FUGITIVE METHANE DETECTION AND LOCALIZATION WITH SMALL UNMANNED AERIALSYSTEMS: CHALLENGES AND OPPORTUNITIES

DEREK HOLLENBECK, YANGQUAN CHENDepartment of Mechanical Engineering, University of California, Merced.

Contact: {dhollenbeck,ychen53}@ucmerced.edu Acknowledgements: NSF NRT Fellowship (http://www.nrt-ias.org)

REFERENCES

[1] Matheou et al. Environ Fluid Mech., 2016.

[2] Li et al. Int. Conf. on Rob. and Biomim., 2009.

[3] Smith et al. ICUAS Miami., 2017.

[4] Farrell et al. Env. Fluid Mech., 2002.

[5] Nurzaman et al. PLos ONE, 2011.

INTRODUCTIONNatural gas is one of our main methods to gener-ate power today. Utility companies that providethis gas are tasked with maintaining and survey-ing leaks. These leaks are referred to as fugitivemethane emissions and detecting these fugitivegases can be pivotal to preventing incidents suchas the San Bruno explosion, killing 8 and injur-ing dozens due to a gas leak going undetected.Recently, using NASA technology onboard lowcost vertical takeoff and landing (VTOL) smallunmanned aerial systems (sUAS) we can detectfugitive methane at 1 ppb (parts per billion) lev-els.

CHALLENGES IN DETECTIONGeneral challenges include: FAA regulations (noflights over people), battery life, and complex dy-namic plume behavior. Factors that impact de-tection can be: propeller wash, sensor placement,wind, and mechanical/electrical noises. Evendistance to source and flight altitudes can changethe probability of detection (Sigmoid like) scal-ing with topology and atmospheric stability. Lo-calization by CFD approaches are costly makingreal-time estimations and visualizations difficult.

QUASI-STEADY INVERSIONFollowing the work by Matthes et al (2005),Carslaw (1959), and Roberts (1923) the solutionto a single point source advection diffusion equa-tion (ADE) can be solved for a dynamic systemapproximately by making a quasi-steady state as-sumption if the variance and transient behaviorof the wind small. W0 is the Lambert function.

∂C

∂t−D∂

2C

∂x2i

+ v∂C

∂xi= 2q0δ(t− t0)δ(xi − xi0)

C(xi, x0, qo)i =q0 exp( v(xi−x0)

2D )

π23Dd

di(Ci, x0, q0) ≈ 2D

vW0(

vq0

4πD2Ciexp(

v

2D(xi−x0)))

minq0,x0

:m∑

i,j=1

(y0,i(x0, q0)− y0,j(x0, q0))2

ADAPTIVE SEARCH MODELIn the foraging literature the Levy walk has beenshown to be effective at searching sparse environ-ments. However, Brownian motion is more effi-cient in dense areas. This adaptive search model[5] can switch dynamically from Levy to Brown-ian based on finding targets using tumble proba-bility P (x(t)), x(t) is governed by the stochasticdifferential equation (SDE) below

P (x(t)) = e−x(t), 0 ≤ x ≤ 5

x = −∂U∂x

A+ε,

U = (x− h)2, ε :

{H = 1

2 ,N(0, σ)

H 6= 12 , fGn

A = max(Amin, α(t))

αk = Cααk−1 + ktF

{F = 1, found target

F = 0, otherwise.

we extend [5] by adding, fGn, defined as Yj =BH(j+ 1)−BH(j) and fraction Brownian motionis given below.

ADAPTIVE SEARCH AND LOCALIZATION

The adaptive search model has shown to adjustfrom Brownian motion to Levy walks in a 2D ran-dom search. By reducing the problem to a 1Dpath problem (i.e. survey route) adding decisiontrees and modeling fugitive gas with a small timescale filament model [4] we have the opportunityto optimize random search for application. Gatherenough information to form a sample(s) to use inthe inversion method for a Zeroth order approxi-mation of source localization (x0,y0) and quantifi-cation (q0).

BH(x) ≈∑

φ(x− y)B(∆y) φ(x) =Γ(H + 1− d/2 + ||x||)

Γ(||x||+ 1)Γ(H + 1− d/2)≈ ||x||H−d/2

Γ(H + 1− d/2)

EXPERIMENTAL RESULTS

Using the quasi-steady inversion method on ex-perimental data we can see the results from justtwo samples (blue) in the presence of two sources(red). Only taking a small section of raw datafrom each longitudinal pass we can approximatethe source (green) from our measurement withthe OPLS [3].

FUTURE RESEARCHThis work hopes to optimize this adaptive searchstrategy efficiency η = N/L (N is the numberof targets found and L is the total distance trav-eled) through transition parameters (Cα, Amin,and kt) the potential (h), and the choice of noise(i.e. Gaussian or fGn) by means of evolution-ary algorithms. Furthermore, we want to answerhow the level of noise σ and how the Hurst pa-rameter H , stochastically shift the tumble prob-ability through x(t). Once we have an optimalmodel we look to compare with current methods(i.e. Zig-Zag, spiral surge [2]), and other gradi-ent or flux based approaches (stochastic gradientdescent, fluxotaxis, infotaxis etc.).