skip to content

Composites and Coatings Group

Department of Materials Science & Metallurgy

The volcanic ash problem

Airborne volcanic ash can be a significant threat to the safe operation of gas turbine powered aircrafts [1]. In recent years, especially during the past 20 years, the threat from volcanic ash to the aviation industry has been recognized and documented [2]. The 2010 eruption of Eyjafjallajokull (Iceland) and the 2011 eruptions of Grimsvotn (Iceland) and Cordon Caulle (Chile) have caused millions of pounds worth of damage to the airline industry worldwide. Overall 95,000 flights had to be cancelled all across Europe during the six days of Eyjafallajokull eruption alone costing the airline industry millions of pounds.

Adhesion of such particulate can lead (at relatively high levels) to disruption of gas flow through the engine, ranging from the blockage of cooling channels [3,4] to complete inhibition of air passage and combustion.  At lower levels, short- or medium-term damage can be caused to protective ceramic coatings, with a particular danger of promoting spallation [5,6].  These effects are potentially of major concern, but they all depend on the likelihood of particle adhesion. As a consequence, the first requirement would be the development of a volcanic ash adhesion criteria based on ash properties (particle size, composition, amorphous content, etc) and engine conditions. To do that, a systematic adhesion testing, and associated computational fluid dynamics (CFD) modelling, is required using a range of temperature and velocity conditions and ash types. This should allow identification of combinations of conditions where a certain type of volcanic ash tends to stick into a solid surface. Such criteria can then be used to predict the likelihood of adhesion in real engines for a given ash type or at least to classify the level of danger [7].


Research Programme

2.1         Volcanic Ashes

Several volcanic ashes (VA), Laki, Hekla, Askja and Eldjga, with a range of composition, amorphous/crystalline content, glass transition (Tg) and melting (Tm) temperatures are being investigated with a controlled particle size distribution. Figure 1 presents a summary of the main characteristics of the studied VA types, composition (Fig.1a), amorphous/crystalline content (Fig. 1b), Tg and Tm (Fig. 1c) and particle size distribution (Fig. 1d).


Figure 1: Volcanic ashes characteristics: (a) composition, (b)X-Ray diffraction pattern, (c) dilatometry measurements and d) representative particle size distribution


The main component in VA is SiO2 although several other elements can be present in their composition as shown in Fig. 1a. It can be observed that Askja and Hekla are mainly SiO2 with some Al, Ca and K whereas Eldgja and Laki present lower content of SiO2 with a range of additional oxides (Fe, Ti, Ca, K, Al, Mg and Na). It would be expected that this differences in composition would affect ash properties such as Tg and Tm. Figure 1b shows the X-Ray diffraction patterns of the 4 selected volcanic ashes. The broad peak around 20-30° represents the amorphous phase present in each of the ashes. Using these ashes, amorphous content ranging from 50 to 100 % can be studied. Glass transition (Tg) and melting temperature (Tm) are very relevant properties of the ashes as it represents the softening temperature of the different types of ash. To evaluate the Tg and Tm of the selected ashes, a dilatometry set-up shown in the insert of Figure 1c was used. A cold isostatic press is used to fabricate a “green” pellet of the VA which is pushed by an alumina push road at a constant load (0.3 N) and heat it up to 1200 °C at 5° C   min-1.  Figure 1c shows the dilatometry curves for the different VA where a contraction of the pellet can be detected due to softening of the amorphous phase on passing through the Tg. Sticking of VA into the jet engine presents high influence of the particle size of the ingested particulates so it is important to control the particle size distribution (PSD) of the studies ashes. Figure 1d shows a representative PSD of the VAs after being sieving through a ~ 40 µm aperture sieve (average of 30 µm).



2.2         Measurement of volcanic ash deposition rates

A new Set-Up has been developed where ash powder is fed at predetermined rates into the Vacuum Plasma Spray (VPS), within a controlled pressure chamber. Gas flow is confined within a tube, simulating the combustion chamber of a turbine, with gas temperature and velocity being in broadly comparable ranges (T ~ 800-1100˚C, V ~ 100 m s-1). Deposition takes place on a static surface (with or without a coating), inclined at selected angles to the axis of the tube - see Fig.2.



Fig.2:  Schematic of Ash Particle Deposition Set-up.


This set up allows controlling temperature and velocity fields as well as the substrate position. Using this set up, the mass fraction of particles incident on the substrate that adheres to it can be measured. This would allow acquiring information concerning the significance of ash characteristics (particularly phase constitution, and hence softening temperature, as well as particle size distribution) in determining whether deposition is likely for given engine operating conditions.



Fig.3:  Proportion of incident particles adhering to the insert, as a function of the substrate temperature for several volcanic ashes


Figure 3 presents the mass proportion of adhering particles of the four volcanic ashes, with an average size of 30 µm particle size, at different substrate temperatures. There is a clear tendency for all the ashes to increase their deposition as the substrate temperature increases. However, it can be observed that there is a difference in their deposition rates; Askja and Hekla tend to deposit less than Laki and Eldgja. This confirms that, under the same conditions (temperature and velocity field), the likelihood of adhesion of one ash is different than the other, meaning that some ashes would be more dangerous for an engine than others.


2.3               Computational fluid dynamics (CFD) modelling

The numerical modelling activities have been carried out using COMSOL computational fluid dynamics software. Modelling runs have been carried out for 3 cases with boundary conditions set so as to give good agreement with the (steady state) experimental measurements of gas temperature and velocity. Comparisons are presented in Fig.4(a) and Fig.4(b) between modelled and measured thermal and velocity fields (along the tube axis) for the three cases.  It can be seen that the modelled fields are broadly consistent with the experimental data.  Fig.4(c) shows a contour map of gas velocity (for an inclined substrate), giving an indication of the nature of the disturbance introduced by the presence of the substrate. The software has also been used to simulate the temperature distribution within and trajectory of particles ingested into the gas stream.




Fig.4:  Experimental and modelled thermal and velocity fields of the gas, showing axial profiles for the three cases, with θ = 90˚, of (a) temperature and (b) velocity, while (c) is a velocity contour map in the (vertical) x-r plane, for Case C, with θ= 60˚, illustrating the perturbations caused by the substrate.


Particle thermal and velocity histories in flight have been also simulated and correlated with experimental data. Figure 5 shows the correlation between particle thermal histories and experimental deposition rates for Laki VA at different substrate temperatures for a normal incidence (θ=90°). It can be seen that deposition rates start to rise sharply as these particles start to reach temperatures (on impact) of ~700-900˚C.  This correlates well with the data shown in Fig.1c, where it can be seen that, for Laki VA, this is the range just above Tg, where, given the high glassy content of this VA (80 %, Fig. 1b), at least most of these particles are expected to become very soft.  It’s clearly not necessary for particles to reach temperatures close to or above the “melting temperature”, which in this case is about 1100˚C, for adhesion to occur.




Fig.5:  Measured adhesion rates and predicted particle parameters at impact, as a function the processing conditions (represented by the substrate temperature), showing the temperature, for normal incidence (θ=90°)


Particle velocity histories have been also correlated with adhesion rates (not shown). It can be seen that there is an increase in velocity with increasing substrate temperature but it’s not very significant and it seems likely that the sharp rise in deposition efficiency as these changes are made is primarily due to the higher particle temperatures at impact (particularly as they start to exceed Tg).

Combination of modelling and experimental work would provide a framework for assessment and prediction of the likelihood of adhesion of VA into real jet engines.


2.4         Deformation of VA at high strain rate (high velocity impact)

As shown in Fig. 3, different ashes present different deposition rates, where Laki and Eldgja tend to deposit more than Hekla and Askja. It would be expected that this will mainly depend on particle size, particle temperature and velocity, amorphous and crystalline content among others. The softening temperatures (Tg) for all the ashes are in the range of 600 -800 °C (600°C for Laki, 700°C for Askja and Eldgja and 750°C for Hekla) which correlates fairly well with the deposition ranking order. However, important differences can be observed in the deposition level of the VA which might be an indication that other ash properties should be consider such as viscosity at high temperature. To do that, a high temperature/high velocity impact test has been performed.


Fig.6:  Schematic of high temperature/high velocity impact test set-up


The set-up (Figure 6) compromises a gas gun equipment with an RF unit to heat up the ash pellet. Spherical ash pellet was produced and fired with the gas gun at approximately 100 m s-1 and temperatures of 1200-1300 °C.  The impact was recorded with a high speed camera. The high speed camera reveals that the deformation of the different ashes at high temperature and high strain rate was different from one to another. At 1300 °C, Askja behaves as a “solid” (Fig. 7a) whereas Laki presents a “liquid-like” behaviour (Fig. 7b).



Fig.7:  High speed camera images of (a) Askja and (b) Laki volcanic ash pellets during high impact tests at ~1300 °C (complete videos can be found at: High Temperature, High Speed Impact of Volcanic Ash Pellets on Static Targets, J. Dean, December 2015)


This set up will be used to evaluate the deformation at high strain rate of the different ashes using several temperatures, angles and coated and un-coated substrates.  This will give insight into the sticking of the VA to develop a realistic adhesion criterion. As a consequence, a prediction of the “danger” of certain type of ash could be predicted when ingested by a jet engine.


2.5         Effects of VA ingestion into sintering and spallation of thermal barrier coatings (TBCs)

One of the damages introduced by the ingestion of VA into the jet engines is the reduction of the lifetime of the thermal barrier coatings (TBC) because of the acceleration of its sintering, promoting spallation [5]. The most common composition of these coatings is the 8 Yttria-stabilised zirconia (YSZ), offering chemical stability, low thermal conductivity and relatively high thermal expansivity. Sintering of TBCs, with or without CMAS, leads to microstructural changes that raise the thermal conductivity [8] and the stiffness [9-11], the former caused by growth of the inter-splat contact area [12] and the latter by inter-splat locking and splat stiffening, due to microcrack healing [11]. Coatings become more brittle and less strain-tolerant as sintering proceeds, making them prone to spallation (usually as a result of differential thermal contraction stresses set up during cooling).


2.5.1     Penetration of VA into YSZ coatings

Conventional (vacuum) plasma spray (PS) equipment will be used to produce the 8YSZ-TBC. Evolution of TBC microstructure will be monitored during particulate absorption and subsequent treatments. This will be done via EDX composition profiling on transverse sections, and XRD for phase identification, after serial sectioning. Penetration of VA in the through-thickness direction will be monitored via the profile of species not present in the YSZ coatings. The XRD spectra will allow estimates of the amorphous content, as a function of depth, as well as indicating whether any new crystalline phases are formed.


2.5.2     Stiffness monitoring

The stiffness (in-plane Young’s modulus) of (detached) coatings will be measured and gives a sensitive indication of the progression of sintering. It’s also of practical importance, since, for a given misfit strain (from differential thermal contraction between coating and substrate), it determines the (in-plane) stresses created in the coating and hence the magnitude of the strain energy release rate (driving force) for debonding. It will be measured by 4-point bend testing, using a rig incorporating a scanning laser extensometer. These measurements will allow quantification of the acceleration in sintering induced by VA, at different concentrations.


2.5.3     Spallation lifetime

Spallation resistance will be assessed by exposing specimens to high temperature for extended periods, with periodic quenching. The severity of the heat treatment needed to provoke spallation is a measure of this resistance, to be studied as a function of the VA “infiltration” level. A fracture mechanics-based methodology to lifing prediction will be adopted, with the strain energy release rate for the cooling being set equal to the interfacial fracture energy.


2.5.4     Development of VA-resistant TBC formulations

In addition to improved understanding of VA attack, contributing to the development of guidelines for safe ingestion limits, coating formulations will be sought offering improved protection against such degradation. These could be aimed at reducing the “deposition efficiency” of harmful particulate and/or at countering their deleterious effects after deposition. The former would primarily relate to the free surface topography, while the latter would focus on reducing the effect of VA on sintering and hence on resistance to spallation. The most fruitful line is likely to involve species that form (stable) crystalline phases by absorption of elements present in VA responsible for the creation of a vitreous (ie liquid or grain boundary) phase - ie Al, Mg, Ca etc. Such a “scavenging” species could be introduced uniformly throughout the coating and/or in a more concentrated form at the outer surface.



[1] M. G. Dunn, "Operation of Gas Turbine Engines in an Environment Contaminated with Volcanic Ash," Journal of Turbomachinery, vol. 134, September 2012.

[2] C. G. Levi, J.W Hutchinson, M.H Vidal-Setif, C.A Johnson "Environmental Degradation of Thermal Barrier Coatings by Molten Deposits," MRS Bulletin, pp. 932-941, 2012.

[3] Song, WJ, KU Hess, DE Damby, FB Wadsworth, Y Lavallee, C Cimarelli and DB Dingwell, Fusion Characteristics of Volcanic Ash Relevant to Aviation Hazards.Geophysical research letters, 2014. 41(7): p.2326-2333.

[4] Ai, WG, N Murray, TH Fletcher, S Harding, S Lewis and JP Bons, Deposition near Film Cooling Holes on a High Pressure Turbine Vane. Journal of Turbomachinery-Transactions of the Asme, 2012. 134(4).

[5]Shinozaki, M and TW Clyne, The Effect of Vermiculite on the Degradation and Spallation of Plasma Sprayed Thermal Barrier Coatings. Surface and Coatings Technology, 2013. 216(0): p.172-177.

[6] Lee, KI, LT Wu, RT Wu and P Xiao, Mechanisms and Mitigation of Volcanic Ash Attack on Yttria Stablized Zirconia Thermal Barrier Coatings. Surface & Coatings Technology, 2014. 260: p.68-72.

[7] C. Taltavull, J. Dean & T.W. Clyne. Advanced Engineering Materials (2015) DOI: 10.1002/adem.201500371.

[8] Rahaman, MN, Gross, JR, Dutton, RE and Wang, H, Phase stability, sintering, and thermal conductivity of plasma-sprayed ZrO2-Gd2O3 compositions for potential thermal barrier coating applications. Acta Materialia, 54: (2006) 1615-1621.

[9] Tsipas, SA, Golosnoy, IO, Damani, R and Clyne, TW, The Effect of a High Thermal Gradient on Sintering and Stiffening in the Top Coat of a Thermal Barrier Coating (TBC) System. J. Therm. Spray Technol., 13: (2004) p370-376.

[10] Siebert, S, Funke, C, Vassen, R and Stover, D, Changes in Porosity and Young's Modulus due to Sintering of Plasma-Sprayed Thermal Barrier Coatings. J. Mater. Proc. Technol., 93: (1999) 217-223.

[11] Thompson, JA and Clyne, TW, The Effect of Heat Treatment on the Stiffness of Zirconia Top Coats in Plasma-Sprayed TBCs. Acta Materialia, 49: (2001) 1565-1575.

[12] Golosnoy, IO, Tsipas, SA and Clyne, TW, An Analytical Model for Simulation of Heat Flow in Plasma Sprayed Thermal Barrier Coatings. J. Ther. Spray Techn., 14: (2005) 205-214.

Dr Catalina  Taltavull
Research Associate - University of Cambridge