2012 Sixth International Conference on Sensing Technology (ICST)

Gas sensing using acoustic attenuation with

improved resolution
Ajit Singh Prof. Maringanti Radhakrishna
Electronic Division Electronic Division
Indian Institute of Information Technology Indian Institute of Information Technology
Allahabad, India Allahabad, India
Email: ajitsingh@iiita.ac.in Email: mkrishna@iiita.ac.in

Abstract—Sensors being used for gas sensing in underground reflection of current gas concentration level. Characteristic of
coal mine require frequent recalibration due to highly humid gas leakage in coal mine is different from these assumptions.
and toxic environment prevalent in UG coal mines. If we intend Leaking methane being lighter than air will diffuse towards the
to deploy a WSN which can continuously monitor the UG mine
environment then we require sensors which are able to withstand roof forming a thin layer [1]. Carbon dioxide being heavier
the environment of UG coal mine. Acoustics gas sensors are than air will diffuse towards the floor. Assumption that leaked
more rugged and would be able to work for prolonged period gas diffuses quickly is also not appropriate. Rate of diffusion
in UG coal mine environment. Acoustic sensors work on the will depend on volume of the gas being leaked. So a sensing
principle that acoustic signal velocity and amplitude are affected node placed at fixed position would not be able to detect a gas
by changing composition of air. Attenuation of acoustics waves in
presence of a target gas can be used to detect its presence and give leak or detect it only after high concentration of leaked gas
its concentration level. We present an acoustic signal processing have been developed at places. Therefore a sensing system is
approach which can measure attenuation of acoustic signal with desired which should have a wider area of coverage.
improved resolution. Proposed approach takes advantage of the In contrast to existing sensors, an acoustic sensor is sensitive
fact that when triggered with a transmitted pulse of acoustic to gas composition change in a wider area as acoustic beam
sensor resonance frequency, the receiver output is a damped
ringing waveform. We measure the area under envelop of the travel from transmitter to receiver traversing through air. It
received waveform to provide increased resolution. is an area sensor. It can be therefore used to monitor gas
leakage in large area, the distance between transmitting and
I. I NTRODUCTION receiving sensors. Acoustics based gas sensor have been used
Prolonged and continuous gas sensing in UG coal mines in the past, some for specifically sensing gas concentration
throws a unique set of challenges. Currently used gas sensors level in underground coal mines like methane whistle, used
are not suitable for above as highly humid and toxic environ- in German coal mines. Acoustic phenomena that are used for
ment prevalent in UG coal mines makes them inoperable after gas sensing are acoustic resonance [2], surface acoustic waves
a certain period of time. To be useful the sensors therefore [9], acoustic attenuation [6], [12] and speed of sound [14].
have to be brought to surface for restoration of their sensing Gases in a mixture can be identified using acoustic attenuation
ability and recalibrated. In UG coal mine galleries two sensing [3], [6], [13]. Prototypes for sensing acoustic attenuation for
scenarios are visualized for gas sensors. In the first scenario gas mixture were evaluated [3], [7] establishing that using
they will be used to detect a leak of target gas from a point acoustic attenuation for identifying change in gas composition
source. In second scenario they will measure the concentration is a viable option. For analysis a controlled environment
level of target gas, where leaked gas source is not known, it with stable temperature and humidity condition was designed.
could be slowly leaking from multiple points and therefore Acoustic resonance and SAW based sensors are even more
it has diffused(assumed evenly) in the environment air. Gas accurate [2], [9], but there working is strongly affected by
sensor need to detect the gas leakage in the first case, whereas stability of the environment in the enclosing they are placed
it has to detect slow build up gas concentration level in the into.
second case. Measuring acoustic attenuation with sufficient accuracy is
Currently used electrochemical, pellistors and Infra Red a challenge and this challenge increases if measurement are
gas sensors suffers from many limitation, for prolonged and not being done in controlled environment, which would be
continuous gas sensing in UG coal mines [10], [11]. They required if acoustic sensors are to be used in real life situations.
can be said to be point sensor i.e. they are sensitive to the gas Our objective is to evaluate suitability of acoustic attenuation
concentration level in their close vicinity. For the first scenario parameter for gas sensing when transceiver pairs are placed
gas sensor need to be properly placed to be able to sense larger distance apart, of the order of few meters. We are
the leakage. In second scenario, it is expected that leaking aware that there won’t be tight control over temperature and
gas diffuse in the tunnel air evenly and quickly and therefore humidity parameters, correspondingly measurements being
measurement made at any location in neighbourhood is correct made will have larger errors. This paper analyses whether

978-1-4673-2248-5/12/$31.00 ©2012 IEEE 543

acoustic attenuation can give sufficient information that it can
be used in improving the accuracy of acoustically sensing
leak or presence of gases. Beside attenuation, other acoustic
parameter will also be used, but in this paper only acoustic
attenuation change is being evaluated. Fig. 1. Received Signal processing
Another significant way in which our analysis is different
from existing work is that, the concentration levels of gases
which are used in existing analysis are an order of magnitude
more than what will be required to be detected in UG coal
mines. In first scenario the change in concentration level is
going to vary depending on distance of sensing unit and the
leak rate. Still it would be in the range of 5-10% by volume
averaged over the distance between sensing units. In Second
scenario, the change in concentration level that has to be
detected in UG coal mine would be even smaller, e.g. 2%
(lower flammable limit) by volume change in concentration
level for methane with a minimal resolution of 0.25% by vol-
ume. Therefore there is a need to measure change in acoustic
parameters with improved resolution. Acoustic attenuation in
itself, for these very low concentration level changes would Fig. 2. Squared received Signal(Blue) and its envelop(Red)
not be a reliable parameter for identifying and predicting gas
leaks, but they can be used to increase the confidence in the
prediction made using other acoustic parameters like time of For making a measurement, transmitting unit is excited with
flight. 10 square pulse of 40 kHz frequency. The signal traverse
through the air and then is sensed by the receiver. Received
Acoustic sensing can only be used for gases whose affect
signal is filtered using 500 kHz anti aliasing filter and dig-
on acoustic signal is significant like methane,Carbon diox-
itized. After digitization acquired signal is passed through
ide,ammonia etc in comparison to air. In this paper acoustic
a digital 35 kHz high pass filter (figure 1). For real time
signal attenuation in methane and carbon dioxide is measured.
measurement of area under curve of received signal a FPGA
Methane is the prime gas whose concentration level has to be
based parallel implementation of signal processing algorithm
monitored in UG coal mine. To improve the resolution instead
is used. Objective of using an FPGA based set-up is to improve
of measuring peak to peak deviation in amplitude of received
the resolution of the sensing system. Further signal processing
signal, received signal intensity is measured by measuring
based on other aspect of the acoustic signal will be integrated
area under curve of the received signal waveform. Affect of
with the current system. An FPGA based system will help
environmental parameters like temperature and humidity has to
in adding parallelism and reducing the signal processing time
be considered in the analysis during measurements of acoustic
thereby enabling real time and fast sensing. Thereafter received
attenuation. Change due to these parameters can be used to
signal envelop is generated using decimation and low pass
improve confidence in predictions. Fortunately in the UG coal
filtering. Area under curve is then calculated using trapezoidal
mine environment which are highly humid, humidity level
rule (figure 2).
more or less remains constant which is close to 100%, so
effect of humidity can be readily pre calculated and included III. DATA ANALYSIS AND R ESULTS
in prediction being made. Also in Underground environment
temperature is also stable in a narrow range or fluctuate very The absorption of sound in fluids is due to the inability
slowly. of molecular degree of freedom to follow acoustic fluctuation
[4], [5] i.e. acoustic signal attenuation is related to molecular
II. T EST S ETUP structure of the gases through which it travels. In comparison
to air which largely consist of nitrogen and oxygen which
A test set-up consisting of a ultrasonic 40 kHz transmitter are diatomic gases, acoustic signal attenuate more in carbon
and receiver sensor placed fix distance apart facing each other dioxide and methane which are polyatomic gases [3], [6],
is designed for measurement. To evaluate for first scenario a [7], [13]. When target gas is introduced in the environment
syringe is filled with methane or Carbon dioxide and gas is received acoustic signal characteristic changes. This change
injected in the path between receiver and transmitter to mimic can be measured using received signal amplitude variation
behaviour of a gas leak. For second scenario after injection of or we can measure signal intensity by taking integral of the
measured amount of gas, a fan is used to mix the gas evenly square of the signal. We have analysed advantage of measuring
in the chamber. After being thoroughly mixed with the air, received signal area under curve and using this parameters to
methane or Carbon dioxide does not quickly segregate and detect change in gas concentration level.
form a layer at the ceiling or the floor of test chamber [1]. For our analysis we have measured and compared deviation

544
in peak amplitude of received acoustic signal and change in
signal intensity measured by taking area under curve of the
received damped ringing signal waveform, for the correspond-
ing change in gas composition. In first scenario received signal
maximum amplitude when gas is not leaking is compared with
maximum amplitude of signal when gas(methane and CO2)
is leaking to identify effect of leaking gas on the acoustic
signal amplitude. For a small change in gas concentration the
corresponding change in amplitude is negligible. If we take
maximum, minimum or average amplitude of received signal Fig. 3. Max Amplitude in Methane(Red),Carbon Dioxide(Green) and
result would be similar to (figure 3) and would be inconclusive Air(Blue)
for low concentration change in gas composition. In second
scenario variation is even more inconclusive as concentration
changes are even lower.
This is improved by taking into consideration the fact that
the received signal is a damped ringing waveform not a single
pulse. The waveform peak and the ringing duration is going
to be dependent on the received acoustic signal intensity
which in turn is dependent on the gas composition of the
medium through which it travels. Area under the envelop of
the received waveform is used to measure the received signal
intensity and correspondingly change in gas composition. For Fig. 4. Received signal Area Under Curve in Methane(Red),Carbon Diox-
ide(Green) and Air(Blue)
first scenario, as evident from the (figure 4), relative changes
in area under curve for gas leak is much more compared to
relative change in the peak amplitude of the received signal
IV. C ONCLUSION
waveform. Variation in the measurements is due to limitation
of the test set-up to provide constant gas release. For second Acoustic measurement of gas leak require that we are able
scenario, an improved test set-up has to be used to quantify to measure attenuation of acoustic signal in gas with high
change in area under curve with corresponding precise change degree of accuracy. Measuring area under curve of received
in target gas concentration level. signal enable us to measure change in signal strength with
The lower flammable limit for methane in UG coal mine is improved resolution. Further work is need to evaluate and
5% by volume but most of the commercially available methane quantify change in area under curve with exact change in gas
sensors use 2% by volume as the lower limit, i.e. the limit concentration level for the second scenario. Also effect of high
at which the alarm is generated. Proposed approach would humidity is to be evaluated and quantified. Proposed approach
be useful for situations where concentration level change can still be used to detect gas leakage not only in UG coal
of 5-10% by volume are involved, but its utility for lower mines but in other places as well where possibility of a gas
concentration changes seems limited. Further experimentation leak is there, like petrochemical plants. Advantage of acoustic
with more precise test set-up is required to quantify the lower sensing approach is that a single sensing system can be used
limit of concentration level change where acoustic attenuation to sense many different gases. Using acoustic sensors for gas
change, measured by taking area under curve of the received sensing in UG coal mines will enable real time, fast and more
signal can be reliably used to make a prediction. accurate sensing of gas leakage.
Acoustic attenuation measurement even if it is not reliably
able to measure change in gas concentration upto required R EFERENCES
resolution, will be able to assist in improving the confidence
[1] Kissell, Fred. N., Information Circular 9486,CDC NIOSH, Handbook
in other acoustic measurements. For e.g., Acoustic signal For Methane Control in Mining, (2006).
attenuation is greatly affected by change in humidity levels [2] Garrett,S , Sonic Gas analyzer for Hydrogen and Methane , Proceedings
[8] but has negligible affected due temperature change. This of the Acoustics, 2008, Paris, France, 29 June - 04 July, (2008).
[3] Petculescu, Andi, Future trends in acoustic gas monitoring and sensing,
feature can be used to negate the error in time of flight Journal of Optoelectronics and advanced materials, 8 (1), 217–221,
measurements due to temperature change. In this analysis all (2006).
measurement are taken in dry condition but measurements will [4] Henderson, Malcom C., Sound In Air: Absorption and Dispersion ,
be severely affected if the environment have high humidity SOUND , 2 (6), 28–36, (1963).
[5] Evans, L.B., Bass, H.E. and Sutherland, L.C., Atmospheric Absorption
levels. Humidity sensors can be used to measure current hu- of sound: Theoretical predictions, Journal of Acoustical society of
midity level and humidity level measurements can be included America, 51 (5) part(2), 1565–1575, (1972).
in the analysis. Temperature also can be measured using [6] Ejakov, Sally G., Phillips, Scott, Dain, Yefim, Lueptow, Richard, M. and
Visser, Jacobus H., Acoustic attenuation in gas mixtures with nitrogen
a temperature sensor and its effect can be included in the : Experimental Data and Calculations, Journal of Acoustical society of
analysis. America, 113 (4) part(1), 1871–1879, April (2003).

545
 [7] Petculescu, Andi, Hall, B., Fraenzle, R.,Phillips, S. and Leuptow, R.
M., A Prototype acoustic gas sensor based on attenuation(L), Journal
of Acoustical society of America, 120 (4), 1779–1782, (2006).
[8] Bohn,Dennis A. , Environmental Effects on the speed of sound , 83rd
convention of the audio engineering society, 1987, New York, USA, 16
- 19 October, (1987).
[9] Shuh-Haw Sheen, Hual-Te Chien, Apostolos C. Raptis, Ultrasonic
techniques for detecting helium leaks, Sensors and Actuators ,Elsevier
Science, 71, 197-202, (2000).
[10] Taylor,C.D., Chilton,J.E. and Martikainen, A.L., Use of infrared sensors
for monitoring methane in underground mines , 12th US/North American
Mine Ventilation Symposium, 2008, Wallace, USA, 9 - 11 June, (2008).
[11] hester, Colin, Emerging Technology in gas monitoring equipment ,
QCO/DME Coal Industry Safety Conference in conjunction with Haz-
coal Management, 1996, Queensland, Australia,(1996).
[12] Dain, Y. and Leuptow, Richard M., Acoustic attenuation in three-
component gas mixtures Theory, Journal of Acoustical society of
America, 109 (5), 1955–1964,May (2001).
[13] Dain, Yefim and Leuptow, Richard M., Acoustic attenuation in three-
gas mixtures Results, Journal of Acoustical society of America, 110
(6), 2974–2979,December (2001).
[14] Tobias, Peter, Apparatus and method for using the speed of sound in
photo acoustic gas sensor measurement, Honeywell International Inc.,
Morristown NJ(US),US Patent Pub No: 0147051 A1, Dec. 11 2008, Jun
17 2010.

546