**10. Experimental results**

236 Environmental Monitoring

As for the stabilisation time, several experiments were performed to qualify the PID performance; it was found that at low concentration (tens or hundreds ppb), which represents the area of operation of the VOC detectors in our application and when operated in the diffusion mode, the PID exhibits a stabilisation time of some minutes after a poweroff/power-on cycle. A typical PID duty cycled response after storage is represented in Fig. 11. The experimental stabilisation curve is compared with a 80 s decay-time exponential function showing an excellent fitting. After a warm-up of several hours the PID was powered-off for 15 minutes and then powered-on again; thie sequence simulated a 15 minute sampling interval, which was the initial target of our application; in this experiment ambient concentration was around 50 ppb, which represents the average concentration

Fig. 10.Calibration curves for a PID with low sensitivity before (blue) and after (red)

As observed in Fig. 11, a 300 seconds stabilisation time is needed prior the PID can reach a stable read-out value. This experiment shows that a 15 minutes sampling interval calls for a 5 minutes stabilisation time, thus resulting in some 30% duty-cycle. A duty-cycled operation, as compared with a continuous power-on operation, is desirable in principle to prolong both the battery- and lamp-life; however, the benefit of energy saving allowed for by the 30% duty cycle is marginal, when compared with the advantage of achieving a more time-intensive monitoring of VOC concentration, as provided by continuous power-on operation. In terms of energy resources, continuous power-on operation requires some 35 mAh charge, which corresponds to 1 month of full operation with a 30 Ah primary energy source; the corresponding power consumption of 360 mW@12 Vdc can be balanced using a 5

The UV lamp expected life is more than 6000 hours of continuous operation; we expect at least a quarterly service for the PIDs, due to environment contamination and related lamp

where the PID is supposed to be set up.

linearisation

W photovoltaic panel.

Data from the field are forwarded to a central database for data storage and data rendering. A rich and proactive user interface was implemented, in order to provide detailed graphical data analysis and presentation of the relevant parameters, both in graphical and bidimensional format. Data from the individual sensors deployed on the field can be directly accessed and presented in various formats by addressing the appropriate sensor(s) displayed on the plant map, see Fig 12 left.

The position of each SN and EN unit is displayed on the map; by positioning the mouse pointer over the corresponding icon, a window opens showing a summary of current parameter values.

A summary of the sensor status for each deployed unit can be obtained by opening the summary panel, Fig. 12, right. The summary panel reports current air temperature/humidity values, along with min/max values of the day (left lower, in Fig. 12), wind speed and direction (left upper, in Fig. 12), and VOC concentration (right, in Fig. 12), in the last six hours. A graphic representation of data gathered by each sensor on-the field can be obtained by opening the graphic panel window, see Fig. 13.

The graphic panel allows anyone to display the stored data in any arbitrary time interval in graphic format; up to six different and arbitrarily selected sensors can be represented in the same graphic window for purpose of analysis and comparison.

Real-Time Monitoring of Volatile Organic Compounds in Hazardous Sites 239

The effect of a sudden wind speed increase, light green line, is shown on the right graph of Fig. 14 right. It can be observed a wind speed increases to some 5m/s and more, green line, around 10 pm; accordingly, the VOC concentration detected by the three PIDs deployed in the plant is suddenly decreased. It should be noted that the three PIDs are located several

Fig. 15. Multi-trace read-outs of the six VOC sensors deployed around the ST40 plant

concentration levels demonstrating the effectiveness of the calibration procedure.

obtained for each of the sensor represented in the graphic window, lower right.

recording, there were some maintenance works going on in the plant's area.

In Fig. 15, the read-outs of the 6 VOC sensors deployed around the ST40 plant are represented; it should be noted the very good uniformity among the background

The user interface can perform various statistics on the data items; in the graphic panel, the user can enter the inspection mode, see the button on the lower right in Fig. 16, and set an user defined inspection window (in white); the window can be set over an arbitrary time interval; parameters like max/min, arithmetic mean and maximum variation can be then

The sensitivity of the PID sensor is demonstrated in Fig. 17, where the traces of two different PIDs are shown. The PIDs are located some 500 meters far apart. At the time of data

The VOC components due to maintenance works were detected by the PIDs and recorded as small variation of the concentration around the mean value during the working hours (from 8 am to 6 pm, roughly), to be compared with the more smoothed traces recorded during the night. A diagnostic panel is available to evaluate the system Quality of service (QoS) and the gathered data reliability, see Fig. 18; connectivity statistics are displayed along with the

Fig. 14. Correlation between wind speed and VOC concentration

hundred meters far apart each other.

Fig. 12. Plant lay-out and details of the sensors

In Fig. 13 left, the VOC concentration traces of three different detectors are represented in a period of one day; in Fig. 13 right, the same data are displayed in a period of 30 days. By using the pointer, it is possible to select a time sub-interval and to obtain the corresponding graphic representation at high resolution.

Fig. 13. Representation of sensor data in graphic format

In Fig. 13 left, the VOC concentration background is around 50 ppb; thanks to the very intensive sample-interval, 1 minute, the evolution of the concentration in time, along with other relevant meteo-climatic parameters can be very accurately displayed; it should be noted that the spikes which can be observed in the blue trace, Fig. 13 left, have a duration of some 3 minutes. The multi-trace graphic feature is very useful to perform correlation between different parameters. In Fig. 14 two examples of correlation between WSD and VOC concentration are shown. In Fig. 14 left, the VOC concentration, green line, exhibits a night/day variation; this is compared with the wind speed, rosé line, which increases during the day hours and decreases during the night hours, very likely due to the thermal activity. As it can be observed, in fact, wind speed and VOC concentration are in phase opposition, i.e. the greater the wind speed, the lower the average VOC concentration in the plant, that is in good agreement with what one can expect.

In Fig. 13 left, the VOC concentration traces of three different detectors are represented in a period of one day; in Fig. 13 right, the same data are displayed in a period of 30 days. By using the pointer, it is possible to select a time sub-interval and to obtain the corresponding

In Fig. 13 left, the VOC concentration background is around 50 ppb; thanks to the very intensive sample-interval, 1 minute, the evolution of the concentration in time, along with other relevant meteo-climatic parameters can be very accurately displayed; it should be noted that the spikes which can be observed in the blue trace, Fig. 13 left, have a duration of some 3 minutes. The multi-trace graphic feature is very useful to perform correlation between different parameters. In Fig. 14 two examples of correlation between WSD and VOC concentration are shown. In Fig. 14 left, the VOC concentration, green line, exhibits a night/day variation; this is compared with the wind speed, rosé line, which increases during the day hours and decreases during the night hours, very likely due to the thermal activity. As it can be observed, in fact, wind speed and VOC concentration are in phase opposition, i.e. the greater the wind speed, the lower the average VOC concentration in the

Fig. 12. Plant lay-out and details of the sensors

graphic representation at high resolution.

Fig. 13. Representation of sensor data in graphic format

plant, that is in good agreement with what one can expect.

Fig. 14. Correlation between wind speed and VOC concentration

The effect of a sudden wind speed increase, light green line, is shown on the right graph of Fig. 14 right. It can be observed a wind speed increases to some 5m/s and more, green line, around 10 pm; accordingly, the VOC concentration detected by the three PIDs deployed in the plant is suddenly decreased. It should be noted that the three PIDs are located several hundred meters far apart each other.

Fig. 15. Multi-trace read-outs of the six VOC sensors deployed around the ST40 plant

In Fig. 15, the read-outs of the 6 VOC sensors deployed around the ST40 plant are represented; it should be noted the very good uniformity among the background concentration levels demonstrating the effectiveness of the calibration procedure.

The user interface can perform various statistics on the data items; in the graphic panel, the user can enter the inspection mode, see the button on the lower right in Fig. 16, and set an user defined inspection window (in white); the window can be set over an arbitrary time interval; parameters like max/min, arithmetic mean and maximum variation can be then obtained for each of the sensor represented in the graphic window, lower right.

The sensitivity of the PID sensor is demonstrated in Fig. 17, where the traces of two different PIDs are shown. The PIDs are located some 500 meters far apart. At the time of data recording, there were some maintenance works going on in the plant's area.

The VOC components due to maintenance works were detected by the PIDs and recorded as small variation of the concentration around the mean value during the working hours (from 8 am to 6 pm, roughly), to be compared with the more smoothed traces recorded during the night. A diagnostic panel is available to evaluate the system Quality of service (QoS) and the gathered data reliability, see Fig. 18; connectivity statistics are displayed along with the

Real-Time Monitoring of Volatile Organic Compounds in Hazardous Sites 241

the overall GPRS efficiency figure. EN unit status and connectivity are displayed in the

The diagnostic panel identifies any lack of connectivity and/or reliability of each single SN

In addition to the graphic format, data items can be represented in a bi-dimensional format. It is quite difficult to correlate the data in graphic format from different sensors deployed over the plant; a helpful bi-dimensional picture of the area based on an interpolation of algorithms has been implemented, resulting in a very synthetic representation of the parameters of interest over the plant in pseudo-colours. The sensors are basically punctual and, thus, are only representative of the area in their proximity. For that reason the interpolation would be only effective if an adequate number of sensors is deployed on the field, so that the area is

This requirement would result in an unnecessarily high number of units to be deployed. A more effective approach is to take into account the morphology and functionality of the

As for the VOC, by instance, the potential sources of VOC emissions in the plant are located in well identified areas like, the chemical plant and the benzene tanks; accordingly, the deployment strategy includes a number (6) of VOC sensors surrounding the chemical plant infrastructure, thus resulting in a virtual fence, capable of effectively evaluating VOC

As for wind speed and direction, which are relevant for correlation with VOC concentration, on the basis of an evaluation of the plant infrastructures, the areas of potential turbulence were identified and the wind sensors were deployed accordingly. Both SN and EN units were equipped with RHT sensors, whose cost is marginal. In Fig. 19 two bidimensional pictures of the temperature (left) and RH (right) in the area of the plant are represented. Not surprisingly, both temperature and RH are not uniformly distributed; according to the colour scale of air temperature blue means lower temperature and red means higher temperature; in this case the temperature ranges from 28°C (blue) to 31°C (red). Two areas of higher temperature are clearly identified, one on the left around the chemical plant ST40

subdivided into elementary cells, *quasi- homogeneous* in terms of the parameter values.

different areas of the plant and deploy the sensors accordingly.

emissions on the basis of the concentration pattern around the plant itself.

columns 4 and 9 from left, while power supply status is showed in column 5 from left.

or EN unit for immediate service action.

Fig. 18. The diagnostic panel

current status of connectivity for each of the SN and EN units. The status of the GPRS connectivity and the related statistics are represented in column 3 and 6 from left, respectively.

Fig. 16. Statistical parameters analysis

Fig. 17. Day/night VOC read-outs

As it can be observed, GPRS connectivity in excess of 99% is obtained, because of the periodic restart of the SN unites which do not get connected for a short time interval, and thus reducing the overall GPRS efficiency figure. EN unit status and connectivity are displayed in the columns 4 and 9 from left, while power supply status is showed in column 5 from left.

The diagnostic panel identifies any lack of connectivity and/or reliability of each single SN or EN unit for immediate service action.


Fig. 18. The diagnostic panel

240 Environmental Monitoring

current status of connectivity for each of the SN and EN units. The status of the GPRS connectivity and the related statistics are represented in column 3 and 6 from left,

As it can be observed, GPRS connectivity in excess of 99% is obtained, because of the periodic restart of the SN unites which do not get connected for a short time interval, and thus reducing

respectively.

Fig. 16. Statistical parameters analysis

Fig. 17. Day/night VOC read-outs

In addition to the graphic format, data items can be represented in a bi-dimensional format. It is quite difficult to correlate the data in graphic format from different sensors deployed over the plant; a helpful bi-dimensional picture of the area based on an interpolation of algorithms has been implemented, resulting in a very synthetic representation of the parameters of interest over the plant in pseudo-colours. The sensors are basically punctual and, thus, are only representative of the area in their proximity. For that reason the interpolation would be only effective if an adequate number of sensors is deployed on the field, so that the area is subdivided into elementary cells, *quasi- homogeneous* in terms of the parameter values.

This requirement would result in an unnecessarily high number of units to be deployed. A more effective approach is to take into account the morphology and functionality of the different areas of the plant and deploy the sensors accordingly.

As for the VOC, by instance, the potential sources of VOC emissions in the plant are located in well identified areas like, the chemical plant and the benzene tanks; accordingly, the deployment strategy includes a number (6) of VOC sensors surrounding the chemical plant infrastructure, thus resulting in a virtual fence, capable of effectively evaluating VOC emissions on the basis of the concentration pattern around the plant itself.

As for wind speed and direction, which are relevant for correlation with VOC concentration, on the basis of an evaluation of the plant infrastructures, the areas of potential turbulence were identified and the wind sensors were deployed accordingly. Both SN and EN units were equipped with RHT sensors, whose cost is marginal. In Fig. 19 two bidimensional pictures of the temperature (left) and RH (right) in the area of the plant are represented.

Not surprisingly, both temperature and RH are not uniformly distributed; according to the colour scale of air temperature blue means lower temperature and red means higher temperature; in this case the temperature ranges from 28°C (blue) to 31°C (red). Two areas of higher temperature are clearly identified, one on the left around the chemical plant ST40

Real-Time Monitoring of Volatile Organic Compounds in Hazardous Sites 243

As it can be noted, wind direction represented by blue arrows is far by being uniform over the plant, thus denoting turbulences due to the plant infrastructures and surrounding

An end-to-end distributed monitoring system integrating VOC detectors, capable of performing real-time analysis of gas concentration in hazardous sites at unprecedented

The aim was to provide the industrial site with a flexible and cost-effective monitoring tool, in order to achieve a better management of emergency situations, identify emission sources in real time, and collect continuous VOC concentration data using easily re-deployable and

The choice of collecting data at minute time interval reflects the need to identify short term critical events, quantify the emission impacts as a function of weather conditions and

The choice of a WSN communication platform gave excellent results, above all the possibility to re-deploy and re-scale the network configuration according to specific needs, while greatly reducing installation cost. Furthermore, to manage real-time data through a web based interface allowed both adequate level of control and quick data interpretation in

Among the various alternatives available on the market, the choice of PID technology proved to meet all the major requirements. PIDs are effective in terms of energy consumption, measuring range, cost and maintenance, once installed in the field. The installation of weather sensors at the nodes of the main network stations allowed for a better understanding of on-field phenomena and their evolution along with clearer identifcation of

Future activity will include a number of further developments, primarily the development of a standard application to allow the deployment of WSN in other network industries (e.g. refineries) and an assessment of potential applications for WSN infrastructure monitoring of

This work was supported by eni SpA under contract N.o 3500007596. The authors wish to thank W O Ho and A Burnley, Alphasense Ltd., for many helpful comments and clarifications concerning the PID operation, S Zampoli and G Cardinali, IMM CNR Bologna, for many discussions on PID characterisation and E Benvenuti, Netsens Srl, for his valuable

Assistance and support by the Management and technical Staff of Polimeri Europa Mantova

Adler R.; Buonadonna, P. Chhabra, J. Flanigan, M. Krishnamurthy, L. Kushalnagar, N.

Nachman, L. & Yarvis M. (2005). *Design and Deployment of Industrial Sensor Networks: Experiences from the North Sea and a Semiconductor Plant* in ACM SenSys,

time/space scale, has been implemented and successfully tested in an industrial site

vegetation.

**11. Conclusions** 

rationally distributed monitoring stations.

order to manage critical situations.

potential emission sources.

**12. Acknowledgement** 

technical support.

**13. References** 

other environmental indicators.

is also gratefully acknowledged.

November 2-4, 2005, San Diego, CA.

operational process, and identify critical areas of the plant.

and the other on the right around the arrival of the pipeline; this is obviously related to the mechanical activity in those areas. The thermal distribution also influences the air RH as demonstrated in Fig. 19, left. In this case the grey colour means lower RH and the blue colour means higher RH.

The RH values range from 26% to 33%, in this case. The temperature gradient among the different areas in the plant, which in some cases grew to up 5°C, is responsible of some thermal activity possibly affecting the VOC concentration distribution.

Fig. 19. Bi-dimensional map of air temperature (left) and air RH (right) distribution in the area of the plant

Fig. 20. Bi-dimensional map representing VOC concentration in the plant

VOC concentration is mapped in Fig. 20 in pseudo-colours. In this case blue denotes lower concentration, while red denotes higher concentration; it should be emphasized that the red colour has no reference with any risky or critical condition at all, beings only a chromatic option.

As it can be noted, wind direction represented by blue arrows is far by being uniform over the plant, thus denoting turbulences due to the plant infrastructures and surrounding vegetation.
