Landfill Gas Monitoring Network

Development of Wireless Sensor Network Platforms

Fiachra Collins, Dylan Orpen, Eoghan McNamara, Cormac Fay and Dermot Diamond

CLARITY Centre for Sensor Web Technologies, National Centre for Sensor Research, Dublin City University,

Glasnevin, Dublin 9, Ireland

Keywords: Environmental Monitoring, Gas Sensing, Landfill Gas.

Abstract: A wireless sensor network has been developed for the application of landfill gas monitoring, specifically

sensing methane, carbon dioxide and extraction pressure. This collaborative work with the Irish

Environmental Protection Agency has been motivated by the need to reduce greenhouse gas emissions as

well as aiming to improve landfill gas management and utilisation. This paper describes the preliminary

findings of an ongoing trial deployment of multiple sensing platforms on an active landfill facility; data has

been acquired for nine months to date. The platforms have operated successfully despite adverse on-site

conditions, with validity demonstrated by reasonably strong correlation with independent on-site

measurements. The increased temporal and spatial resolution provided by distributed sensor platforms is

discussed with regard to improving landfill gas management practice.

1 INTRODUCTION

Wireless sensor networks are gaining increasing

attention with regard to environmental monitoring;

the extended temporal and spatial resolution

provided by distributed sensing nodes enable a

wealth of environmental parameters to be measured.

The need for such environmental monitoring is

driven by international legislation targeting a

reduction in greenhouse gas emissions and

improving air quality (EU, 2008). Traditionally,

such parameters have been measured using fixed

single-point sampling equipment. For example, the

Irish Environmental Protection Agency (EPA)

operates 29 fixed stations around Ireland for the

purposes of monitoring ambient air quality (EPA,

2012). The emergence of medium and large scale

sensor networks has been evident in recent years; the

first urban network for monitoring carbon dioxide

was announced in Oakland, USA (Cohen et al.,

2012), while air quality sensor networks have been

deployed in urban centres in the UK (Envirowatch,

2012). However, while WSNs are moving to the

foreground of attention with respect to

environmental monitoring, they have yet to

proliferate to their full potential. The principle

obstacles to this proliferation are price-point and

technology dependability. The extrapolation of

current sensor and platform technology costs for

scaled-up deployments is, for the most part, not

economically viable. Furthermore, the deployment

of a distributed network of sensors presents a

challenge in terms of configuration, maintenance

and multiplication of technical issues.

The environmental monitoring discussed in this

paper is specifically with regard to landfill waste

facilities. Landfill gas is primarily composed of

carbon dioxide (CO

) and methane (CH

) generated

from the anaerobic decomposition of biodegradable

waste. Modern landfill facilities contain a network

of pipes connecting perforated wells sunk into the

waste body, where an applied negative pressure

extracts the gases from the waste cells. Landfill gas

poses significant environmental hazards, both in

terms of greenhouse gas contribution and local

hazards such as asphyxiation and combustion. This

was dramatically demonstrated in Ireland in early

2011 where a subterranean fire on a closed landfill

site incurred local controversy and costly

extinguishing and remediation measures.

It can be seen therefore that substantial attention

is required to effectively manage landfill gas and

indeed utilise it for electricity generation where

methane concentrations are sufficiently high. From

the viewpoint of landfill facility operators, there are

two aspects of monitoring landfill gas: firstly at

222

Collins F., Orpen D., McNamara E., Fay C. and Diamond D..

Landﬁll Gas Monitoring Network - Development of Wireless Sensor Network Platforms.

DOI: 10.5220/0004314502220228

In Proceedings of the 2nd International Conference on Sensor Networks (SENSORNETS-2013), pages 222-228

ISBN: 978-989-8565-45-7

 2013 SCITEPRESS (Science and Technology Publications, Lda.)

Figure 1: (i) Exploded view of platform (a) microcontroller (MSP430), (b) sensors, (c) IP68 casing, (d) battery, (e) GSM

module, (f) extraction pump. (ii) Configuration of wireless sensor network deployed on landfill site.

perimeter wells to ensure against gas migration

through the waste body into the surrounding soil

beyond the confines of the site; secondly in-line with

the extraction network to maintain an appropriate

gas composition for optimum engine/flare operation.

The former aspect of perimeter landfill gas

monitoring is what initially motivated this work,

resulting in autonomous monitoring platforms being

developed and validated (Beirne et al., 2010; Fay et

al., 2011; Collins et al., 2012). The latter aspect of

in-line monitoring has been the focus of more recent

efforts and is the subject of this present paper. Gas

management via the site’s extraction pipe network is

one of the crucial aspects of landfill operation in

order to comply with environmental legislation.

Given their hazardous nature, all landfill gases must

be thermally oxidised (i.e. burned) to mitigate

against greenhouse gas emissions and local

pollution. For methane concentrations in excess of

50% v/v, the gas can be used as fuel for a CHP

(combined heat and power) engine, presenting a

financial advantage in recuperating costs by

generating electricity and selling it to the national

grid; otherwise the gas must be burned in a flare.

The composition of the gas must be precisely

controlled for effective combustion, requiring the

adjustment of flow from different waste cells within

the landfill site - this operation is called ‘field

balancing’.

The development of wireless sensor networks is

ideally suited to the landfill application, where a

typical extraction system covers an expansive area

of ground. Moreover, this terrain by nature is

difficult to traverse, implying that remote sensors

deployed in-situ represent an advantage over the

current infrequent manual sampling routines. Gas

composition monitoring would indicate the gas

generation potential of different waste cells as well

as identifying fugitive gas emissions. Measurement

of the extraction pressure can be used as a diagnostic

tool to identify loss in flow due to blockages or

leakages. The distribution of autonomous sensing

platforms in a networked configuration would

enable more informed and precise field balancing

and gas management which, in turn, represents cost

savings by promoting electricity generation and

avoiding engine/flare downtime.

The paper addresses the research challenge of

attaining reliable and accurate sensor performance

without incurring prohibitive expense. Typically,

cost and performance are inherently linked – cheaper

components fail to deliver adequate resolution,

accuracy or in-calibration duration. Traditionally,

long-term reliability is achievable only at a

substantial cost. However, the price of electronics

and sensors is continuously reducing, driven by the

ever-growing market of consumer technologies and

optimised manufacturing techniques. The integration

of such technology into deployable platforms with

WSN capabilities can fulfil the demand in industry

for low-cost long-term sensing. However, market

acceptance is only achievable if the sensing

performance is validated and proven to work. This is

the motivation behind the platform technology

described in this paper, which represents a

progression of work on developing autonomous gas

monitoring platforms with web-based accessibility

in collaboration with the Irish EPA.

The rugged construction of these autonomous

platforms, as shown in Figure 1(i), enables long-

term in-situ monitoring of transient gas events. The

system is deployable with GSM communications

enabling transmission to a remote base-station,

which parses the data onto a database and uploads it

onto an online portal. The performance of these

platforms has been validated by deployment

LandfillGasMonitoringNetwork-DevelopmentofWirelessSensorNetworkPlatforms

223

durations of up to 13 months (Collins et al., 2011).

In addition to the autonomous monitoring operation,

analysis of the data provides a value-added service

by investigating the factors affecting landfill gas

activity, evaluating the contributions of extraction

flow-rate and local weather conditions (barometric

pressure and rainfall). This paper presents the

initialisation of a deployment of multiple sensor

platforms on one site, the first of its type on an Irish

landfill site.

2 PLATFORM DEPLOYMENT

2.1 Deployment Configuration

A wireless sensor network in the form of five

autonomous platforms has been deployed on a single

active landfill site in Ireland. To minimise any

debugging issues and allow for flexibility in locating

the systems, each of the five systems has been

equipped with independent GSM modems for

wireless communications. While this may not satisfy

the classical definition of a wireless sensor network

comprising of a central gateway platform with

peripheral node platforms, the five systems

nevertheless represents the increased spatial

distribution achieved by a sensor network.

There were two configurations of platform

deployed on this site: three units were fitted with

and CO

infrared gas sensors (Premier Series,

Dynament, UK) for gas concentration monitoring,

two units were fitted with pressure sensors (40PC,

Honeywell, UK) for monitoring extraction pressure.

The two configurations of platforms employed

slightly different data acquisition strategies. The gas

platforms ran a sampling routine every 6 hours, with

data from that sampling routine being transmitted

immediately afterwards. The sample routine

consisted of a two-minute sampling period, where

the sensors took a reading every three seconds as the

pump drew the gas from the landfill’s extraction

network, resulting in 40 readings each for CO

and

. The two-minute routine enabled a reasonable

volume (~0.6 litres) of gas to be sampled, thus

avoiding any outliers caused by pockets of

concentrated gas. The full dataset of 80 readings was

encoded to fit within the 160 character SMS limit,

whereupon it was transmitted as a text message to

the remote base-station. As the pressure was by

nature more variable, readings were acquired every

30 minutes, stored on flash memory and transmitted

by SMS every 24 hours. Upon reception in the base-

station (based in the lab in DCU), a java program

decoded the text message, parsed the data to a

database via MySQL and uploaded averaged data to

an online portal which could be accessed from a web

browser.

The configuration of the deployment is shown in

Figure 1(ii). The deployment began in December

2011, with two pairs of gas/pressure platforms

deployed in-line with the extraction pipes leading to

each of the two flares on site (locations #1 and #2 in

Figure 1(ii)). The reasoning behind these locations

was that the readings could be validated against the

flares’ SCADA monitoring systems. A third gas

monitoring system was deployed in a newly capped

cell (location #3 in Figure 1(ii)) to monitor the gas

activity as the waste body matured.

2.2 On-site Issue Resolution

As can be expected for outdoor environmental

deployments, there were a number of issues to be

surmounted including power requirements, wireless

communications and sensor readings validity.

Furthermore, there were site-specific issues inherent

to conditions on the landfill site.

A concern with all remote devices is power

provision. For these monitoring platforms, low

power ‘sleep’ algorithms programmed into the

microcontroller circuitry enabled a 10-week

deployment duration using the 12V 5Ah lead-acid

battery. To avoid system downtime, photovoltaic

panels (BP SX-5M) with charge controllers (Solar

Technology STS01208) were fitted to the three gas

monitoring systems. As seen in Figure 2, solar

charging was found to acceptably sustain the battery

level of the system indicating an indefinite

deployment period in terms of power requirements.

Note the 13.5V upper limit in Figure 2 is not

representative of the actual battery voltage but

instead is the maxed-out voltage potential across the

solar panel during daylight hours. A truer reading of

the battery level is the night-time values - the

minima in Figure 2 (ii). Solar charging was not

necessary for the pressure monitoring systems which

had an estimated battery life of 12 months due to the

lack of the pump and power hungry IR sensors.

Figure 2: Power levels during operation (i) battery

depletion rate, (ii) battery solar recharging.

(i)

(ii)

SENSORNETS2013-2ndInternationalConferenceonSensorNetworks

224

Setting up communications was straight-forward

with acceptable GSM coverage found for the

deployment locations on-site. Intra-network

communication protocols (RF, etc) were not

implemented at this stage due to issues with line-of-

sight and the need to simplify the initial network

operation. This will be the subject of future

development when the scale of the deployments

increases. Communications was the principal

contributor to the running costs of the platforms;

monthly text message costs of €6.30 and €2.70 are

associated with each gas and pressure system,

respectively.

The validity of the sensor readings was

confirmed by occasional spot checks using the

industry standard equipment (GA2000, Geotech

Instruments, UK). Moreover, the gas monitoring

platforms were retrieved to the laboratory to test

their calibration settings after eight months

deployed. As seen in Figure 3, the CO

sensor was

found to have a sensitivity drift of 1.2% and zero

drift of 6.38%; the CH

sensor had a sensitivity drift

of 0.6% and zero drift of 8.8%. This was considered

to be an acceptably small drift considering continual

exposure to high gas levels as well as the presence

of other noxious and corrosive gases such as

hydrogen sulphide (H

S).

Figure 3: Comparison of sensors’ calibration data before

and after 8 months deployed (i) CO

, (ii) CH

A number of issues specific to this site were

encountered and rectified in the early months of the

deployment. Firstly, a high moisture content in the

landfill’s extraction pipes led to condensate forming

in the monitoring systems’ connection tubing, hence

inhibiting the sampling of gas, see Figure 4(i). This

was rectified in two ways: the monitoring systems

were elevated on their mountings such that any

condensate would flow back down into the landfill

pipes and the tubing was insulated by neoprene

sleeves to reduce the differential temperature

between the tubing and the landfill pipes, thus

lessening the extent of condensation. The second

issue related to the tubing, where an abrupt increase

in ambient temperature in April 2012 led to

shrinkage of the tubes until disconnection occurred,

see Figure 4(ii). The length of the tubing was

extended to compensate for this. The third issue

concerned the landfill pipe at location #3; given that

this was a newly capped cell, remedial construction

works were ongoing to stabilise the pipe which was

collapsing from its support ridge due to the weight

of excessive water in the pipe, resulting in the

breakage of connections on numerous occasions, see

Figure 4(iii). These issues were identified from

observing the online data, enabling the monitoring

operation to be resumed quickly.

Figure 4: Site-specific issue rectification (i) Water-

blockage, (ii) Tube shrinkage, (iii) Connection breakage

due to pipe collapse.

(i)

(ii)

(iii)

LandfillGasMonitoringNetwork-DevelopmentofWirelessSensorNetworkPlatforms

225

3 RESULTS & DISCUSSION

To date, the platforms have operated for up to 9

months, with valid readings totalling over 3,000 and

25,000 for gas concentration and pressure,

respectively. Subsets of the data from the sensors in

Locations #1 and #2 are shown in Figure 5 (some

gaps in the data were due to the aforementioned

issues described in Section 2). Distinct ‘events’ are

clearly seen in the data, coinciding in both the gas

concentration and pressure values. Such evidence of

the dynamics of landfill gas concentration has not

been available previously; now, for the first time, the

cause and effect relationship of landfill gas

behaviour can be investigated. Interpretation of this

gas field behaviour would be conducive to more

informed decisions about the impact of field-

balancing.

Figure 5: Subset of monitored data, where ‘events’ are

evident in both gas and pressure readings.

The variations in the constituent gas levels

tended to be correlated, i.e. CO

and CH

tend to rise

and fall at the same time. This indicated that the gas

activity was predominantly controlled by a singular

factor affecting the entire volume of the gas, most

likely the extraction rate towards the flare. However,

it can be seen that CH

at times experienced a

greater differential change possibly due to other

factors, such as varying methanogenic activity of the

waste body associated with the different stages of

waste decomposition and increased moisture content

after rainfall. Clearly, the availability of near real-

time methane concentration variation across the site

can assist in field-balancing to attain the optimum

gas composition for combustion in the flare.

Pressure readings also tended to align with gas

activity. This was an interesting finding: given their

substantially lower cost, greater proportion of

pressure sensors could be deployed for the same

investment. A relationship established between

pressure and gas would enable pressure sensors to be

deployed as low-cost indicative monitors, signalling

a possible change in gas concentration levels which

could then be verified by a smaller number of gas-

specific sensors. Judging from Figure 5, the

correlation between gas and pressure appeared

stronger for location #1 compared to Location #2;

the investigation of the relationship between gas and

pressure is subject to further careful consideration.

Figure 6: Subset of data indicating correlation between

location #1 deployed platforms and SCADA (a) gas

comparison (b) pressure/flow comparison.

Table 1: Correlation analysis of deployed systems and

SCADA measurements (values approaching ±1.0 indicate

stronger correlation).

Deployed

platforms

SCADA

Pressure Flow

0.72 - -0.024 -0.005

- 0.736 0.059 0.107

Pressure - - 0.065 0.218

0.649 - 0.195 0.358

- 0.592 0.337 0.505

Pressure - - 0.376 0.307

SENSORNETS2013-2ndInternationalConferenceonSensorNetworks

226

Correlation analysis was conducted between the

platforms’ data and the flares’ SCADA data. The

correlation for two months (64 data points) is

demonstrated in Figure 6 with correlation

coefficients presented in Table 1. Reasonably strong

correlations for gas concentrations were found for

both locations and their respective SCADA

measurements. This serves to validate the gas data

collected by the deployed platforms. Interestingly,

this correlation is not consistently positive, e.g.

negative correlations in gas levels on 2/3/12 and in

pressure on 27/3/12 as seen in Figure 6. Gas activity

measured locally at a specific point in the gas field

does not necessarily manifest at the flare. This

underscores the incapability of the flare’s SCADA

measurements to fully describe behaviour in gas

activity further along the extraction system – a task

that could be accomplished by distributed sensor

network.

The correlations in gas concentration levels were

slightly higher for location #1 compared to location

#2 (0.73 vs. 0.62 averaged); as this difference is

slight, no major conclusions are being drawn though

this is subject to further investigation. Conversely,

correlations for pressure/flow were substantially

stronger for location #2 (>0.3 vs. <0.2); the

interrelationship was less evident for location #1

possibly due to other gas inlets to that particular

flare.

A lesser correlation is seen between the CH

values and SCADA flowrate compared to that of

(0 vs. 0.1 for location #1; 0.35 vs. 0.5 for

location #2). It is difficult to ascribe variations in a

system as complex as a landfill gas field to a specific

source; one possibility is differing levels of

methanogenic activity in the waste body according

to varying conditions of waste decomposition.

Figure 7: Methane readings from locations #2 and #3.

The simultaneous measurement of gas levels at

different points in the landfill extraction system is

another advantage provided by distributed sensor

platforms. Location #3 was situated on an extraction

line leading to the flare via location #2, see Figure

1(ii). A subset of the CH

data from both locations is

displayed in Figure 7. CH

generation in location #3

is seen to be less than that in Location #2 – this is to

be expected for a new-capped waste cell where

waste decomposition is not at a sufficiently

advanced stage. The elevated CH

levels in location

#2 indicate that the field is balanced from other cells

to provide the appropriate composition for the flare.

Interestingly, CH

levels in location #3 matches

those in location #2 on a few occasions; the authors

estimate that the extraction rate from location #3

was increased, hence reducing the CH

concentration by allowing less time for the gas to

accumulate. This finding illustrates the benefit of

immediate feedback following field-balancing,

allowing for more precise and effective gas

management.

4 CONCLUSIONS

Autonomous monitoring platforms have been

developed for the application of monitoring gas on

landfill sites. This paper describes the deployment of

a wireless sensor network on an active landfill site,

the first of its type in Ireland. The near real-time

access to gas concentration and pressure data via an

online portal enabled the landfill operators and EPA

to characterise, for the first time, the dynamics of

landfill gas activity. The authors aim to continue

working closely with landfill operators,

demonstrating that the integration of data arising

from distributed sensor networks into landfill

operational practice would assist in field-balancing

and gas management. Furthermore, the cost-benefit

of this service will be promoted where the operation

of the engine/flare can be more effectively managed.

At this approximately half-way point in the

deployment duration, the preliminary analysis

presented in this paper illustrates the valuable insight

provided by knowledge of trends in landfill gas

activity; more comprehensive analysis is ongoing

with the deployment projected to continue until

March 2013.

Further development of the platforms will

involve implementing different communications

protocols and reducing platform costs. The current

use of GSM modules for all platforms is not

financially viable when considering the extrapolated

running costs associated with a greater magnitude of

deployed systems. Reducing costs to a viable price-

point is an integral part of the commercialisation

process of this technology; in a sector that is

LandfillGasMonitoringNetwork-DevelopmentofWirelessSensorNetworkPlatforms

227

becoming increasingly financially restricted, the

ideal scenario of installing a sufficient number of

platforms to fully encapsulate the gas field activity is

currently too cost-prohibitive. Despite this, the

authors believe that a viable commercial prospect

exists with the increased temporal and spatial

resolution provided by distributed monitoring

platforms complemented with value-added analysis

and interpretation of landfill gas data.

ACKNOWLEDGEMENTS

The authors wish to acknowledge funding from the

EPA under the STRIVE programme (grant no. 2010-

ET-MS-10), and to thank the EPA and the landfill

facility management for their co-operation.

REFERENCES

European Parliament and Council of the EU, 2008.

Directive 2008/50/EC on ambient air quality and

cleaner air for Europe.

EPA, 2012. Ireland's Environment - An Assessment.

Environmental Protection Agency, Ireland.

Cohen, R., Teige, V., Weichsel, K., 2012. The Berkeley

Atmospheric CO

Observation Network. URL:

http://beacon.berkeley.edu/

Envirowatch, 2012. URL: http://www.envirowatch.ltd.uk

Beirne, S., Kiernan, B., Fay, C., Foley, C., Corcoran, B.,

Smeaton, A. F., Diamond, D., 2010. Autonomous

greenhouse gas measurement system for analysis of

gas migration on landfill sites. In: SAS 2010 - IEEE

Sensors Applications Symposium, Limerick, Ireland.

Fay, C., Doherty, A., Beirne, S., Collins, F., Foley, C.,

Healy, J., Kiernan, B., Lee, H., Maher, D., Orpen, D.,

Phelan, T., Qiu, Z., Zhang, K., Gurrin, C., Corcoran,

B., O’Connor, N., Smeaton, A.F., Diamond, D., 2011.

Remote real-time monitoring of subsurface landfill gas

migration, Sensors, vol. 11, pp. 6603-6628.

Collins, F., Orpen, D., Fay, C., Diamond, D., 2012.

Analysis of landfill gas migration by use of

autonomous gas monitoring platforms. Proceedings of

The 27

International Conference on Solid Waste

Technology and Management, Philadelphia, USA.

Collins, F., Orpen, D., Fay, C., Foley, C., Smeaton, A.F.,

Diamond, D., 2011 Web-based monitoring of year-

length deployments of autonomous gas sensing

platforms on landfill sites. Proceedings of IEEE

SENSORS 2011, Limerick, Ireland.

SENSORNETS2013-2ndInternationalConferenceonSensorNetworks

228