Development of a Wireless Sensor Network for Monitoring a ...

2 downloads 7407 Views 2MB Size Report
The design considerations, hardware and software solutions, and samples of ... (LAN). A networked computer serves as the central monitoring station and ...
Presented at GeoCongress 2006, Feb. 26 - March 1, 2006, Atlanta, GA.

Development of a Wireless Sensor Network for Monitoring a Bioreactor Landfill Asis Nasipuri,1 Kalpathi R. Subramanian,2 Vincent Ogunro,3,* John L. Daniels,3 and Helene A. Hilger,3 1

Electrical and Computer Engineering, The University of North Carolina at Charlotte, 9201 University City, Blvd., Charlotte, NC 28223-0001 2 Computer Science, The University of North Carolina at Charlotte, 9201 University City, Blvd., Charlotte, NC 28223-0001 3 Civil Engineering, The University of North Carolina at Charlotte, 9201 University City, Blvd., Charlotte, NC 28223-0001 * Corresponding author; PH (704) 687-3101; FAX (704) 687 6953; email: [email protected] Abstract Recent studies of aerobic bioreactors have demonstrated their success in expediting stabilization of municipal solid waste (MSW), reducing or eliminating treatment and disposal costs of leachate, and increasing landfill capacity. Such aerobic decomposition is highly dependent on maintaining optimum distribution of moisture and air throughout the highly heterogeneous waste for the duration of the stabilization process. This requires distributed monitoring of the temperature and moisture in the bioreactor. This work presents the development and implementation of an autonomous monitoring system using an array of wireless sensors (motes). Each mote is equipped with embedded microprocessor, flash memory, and a wireless transceiver. Networked data collection along with 3D interactive visualization tools are developed for efficient assessment of the conditions in the bioreactor. Introduction Management of municipal solid waste is an important environmental issue in most cities and municipalities. One of the main management options of municipal solid waste involves subsurface disposal in specially engineered containment structures i.e. landfilling. The goal of a conventional landfill (typically referred to as “dry-tomb”) is to isolate the waste from the environment by preventing the release of leachate from and/or minimizing infiltration of external moisture into the system. However, stabilization of waste within such landfills takes a long time. Therefore, these waste containments system, being constructed facilities, degrade with age and become less effective barrier system in the long-term. An alternative approach is the acceleration of the biodegradation process by creating conditions favorable for microorganisms that are responsible for waste decomposition in landfills (bioreactor concept). Currently, there is a lot of interest in designing “bioreactors – anaerobic and/or aerobic”, where optimal conditions must be maintained artificially by controlled injection and extraction of moisture and/or air into and from the landfill. Bioreactor landfills, particularly aerobic, can save valuable time and space as well as reduce the potential threat of groundwater pollution from traditional landfills (Stessel and Murphy, 1992; Stessel and Bernreuter, 2001). Although such accelerated biodegradation of solid waste have been confirmed in laboratory and field settings, the process is highly dependent on accurate monitoring and optimal control of the distribution of moisture and air in the heterogeneous mixture. In addition, the extent of this control has direct bearing on safety considerations associated with hazards such as landfill fires and the failure of the containment structure due to excessive unit weight of the material (slope failures) (Isenberg, et al., 2001; Kavazanjian, et al., 2001; Reinhart and Townsend, 2001). In few cases, landfills are monitored remotely by an array of sensors that are hardwired to a SCADA (supervisory control and data acquisition) system operated by a system control center. The system control center of the SCADA system of a bioreactor would be required to perform all the supervisory functions in a centralized fashion. Since a typical bioreactor would require hundreds of sensors distributed over several acres operating over extended periods (about 12 - 24 months), such systems involve huge wiring costs and careful planning for laying out the infrastructure of wired interconnections. In addition, the centralized nature of such a system makes it inefficient, burdening the control center with a large amount of data processing tasks. Modern wireless networking technology can greatly simplify the design of a reliable communications infrastructure for the distributed monitoring and control functions required in a bioreactor. Recent advances in embedded systems and communication technology have led to the development of small inexpensive wireless devices that also have limited processing as well as ad hoc networking capabilities. Each device has an embedded

1

microprocessor, flash memory, and a wireless transceiver, which can serve as a “smart” wireless interface to a sensor or an actuator, forming a wireless network node. The small size and low cost of these devices allow them to be used in large numbers with negligible installation costs. Although the communication range of each node is small, the capability of forming an autonomous network with other neighboring nodes using multi-hop ad hoc networking principles allows a system of such devices to perform powerful distributed monitoring and control operations (Estrin, et al., 1999; Intanagonwiwat, et al., 2000; Collins, et al., 2001). Such devices can be used to perform local processing of specific sensor observations, enable collaborative processing amongst a group of sensors, or form a hierarchically structured system involving many nodes and a control center to meet desired objectives (Bulusu, et al., 2000). This paper presents the development and implementation of an autonomous monitoring system using an array of wireless sensors that is applied for monitoring a laboratory meso-scale bioreactor testbed. The system utilizes networked data collection techniques along with 3D interactive visualization tools for efficient assessment of the conditions in the bioreactor. The design considerations, hardware and software solutions, and samples of monitored data are presented. Bioreactor Testbed and Monitoring Objectives A 10’×6’×5’ bioreactor was designed within a geoenvironmental laboratory at UNC Charlotte with the primary objective of deriving models characterizing the effects of the flow if air and moisture through municipal waste. Such a meso-scale bioreactor also provided an excellent testbed to design, test, and evaluate the performance of a wireless sensor-based distributed monitoring system. The bioreactor was loaded in July of 2004 with a mix of municipal solid waste (> 1600 kg) that was carefully prepared from waste steel and aluminum cans, glass, cardboard, newspaper, plastic bottles, and paperboard. The mix proportions were those of the average U.S. MSW composition reported by the U.S. EPA. An elevation schematic of the system is shown in Figure 1. The vertical injection and extraction system is comprised of schedule 40 slotted PVC, and it is bolted through the interior of the bioreactor. Airflow is injected and extracted with 1.5 horsepower blowers that are connected to the PVC with quick-connect couplings. The leachate collection and re-circulation system also consists of schedule 40 slotted PVC and is located at the low point of the bioreactor floor. The leachate is collected and re-circulated into the waste through the injection well. An emergency leachate injection system was also constructed in the floating cover so that the leachate can be gravity fed if needed.

Figure 1. Overall process of the meso-scale bioreactor testbed (after Daniels et al., 2005)

2

The bioreactor was operated under manual control. The primary objectives of this experiment include (a) obtaining the relationships between liquid content, air content, and temperature; (b) developing empirical models for waste settlement; (c) determining the effect of leachate recirculation on stabilizing the waste; obtaining criteria for determining stabilization; and finding the optimum conditions for the aerobic-anaerobic bioreactor. All these objectives were to be derived from analyzing the variation of temperature and moisture obtained from a network of sensors embedded in the bioreactor. In order to facilitate such observations, the sensor network was required to meet the following requirements: • Provide accurate samples of temperature and moisture probe readings, taken at appropriate (unknown) intervals of time • Store and disseminate the data in real time, so that researchers can utilize sensor readings to manually control the bioreactor • Provide a convenient graphical interface for easy interpretation of data obtained from a large set of sensors that are distributed in three dimensions within the bioreactor • Enable such monitoring and display operations for an extended period of time (> 6 months) Wireless Monitoring System In order to achieve these objectives while safeguarding all experimental observations, a two-pronged approach was taken for designing the monitoring the monitoring system. Half of bioreactor was equipped with temperature and moisture probes that were wired to a traditional electronic monitoring system, whereas the other half was equipped with sensor probes that were connected to the wireless monitoring system. The duplication was done to remove dependency to any one of the data collection systems, thus reducing the possibility of losing data loss due to a failure in either system. Network Architecture and Hardware A schematic of the wireless monitoring system is depicted in Figure 2. The locations of these sensors are given in Table 1. There is one temperature sensor and one humidity probe at each location point. Hence, the wireless monitoring system has a total of 12 temperature probes and 12 moisture probes distributed in three layers (geogrids). All sensors are wired to a nearby wireless node (mote) that is equipped with data acquisition board. The motes are programmed to periodically sample the sensor observations and wirelessly transmit the data to a wireless-to-Ethernet interface board located in the laboratory. The interface board provides connectivity to the campus local area network (LAN). A networked computer serves as the central monitoring station and database, which runs an application program to receive all data from the interface board and save them in appropriate data files. Since multiple sensors may be connected to each mote, the motes multiplex the data from each input (sensor) and transmit the collective information together to the interface boards. Appropriate demultiplexing is performed at the monitoring station to identify the sources of various data streams. The wireless components used are second generation experimental embedded devices manufactured by Crossbow with technical collaboration with UC Berkeley. Each mote is equipped with an AtMega128L microcontroller, program memory, 10-bit analog to digital converter (ADC), and 433 MHz ISM band radio. The transmission range of the radio transceiver at the motes is between 100 – 200 feet using a 6.8 inch whip antenna. The motes use a specially designed open-source software platform called TinyOS that provides a convenient platform for developing applications for performing various sensor signal processing and wireless communications/networking. A complete list of hardware components used in the system is given in Table 2. Software Design Issues The key challenges for designing the system include appropriate considerations for conserving battery power at the motes, appropriate signal conditioning and calibration of the sensor probe measurements, and development of graphical visualization software for monitoring. The motes are powered by AA size batteries, which can be depleted within a few days if the sensing, processing, and communication tasks are not conserved. The motes draw approximately 20 mA or current for transmission, 5 mA for receiving, 1 mA during idle/processing periods, and 20 µA during sleep periods. Hence, for extending the life of the batteries, it is desirable to reduce the frequency of transmissions from the motes. This was achieved by maintaining a sampling interval of 45 minutes at all sensors and

3

applying adequate power management schemes that shuts down key processes running in the motes between successive sampling times. The central monitoring station was designed to run an application program to perform the following tasks: • Demultiplex the data from different sensors, assign time stamps, convert mV measurements obtained from the motes into corresponding temperature or humidity units, and save them in separate files. • Host a java applet program for 3D visualization of the temperature and humidity measurements obtained in the bioreactor. This allows users to browse through the data files and obtain a comprehensive view of the sensor measurements.

423

223 313

113

422

222 312

112

LA

421

221 311

111

LAN Temperature sensor Humidity sensor Mote with Data Acquisition board Wireless-to-Ethernet interface board

Figure 2. Schematic diagram of the wireless monitoring system Table 1: Sensor locations Bottom geogrid ID x y 111 1'2" 1'3" 221 3'10" 2'4" 311 6'7" 1'6" 421 9'1" 2'1"

z 9" 9" 9" 9"

ID 111 221 311 421

Middle geogrid x y 1'2" 1'3" 3'10" 2'4" 6'7" 1'6" 9'1" 2'1"

Table 2: List of hardware components used Description Temperature probe Humidity probe Wireless processor/radio node (mote) Data acquisition board Wireless-to-Ethernet Interface

z 9" 9" 9" 9"

Hardware 108 probe Echo20 MPR410 MDA300 MIB510

ID 111 221 311 421

Top geogrid x y 1'2" 1'3" 3'10" 2'4" 6'7" 1'6" 9'1" 2'1"

z 9" 9" 9" 9"

Manufacturer Campbell Scientific Decagon Crossbow Crossbow Crossbow

Performance The wireless monitoring system performed consistently and accurately. The main concern was battery life, which was limited to approximately two months for each mote. In order for the wireless system to be used in a field application,

4

the batteries must last for 12 months or more. This can be achieved by incorporating energy-efficient medium access control protocols that utilize sleep modes in the sensors to conserve electrical power. Temperature readings obtained at sensors 221, 222, and 223 for weeks starting 10/20/2004 and 11/05/2004 are plotted in Figure 3. The rapid fluctuations of temperature are due to finite precision error caused by the ADC. The plots indicate the difficulty in assessing the condition inside the bioreactor by observing variations of signals over time from a few sensors. In Figure 4 we show snapshots taken from the 3d visualization tool that uses volume rendering to construct and update the surface depicting the boundary where the temperature crosses 100°F. The volume rendering was performed from the sensor data using Virtual Toolkit (VTK) software. The snapshots show how the three-dimensional region where the temperature exceeds 100°F varies with time. A java applet to perform the same operation was also designed that can be run online from any computer on the campus network by connecting to the monitoring station using a web browser. 140 130

Sensor 221 Sensor 222 Sensor 223

130

120

120

110

110 Degree F

Degree F

140

Sensor 221 Sensor 222 Sensor 223

100

100

90

90

80

80

70

70 60

60 0

24

48

72

96

120

144

0

168

24

48

72

96

120

144

168

Hours from 11/05/04 1600 hrs

Hours from 10/20/04 2000 hrs

Figure 3. Temperature observed at three centralized sensors at three levels over one week periods.

(a) 0 hours

(b) 6 hours

(c) 12 hours

(d) 18 hours

(e) 24 hours

(f) 30 hours

Figure 4: Snapshots from an online video that shows the variation of a surface enclosing the region within which temperature>102°F. The video was constructed using 3d graphics utilities from sensor data for the period starting 7.00pm, 12/08/2004.

5

Conclusions The distributed monitoring of temperature within the bioreactor system was highly successful. The wireless sensor network and devices performed robustly. Although the wireless system performed reliably, the duration of data collection and transmission is limited by the short battery life of approximately two months for each mote. A new energy-efficient medium access protocol that incorporates sleep modes in the sensors to conserve electrical power is recommended. A java applet to run three-dimensional boundary growth of temperature contours over time was designed and implemented to run online from any computer on the campus network. The data from the wireless system could be interrogated remotely over the internet. Although moisture data were also acquired, the results were inconclusive. Acknowledgements The authors wish to acknowledge the invaluable help received from our students Brian Toothman, Jeremy Yager, Jerry Zacharias, and John Filipi from Electrical and Computer Engineering Department and Jessica Montgomery, Marie Schmader, Chris Stahl and Chris Friel from Civil and Environmental Engineering Department. This work would not be possible without their help in the laboratory. This work was partially supported by funding received from NSF grant EIA-0130799, a Junior Faculty Grant received from UNC Charlotte, a Major Grant received from College of Engineering, UNC-Charlotte, and supports from Civil Engineering Dept. and Global Institute for Energy and Environmental Systems (GIEES), UNC-Charlotte and S&ME of Charlotte, NC. References Bulusu, N., Heidemann, J., and Estrin, D. (2000). “GPS-less low cost outdoor localization for very small devices.” IEEE Personal Communications Magazine, pp. 28–34. Collins, R.T., Lipton, A.J., Fujiyoshi, H., and Kanade, T. (2001) “Algorithms for cooperative multisensor surveillance.” Proceedings of the IEEE, 89(10), pp. 1456–1477. Daniels, J.L., Schmader, M.B., Ogunro, V.O., and Hilger, H.H. (2005). “Laboratory-scale aerobic landfill bioreactor: A precursor to modeling and full-scale investigation.” Conference proceeding of Geo-Frontier 2005, Austin, TX: GSP 142 Waste containment and remediation, K. Reddy, ed., Geo-Institute, ASCE, CD-ROM. Estrin, D., Govindan, R., Heidemann, J., and Kumar, S. (1999). “Next century challenges: Scalable coordination in sensor networks.” In ACM/IEEE International Conference on Mobile Computing and Networking (MOBICOM), pp. 263–270. Intanagonwiwat, C., Govindan, R., and Estrin, D. (2000). “Directed diffusion: A scalable and robust communication paradigm for sensor networks.” In ACM/IEEE International Conference on Mobile Computing and Networking (MOBICOM), pp. 56–67. Isenberg, R.H., Law, J.H., O’Neil, J.H., and Denver, R.J. (2001). “Geotechnical aspects of landfill bioreactor design: is stability a fatal flaw?” In Proceedings of the 6th Annual Landfill Symposium, pp. 51–62. Kavazanjian, E., Henderson, D.M., and Corcoran, G.T. (2001). “Strength and stability of bioreactor landfills.” In Proceedings of the 6th Annual Landfill Symposium, pp. 63–70. Reinhart, D.R., and Townsend, T. (2001). “Aerobic vs. anaerobic bioreactor landfill case study - the new river regional landfill.” In Proceedings of the 6th Annual Landfill Symposium, pp. 5–15. Stessel, R.I., and Bernreuter, J. C. (2001). “A review of aerobic biocell research and technology.” In Proceedings of the 6th Annual Landfill Symposium, pp. 99–114. Stessel, R.I., and Murphy, R.J. (1992). “A lysimeter study of the aerobic biocell research and technology.” Waste Management and Research, 10, pp. 485–503.

6