechargeable. e total power consumption of the hardware is analyzed, as in Table 5.
us, Li-ion is the obvious choice considering its high energy density and almost zero memory effect. A Li-ion battery typically provides 3.6-3.7 V. Two 18650 Li-ion batteries with required mAh power the air audit system [12][13][14].

Software Design and Layout
e hardware schematic design is done using xDX Designer from Mentor Graphics. e MSP432 processor is programmed using Energia IDE, which is an open-source electronics prototyping platform for TI products.

xDX Designer.
e xDX designer makes available an entire schematic design elucidation for design formation, description, and reprocess. It also provides all necessary things required for designing circuits. e home page of the xDX designer is shown in Figure 7. e circuit schematic is drawn in xDX Designer, and then it is converted into PCB using the PADS package from Mentor Graphics. e software has provisions for netlist and Bill of Materials generation. A part of the schematic of the power quality analyzer is shown in Figure 8. e printed circuit board (PCB) of the 3-phase power quality analyzer using which the proof of concept was verified is shown in Figure 9.

Energia IDE.
Energia, which is an open-source IDE for programming TI products, was used in this project. e editor window is shown in Figure 10. e code can be written in simple embedded C language and can be compiled to a specific TI board by selecting it. en, it is dumped into the microcontroller using its serial port. ere is also good support for this platform as developers are bringing in new libraries for specific boards every day. xDX Designer is mostly used in industries for precision embedded drawing. Since it is a piece of paid software, the learning process was not simple. On the other

Application Development
A mobile application was designed using MIT App Inventor to receive the logged data to the mobile phone over WiFi.
e hardware unit consists of a Cypress WiFi module for communicating. e opening screen of the app is shown in Figure 11. e "Enter IP (Internet Protocol) address here" text box can be directly typed in or the dropdown button can be clicked to open a list of already saved IP addresses. e connected status text box indicates if the phone is connected to the IP address entered in the IP text box. e text boxes  indicating voltage, current, etc., are instantaneous, and update time can be configured. e list of saved addresses used at the time of development is given in Figure 12.
Even after the mobile gets connected to the data logger system, the logger system does not respond unless a particular command is sent from the app. e command is sent when the "Get Data" button is pressed. e logger system responds back with a large CSV string which is written into a file with date as the file name. e last data are updated in the text boxes indicating voltage, current, power factor, etc. e data from the file are again used to plot graphs using Highcharts® which is an interactive JavaScript chart for web pages.
When the "Plot Graph" button is pressed, the browser is opened and the line graph for root mean square (RMS) voltage, RMS current, and power factor is displayed. Provisions for opening the saved file are also given within the app. When the "Open File" button is pressed, an external application for opening .CSV files is launched. e file is saved in the internal memory of the mobile phone. e file can also be deleted from within the app using the "Delete File" button. e "About" button opens a page which instructs how to use the app, and the "Exit" button is used to close the app without crashing. e app screen after receiving a dataset is shown in Figure 13. An example of Highcharts ® line graph is shown in Figure 14. e mobile application was developed, which captures the real-time power measurement data, and it also displays the data in the mobile app user interface. e data are stored along with its time stamp. ese huge data are stored in the form of .CSV     file. e data in .CSV files are used for report generation and plotting of interactive charts, which help us to monitor the data remotely [15].
Furthermore, for the air quality auditing, few methods such as autoregressive integrated moving average (ARIMA), support vector machine (SVM), random forest (RF), and logistic regression (LR) are elucidated in [16]. A data fusion method for indoor air quality monitoring is experimented in [17]. 5G modeling [18], participatory method [19], and Internet of ings sensing [20] are the topologies which have also been considered for the air quality audit. ese methods are not intended for compressors which is the major scope of this paper.

Conclusion
In this paper, a thorough understanding of drawing the circuit schematic for PCB development was obtained. e understanding towards selection of components for a particular application was obtained while working in the air audit system. A proof of concept for the air audit system using the already existing power quality analyzer and evaluation boards is implemented. A mobile application was also developed using MIT App Inventor for displaying the logger data in a mobile phone. e data were also pictorially analyzed in the form of graphs using Highcharts ® . An attempt has been made in developing a complete standalone module for the air audit system considering lot of features, such as power consumption, flexibility, and accuracy. e hardware design process is constantly improved depending on the industrial standards and innovative ideas. e overall objective is to bring this module as a product in the market through which customers or sales engineers can be able to analyze the compressors.

Data Availability
e required research data can be obtained from the corresponding author upon request.

Conflicts of Interest
e authors declare that they have no conflicts of interest. Mathematical Problems in Engineering 11