**8. System implementation**

Based on the previous requirements the WSN-based VOC monitoring system prototype was implemented and tested at *Mantova, Italy,* petrochemical plant.

The aim was to test a new distributed instrument for collecting VOC emission data in realtime with a high degree of flexibility and scalability, thus transferable to other monitoring

Real-Time Monitoring of Volatile Organic Compounds in Hazardous Sites 231

The GPRS unit operates on the basis of a proprietary communication protocol over TCP/IP, with DHCP. Dynamic re-connectivity strategies were implemented to provide an efficient and reliable communication with the GSM base station. All the main communication parameters like, IP address, IP port (server and client), APN, PIN code and logic ID can be

The system is based on an embedded architecture with high degree of integration among the different subsystems. The unit is equipped with various interfaces including LAN/Ethernet (IEEE 802.1) with TCP/UDP protocols, USB and RS485/RS422, in addition to a wireless interface, which provides short range connectivity. The sensor acquisition board is equipped with 8 analogue inputs, and 2 digital inputs. The SN unit is also equipped with a Wireless Interface (WI), represented in Fig. 5, providing connectivity with the EN units. The WI operates in the low-power, ISM UHF unlicensed band (868 MHz) with FSK modulation, featuring proprietary hardware and communication protocols. Distinctive features of the unit are the integrated antenna, which is enclosed in the box for improved ruggedness, and a PA and LNA for improved link budget. The PA delivers some 17 dBm to the antenna, while the receiver Noise Figure was reduced to some 3.5 dB, compared with the intrinsic 15 dB NF of the integrated transceiver. As a matter of fact, a connectivity range in line-of-sight in excess of 500

This results in a reliable communication with low BER, even in hostile e.m. environments. The energy required for the operation of the unit is provided by a 80 Ah primary source and by a photovoltaic panel equipped with a smart voltage regulator. Owing to a careful lowpower design, the unit could be powered with a small (20 W) photovoltaic panel for

A picture of one of the SN unit installed at the Mantova plant is represented in Fig. 6, left. The battery and photovoltaic panel are clearly visible; the GPRS unit is the grey box close to

remotely controlled.

Fig. 5. Block diagram of the wireless interface

undiscontinued and unattended operation.

meters was obtained.

stations as needed, reconfigurable, in terms of data acquisition strategies, and economically sustainable as compared to traditional fixed monitoring stations.

Fig. 3. The hybrid cluster-three network configuration

Critical locations were identified along the perimeter of the industrial sites, and within some specific relevant internal areas potentially involved in emissive processes. Seven SNs and 10 ENs to be described in the following have been deployed so far.

### **8.1 The SN unit**

Each SN unit typically consists of the five components such as sensor unit, analogue digital converter (ADC), central processing unit (CPU), power unit, and communication unit. Communication unit's task is to receive command or query and transmit data from CPU to outside world. CPU is the most complex unit; it interprets the command or query to ADC, monitors and controls power if necessary, processes received data and manages the EN wake-up.

The block diagram of the SN unit is represented in Fig. 4. It consists of a GPRS antenna and GPRS/EDGE quadriband modem, a sensor board, a wireless unit and a micro-controller ARM-9, operating at 96 MHz clock.

Fig. 4. Block diagram of the Sink Node Unit

stations as needed, reconfigurable, in terms of data acquisition strategies, and economically

Critical locations were identified along the perimeter of the industrial sites, and within some specific relevant internal areas potentially involved in emissive processes. Seven SNs and 10

Each SN unit typically consists of the five components such as sensor unit, analogue digital converter (ADC), central processing unit (CPU), power unit, and communication unit. Communication unit's task is to receive command or query and transmit data from CPU to outside world. CPU is the most complex unit; it interprets the command or query to ADC, monitors and controls power if necessary, processes received data and manages the EN

The block diagram of the SN unit is represented in Fig. 4. It consists of a GPRS antenna and GPRS/EDGE quadriband modem, a sensor board, a wireless unit and a micro-controller

sustainable as compared to traditional fixed monitoring stations.

Fig. 3. The hybrid cluster-three network configuration

**8.1 The SN unit** 

ARM-9, operating at 96 MHz clock.

Fig. 4. Block diagram of the Sink Node Unit

wake-up.

ENs to be described in the following have been deployed so far.

The GPRS unit operates on the basis of a proprietary communication protocol over TCP/IP, with DHCP. Dynamic re-connectivity strategies were implemented to provide an efficient and reliable communication with the GSM base station. All the main communication parameters like, IP address, IP port (server and client), APN, PIN code and logic ID can be remotely controlled.

Fig. 5. Block diagram of the wireless interface

The system is based on an embedded architecture with high degree of integration among the different subsystems. The unit is equipped with various interfaces including LAN/Ethernet (IEEE 802.1) with TCP/UDP protocols, USB and RS485/RS422, in addition to a wireless interface, which provides short range connectivity. The sensor acquisition board is equipped with 8 analogue inputs, and 2 digital inputs. The SN unit is also equipped with a Wireless Interface (WI), represented in Fig. 5, providing connectivity with the EN units. The WI operates in the low-power, ISM UHF unlicensed band (868 MHz) with FSK modulation, featuring proprietary hardware and communication protocols. Distinctive features of the unit are the integrated antenna, which is enclosed in the box for improved ruggedness, and a PA and LNA for improved link budget. The PA delivers some 17 dBm to the antenna, while the receiver Noise Figure was reduced to some 3.5 dB, compared with the intrinsic 15 dB NF of the integrated transceiver. As a matter of fact, a connectivity range in line-of-sight in excess of 500 meters was obtained.

This results in a reliable communication with low BER, even in hostile e.m. environments. The energy required for the operation of the unit is provided by a 80 Ah primary source and by a photovoltaic panel equipped with a smart voltage regulator. Owing to a careful lowpower design, the unit could be powered with a small (20 W) photovoltaic panel for undiscontinued and unattended operation.

A picture of one of the SN unit installed at the Mantova plant is represented in Fig. 6, left. The battery and photovoltaic panel are clearly visible; the GPRS unit is the grey box close to

Real-Time Monitoring of Volatile Organic Compounds in Hazardous Sites 233

The communication between EN unit and VOC detector board is based on a RS485 serial interface, providing high immunity to interference and bidirectional communication

Thanks to the efficient communication protocols and effective power management strategies, the EN unit has a battery life on some two months of continuous VOC detector operation at 1 minute transmission data-rate, only relaying on primary energy resources. The technologies described above allow for the implementation of monitoring procedures in different ways, namely real-time sampling, continuous or discontinuous measurement, VOC analysis with specific concentration of single compounds, to name a few.The secondary energy source plays a key role in ensuring the stand-alone and unattended operation of the sensor network infrastructure. The photovoltaic power supply unit includes a charge

capability, as required for remote configuration/re-configuration of the unit.

Fig. 7. Block diagram of the End Node Unit

Fig. 8. Energy balance of the photovoltaic subsystem

the photovoltaic panel, and the WI is the white box on the top. The wind sensor and the RHT sensor with the solar shield are also visible. A concrete plinth serves as base for the unit, thus avoiding the need of excavations, which could be troublesome in the context of the plant due to pollution and contamination issues.

A picture of an EN unit is represented in Fig. 6, right. The photovoltaic panel along with the power supply and sensor board units are visible in the middle, while the VOC detector unit, protected by a metallic enclosure, is visible at the bottom. Also in this case a concrete plinth serves as the base for the unit.

Fig. 6. SN (left) and EN (right) units installed in proximity of the pipeline and of the chemical plant

#### **8.2 The EN unit**

The block diagram of the EN is represented in Fig. 7; it consists of a WI, similar to that previously described, and includes a VOC sensor board and a VOC detector. The WI unit is visible on the pole-top. Additionally, that solution allows wired connectivity of multiple VOC unit to the same EN, thus increasing modularity and flexibility of the architecture. The acquisition/communication subsystem of the EN unit is based on an ARM Cortex-M3 32 bit micro-controller, operating at 72 MHz, which provides the required computational capability compatible with the limited power budget available.

To reduce the power requirement of the overall subsystem, two different power supplies have been implemented, one for the micro-controller and one for the peripheral units; accordingly, the microcontroller is able to connect/disconnect the peripheral units, thus preserving the local energy resources. The VOC detector subsystem, in particular, is powered by a dedicated switching voltage regulator; this provides a very stable and spikefree energy source, as required for proper operation of the VOC detector itself.

the photovoltaic panel, and the WI is the white box on the top. The wind sensor and the RHT sensor with the solar shield are also visible. A concrete plinth serves as base for the unit, thus avoiding the need of excavations, which could be troublesome in the context of

A picture of an EN unit is represented in Fig. 6, right. The photovoltaic panel along with the power supply and sensor board units are visible in the middle, while the VOC detector unit, protected by a metallic enclosure, is visible at the bottom. Also in this case a concrete plinth

The block diagram of the EN is represented in Fig. 7; it consists of a WI, similar to that previously described, and includes a VOC sensor board and a VOC detector. The WI unit is visible on the pole-top. Additionally, that solution allows wired connectivity of multiple VOC unit to the same EN, thus increasing modularity and flexibility of the architecture. The acquisition/communication subsystem of the EN unit is based on an ARM Cortex-M3 32 bit micro-controller, operating at 72 MHz, which provides the required computational

To reduce the power requirement of the overall subsystem, two different power supplies have been implemented, one for the micro-controller and one for the peripheral units; accordingly, the microcontroller is able to connect/disconnect the peripheral units, thus preserving the local energy resources. The VOC detector subsystem, in particular, is powered by a dedicated switching voltage regulator; this provides a very stable and spike-

free energy source, as required for proper operation of the VOC detector itself.

Fig. 6. SN (left) and EN (right) units installed in proximity of the pipeline and of the

capability compatible with the limited power budget available.

the plant due to pollution and contamination issues.

serves as the base for the unit.

chemical plant

**8.2 The EN unit** 

Fig. 7. Block diagram of the End Node Unit

The communication between EN unit and VOC detector board is based on a RS485 serial interface, providing high immunity to interference and bidirectional communication capability, as required for remote configuration/re-configuration of the unit.

Fig. 8. Energy balance of the photovoltaic subsystem

Thanks to the efficient communication protocols and effective power management strategies, the EN unit has a battery life on some two months of continuous VOC detector operation at 1 minute transmission data-rate, only relaying on primary energy resources. The technologies described above allow for the implementation of monitoring procedures in different ways, namely real-time sampling, continuous or discontinuous measurement, VOC analysis with specific concentration of single compounds, to name a few.The secondary energy source plays a key role in ensuring the stand-alone and unattended operation of the sensor network infrastructure. The photovoltaic power supply unit includes a charge

Real-Time Monitoring of Volatile Organic Compounds in Hazardous Sites 235

ppb and after a certain time of power-off, the detector requires a stabilisation time of several

As for the calibration issue, a linearisation procedure was developed based on a behavioural model of the PID2; accordingly, the voltage read-outs received by the detector, *Vn*, are prior preprocessed by multiplying with a non-linearity compensation factor, *α(C)*, function of the

*n Cv S n )V n = ( C cn V*

where *Vcn* is the read-out corrected by the non-linearity compensation factor *α*, *Cn* is the concentration in ppm and *Vn* is the *nth* read-out in mV, and *Sv* is the PID sensitivity in mV/ppm. Equation (1) shows that, after compensation, the values *Vcn* can be easily mapped

In Fig. 9 and 10 the linearised calibration curves in the range 0-500 ppb are presented for two different PIDs. Fig. 9 represents the experimental calibration curve (read-out vs concentration) of a PID with a relatively high sensitivity, 150 mV/ppm. The non-linearity in

Fig. 9. Calibration curves for a PID with high sensitivity before (blue) and after (red)

The result of the linearisation process, according to the previously outlined procedure, is represented by the red line. Fig. 10 represents the same as Fig. 9 for a PID with relatively low sensitivity (50mV/ppm). In both cases, the linearisation procedure proved to be effective. The main advantage of the described approach is that for performing the PID calibration, one single parameter is needed, i.e. the value of the PID sensitivity, which is measured at ppm concentrations; this makes much simpler and less costly the calibration

(1)

minutes, thus preventing from operating it at minutes duty-cycles.

concentration *C:* 

linearisation

process.

2 GF Manes, unpublished results

in the corresponding concentration value.

the range 0-200 ppb is clearly observed, blue line.

regulator which was specifically designed to provide maximum energy transfer efficiency from the panel to the battery under any operative condition. In Fig. 8 upper left, the weekly graph of the power absorbed/generated by the photovoltaic power supply is represented; the blue line represents the positive balance, i.e. the panel is charging the battery, while the red line represents the negative balance, i.e. the primary source is supplying energy to the subsystem. In Fig. 8, bottom left, a comparison between the current generated by the system and the solar radiation under very clean daylight condition is presented; the right sheet represents the energy budget statistics generated by the system for one of SN unit. In Fig. 8 right, a summary of the daily, weekly and monthly energy balance is represented; more detailed analysis and diagnostics are available.
