Deposition of Volcanic Ash within Gas Turbine Aeroengines


This report presents experimental investigations of the high temperature, high strain rate deformation behaviour of four types of Volcanic Ash (VA) collected from four different Iceland volcanic eruptions. Near spherical, non-porous, VA pellets (~ 6 mm diameter) of all four ashes were heated to different a range of temperatures (~1200 °C – 1500 °C) and fired at high velocity (~ 100 m s-1) towards a stainless steel target using a high pressure gas gun. The deformation behaviour of the different VA types was recorded using high speed photography. The experiments were repeated, with the target substrate also heated to high temperatures (~ 500 °C). It was observed that the behaviour of the ash was independent of substrate temperature as was predicted using a simple analytical heat flow model.
The four ashes were separated into two distinct groups based on their composition. Two ashes contained high Si concentrations (>20 at%) and two ashes contained lower levels of Si (>20 at%) with higher levels of divalent metal cations including Ca2+, Mg2+ and Fe2+.  The deformation behaviour of the 4 ashes was seen to fall into two distinct regimes which corresponded with the compositional groupings described. At high strain rates, the ashes with the higher Si content exhibited much higher viscosities and behaved visco-elastically on impact. The ashes with lower Si content displayed much lower viscosities, and behaved much more like a fluid. This finding helps further the understanding of the measured deposition behaviour of the four different ashes, as described by Dean et al [27], and provides an explanation that could not have been possible with a simple analysis of the glass transition (Tg) temperatures.

Contents Page

1. Introduction and Review of Literature
2. Materials and Methods
3. Results
4. Discussion
5. Future Work
6. Conclusions
7. References

1. Introduction and Review of Literature

1.1 Background
It has long been known that the presence of Volcanic Ash (VA) can damage jet aero engines that pass through polluted atmospheres. The extent of the damage caused can vary, and in the past VA has caused a number of high profile aviation accidents. The details of 7 of the more notable encounters between aircraft and VA are highlighted in Table 1 [1]. Interest in this area of research has grown steadily over the past 20 years as the problems VA presents become more widely encountered and reported.
The determination of safe flying conditions is currently controlled by the UK Civil Aviation Authority, who describe a “safe” flying atmosphere being one in which the concentration of volcanic ash is below 2 mg m-3 [2]. Crucially, the guidelines currently make no distinction between ashes of different composition or different particle size distribution. A study of the literature suggests that both of these parameters could affect the likelihood of a particular VA causing damage. Sections 1.3.2 and 1.3.3. 
The effect of this unspecific prohibition on flying can be extremely detrimental to both the aviation industry and consumers. The Eyjafjallajökull eruption of 2010 was the most recent incident of a volcanic event leading to the closure of vast regions of airspace. The data shown in Table 2 [3] serve to highlight just how damaging the social and economic impacts of air space closure can be to both the aviation industry and consumers. A better understanding of the different VA parameters important when determining likely levels of engine damage could allow for a modification to the current, non-specific blanket ban. This could not only make flying safer, but, in certain circumstances could permit safe flight that otherwise would have been prohibited under the current regulations. This has obvious social and economic benefits.

1.2 Publications from Industry

The obvious benefits arising from this field of research has promoted significant interest from outside of academia. The difficulties that VA present to pilots were reported as early as 1992 with Boeing releasing a video [4] highlighting just how difficult it is to circumnavigate polluted atmospheres. Even today there is little evidence to suggest that detection of VA and subsequent alteration of the flight path could help make flight through VA clouds safer. 
A series of “Safe to Fly” charts published by Rolls Royce reflect advances in VA understanding. These charts demonstrate how Rolls Royce, the primary manufacturer of jet aeroengines, currently view the safe limits of operation. The chart in Figure 1 [5], first published in 2012, shows a reduction in the uncertainty of the recommendations when compared to similar figures released in 2010 [6].
Following requests from the UK MoD, Rolls Royce also report on a design of a new plane which is “unaffected [when flying] through volcanic ash.” This provides significant scientific opportunities for more in depth analysis as this was used to collect samples during the 2010 Icelandic eruption [7]. Prior to this development in the technology, collecting samples and data relating to atmospheric changes following volcanic eruptions was difficult. The UK Met Office has, in the past, used the Numerical Atmospheric-dispersion Environment (NAME) to estimate the likely VA concentrations experienced during known encounters between aircraft and VA [8]. Despite the success of this modelling technique it will not be as accurate as collecting samples directly using drones or planes.

1.2 Different Damage Mechanisms.
Ingestion of any foreign material into jet engines has the potential to cause damage. Some early literature published in this area reports four different damage mechanisms which can be caused by foreign material in the atmosphere [9, 10, 11]
1. Erosion of Compressor and Turbine Components
2. Deposition on hot section components
3. Deposition on fuel nozzles
4. Deposition on vane and blade cooling holes.
The majority of non-volcanic material present in relevant atmospheres has melting temperatures that ensure this type of material is unlikely to soften as it is ingested through the engine. VA, on the other hand, softens at relatively low temperatures (< 1300°C) [12] [13]. The increased likelihood of softening and subsequent adhering of material to solid surfaces in the engine makes VA particularly hazardous.
As engine operation temperatures have continued to increase, the deposition of soft and “sticky” particulate matter is now understood to be the most prominent damage mechanism experienced by engines traversing VA atmospheres. It is this damage mechanism that has been investigated in this project.

1.3 Engine Considerations
Having established further the mechanisms by which VA causes damage, there was a focus within the literature to evaluate what ash characteristics and engine operation conditions were more likely to increase VA susceptibility.
1.3.1 Engine Temperature
Considering the importance of the softening temperatures of foreign material, it was hypothesised that the temperature of engine operation would be a key variable affecting the likelihood and extent of deposition damage. Indeed, work published by Kim et al [14] Crosby et al [15] and Bons et al [16] found that the deposition rate of certain types of VA did increase with an increased temperature. This result has been confirmed across the literature. The greatest challenge in investigating this relationship was devising the most appropriate experimental model of a full sized, commercial engine. The use of an “accelerated deposition facility” in the work reported by Crosby was a more realistic system to the “Hot Section Test System” used by Kim.
1.3.2 Surface Roughness
As well as temperature, Wammack [17] reports on how the surface roughness of engine components can change VA deposition behaviour. It was found that rougher components showed an increased chance of deposition occurring. A similar relationship was reported by Shinozaki [13] who showed qualitatively that deposition was more likely to occur on regions of existing deposition, which could be likened to an increase in roughening on the surface.
1.3.3 Susceptible Engine Components
The temperature and roughness relationships as reported in [13 – 17] were correlated with work which looked at specific regions within engines where deposition is more often located. Figure 2 is included as a schematic highlighting where these key components are commonly located [18]. The hottest part of the engine is the combustion chamber, experiencing temperatures in excess of 1600°C. Brun [19] confirms that deposits do often form within the combustion chamber, and in some cases can be so extreme as to plug the cooling holes completely. Similarly, Ai [20], reports on how deposition is also commonly located on blade surfaces and nozzle guide veins.
1.3.4 Thermal Barrier Coatings
It is clear how the deposition described by Brun et al and Ai in section 1.3.3 can lead to engine failure, for example, overheating arising from blocked cooling holes. The types of deformation and engine failure described in section 1.3.3 are associated with deposition in larger volumes.  Lower levels of deposition can also be problematic. In these sorts of cases however, engine damage would more commonly be associated with the spallation of the thermal barrier coatings (TBC) protecting metallic engine components. TBCs are of great interest and have been widely studied. The specific response of TBC to VA deposition is reported by Shinozaki [21] Shulz [22] and Lee [23]. !!!! Drexler [24] reports specifically on TBCs…..

1.4 Ash Properties
Section 1.3 discussed the different engine characteristics that can affect deposition and describes to what extent different levels of deposition can lead to damage. This understanding has obvious value in explaining in greater detail the mechanisms surrounding VA damage. Despite this, the literature is somewhat limited in the useful information it can provide to the aviation industry on preventing disruption following volcanic eruptions. The operating conditions and construction of modern engines is now so finely tuned for efficiency and performance that it is unlikely significant tolerances can be manufactured to cope with ash deposition. It is necessary, therefore to perform a full investigation on the effects of different ash cloud properties to determine whether some atmospheric conditions could be safer than others.
1.4.1 Particle Size Distribution
It has long been known the concentration of VA particulates in the atmosphere is of primary concern when determining safe flying conditions. The negative effect of an increasing concentration has resulted in the current guidelines being based solely on this parameter.
In more recent literature, however, it has become apparent that the properties of the VA itself can have a significant impact on its propensity to inflict damage.
The particle size distribution of ash clouds is now understood to have significant importance in determining the risk of damage. Work published by Taltavull [25] and Shinozaki [13] broadly agree on the relationship between PSD and deposition rate. Shinozaki described, fairly qualitatively that as the average particle size is increased, the rate of deposition also increases, as is shown in Figures 3.1 – 3.4. The more recent work performed by Taltavull goes further in examining the relationship between deposition rate and PSD; a more quantitative analysis is presented. It is reported that particles between ~10-30 µm are most likely to deposit. Using a number of modelling techniques, it is also predicted that small particles (~2 µm – 3 µm) are unlikely to strike surfaces and particles beyond a critical size (>100 µm) are unlikely to reach the temperatures required for deposition.
1.4.2 Glass Transition Temperature
As described in Section 1.2, the softening of VA at relatively low temperatures (compared to other non-volcanic material) is the reason for the increased risk of deposition damage. Song [26] highlights that the problem of VA deposition is exacerbated by the softening of the VA at temperatures below the melting point. The high volume fraction of amorphous phase observed in some ash samples can cause VA to soften significantly as it is heated through a glass transition temperature (Tg). The Tg values are likely to be lower than the typical operational temperature of a jet engine and hence pose a problem. Shinozaki [13] confirms this issue by publishing the X-Ray diffraction pattern (Figure 4) and DSC trace (Figure 5) of a particular VA showing the significant volume of amorphous phase (Vα) and corresponding Tg value of ~ 600°C.
It could be concluded from this analysis that the proportion of amorphous phase, and Tg value would be important characteristics. Both papers, however, only investigate ash of only one composition. A natural progression from this work would be to investigate ashes of different compositions and to confirm any relationships with Tg or Vα.

1.5 Most recent work
Until recently the behaviour of ashes of different glass transition temperatures had not been reported in any detail. Recent work by Dean et al [27] has attempted to understand further the effect of differing ash compositions on deposition characteristics. It is from this work, performed in the Gordon Laboratory, that this project directly builds upon, and as such this paper has been examined in some detail.
The work reported by Dean et al uses the same four ash types, collected from different Icelandic volcanoes, that were used for investigation into this project. A brief geological description of each eruption is as follows:
1. Laki – fissure eruption in 1783-4
2. Eldgja – fissure eruption of 934
3. Hekla – highly active strato volcano erupted in 2000
4. Askja – 1875
For the purposes of this report the different ash types investigated will be labelled using the name of the eruption source.
Table 3 shows the Tg values, melting temperatures (Tm) and Vα proportions of the 4 ashes. (A full detailed chemical and physical analysis of the four ash types is presented in Section 2.1) Having analysed the relevant values from Table 3 and consulted the literature relating to the glassy transition characteristics (Section 1.4.2) it was predicted that:
1. The four ashes would not show large differences in deposition behaviour as the differences in glass transition temperature are small.
2. If a significant experimental difference was to be observes, it would likely show that Hekla and Askja would exhibiting the highest deposition rates due to the significantly greater Vα values. 
Figure 6 shows the measured deposition rates as a function of temperature for the 4 different ash types. This experiment was performed using a constant particle size of d = 25 µm. As can be seen from this figure the predictions above are not accurate:
1. There are large differences in deposition rates
2. Hekla and Askja actually exhibit the lowest deposition rates.
It was concluded that information relating to the glass transition temperature is, alone, insufficient to describe the high temperature, high strain rate, deformation and deposition behaviour of volcanic ash. There must be other reasons why differences in VA composition result in different deposition behaviour.
1.6 Aims of this project:
As described in Section 1.5, the most recent work conducted within the department has failed to fully explain the deposition behaviour of four different ash types. This would suggest there is a fundamental lack of understanding in the mechanism of deposition.
The aims of this project are:
1. To identify the qualitative differences in the dynamic behaviour of four different ashes in high temperature, high velocity deformation events.
2. To characterise further the properties of the different ashes to understand better the deposition behaviour already observed.
The project aims to build upon work already published and work investigated in the department. The true long term aim of the project is to contribute to a better understanding of the dangers of flying aeroengines through ash particulate matter. In turn there could be a modification to the legislative rules that permit flight in certain circumstances and reduce the damage volcanic eruptions can have on the aviation industry.

2. Materials and Methods
2.1 Description of Key Materials and Key Operational Units
2.1.1 Volcanic Ash
The four different volcanic ashes were collected by Dr Margaret Hartley from the department of Earth Sciences, The University of Cambridge. Prior to beginning this project, Dean et al [27] had performed a detailed chemical and physical analysis of the four ash types using a number of different techniques. These characterisations are crucial in explaining the observed deformation behaviour described in Section 3.XX, and are highlighted below. Composition
EDX techniques were used to investigate the bulk composition of the four ash types (Figure 7). For the purposes of this investigation it was useful to separate the four ashes into two different pairs, based on differences and similarities in their composition. The details of the two groups, and the constituent ash types is summarised in Table 4.
Figure 7.
Table 4. Volume fractions of amorphous and crystalline phase
X-Ray diffraction was used to determine the volume fraction of amorphous (Vα) and crystalline (Vc) phases present in the material. Figure 8 shows the measured diffraction spectra of the four different ashes.
Figure 8.
Hekla and Askja are composed of close to 100% amorphous phase. Eldgja and Laki do contain some crystalline content (~ 60% and ~ 80% respectively.) This grouping is consistent with the compositional groupings described in Table 4. Glass transition temps (as above)
Dean et al used a Netzsch Dilatometer to measure Tg values of the four ashes. Figure 9 shows the DSC plot of displacement against temperature for a ramp of 5 °C min-1. Where relevant, Tm values have also been included. Of course, the fully amorphous Hekla and Askja ashes do not show a physically significant melting temperature.
Figure 9.
2.1.2 Gas Gun
A high power pressurised gas gun (GG) was used to project VA pellets at a stainless steel target substrate. The gun uses a two chamber system and pressurised gas (~ 25 MPa ???) to fire small projectiles along a steel barrel and towards a target. The GG was required to fire the pellets at  velocities that would generate the appropriate impact strain rate regime.  A labelled picture of the Gun is illustrated in Figure 10.
2.1.3 Camera and Lights
High speed photography was used to record the impacts events and associated software used to analyse the moving images. A XXX High Speed camera was orientated approximately perpendicular to the flight of the pellets. It was used to video record the latter stages of the flight of the pellet and its subsequent impact and rebound following collision with the target.  
The recorded impact velocities were ~100 m s-1, although full details of the impact velocities for each trials are included in Section RESULTS. To successfully capture sufficient detail at this velocity, a powerful light source was required to illuminate the moving projectile and the target. Two high power XXX LED ARRAYS were used in conjunction with the camera. 
The XXX software was used to analyse the impact and rebound velocities of the projectiles, as well as measuring the approximate time of contact between the projectile and the plate.
2.1.4 Induction heater
For the majority of the project the VA pellets were required to be heated. A Cheltenham Induction Heating Ltd system, operating at different frequencies was used to heat the VA. Section 2.2.5 describes in detail how this was achieved.

2.2 Experimental Methods
2.2.1 Method of Camera optimisation
Before attempting to obtain moving images of high temperature any ash deformation, the arrangement and settings of the camera, lights and GG needed to be optimised. To avoid wasting the limited supply of ash powder, steel ball bearings were used initially as test projectiles. These steel ball bearings were also more reflective and were easier to capture than the duller VA pellets.
The quality of the images recorded depended heavily on the camera parameters chosen. It was possible to alter the frame rate (number of still images recorded per unit time), the resolution (number of pixels in the filmed region) and the exposure time (the time the shutter is open for every still recorded). There were significant trade-offs with these parameters; a lower exposure time, for example would result in a darker, yet more well defined image.
Figures  11.1 – 11.x shows the recorded impact for one of the first ball bearing impact tests. It showed that good quality images were possible with impact velocities up to XXX m s-1. It was apparent that achieving impact velocities of VA in the region of 100 m s-1 and successfully recording the impact behaviour was possible, and moving forward with the experiments was suitable. 100 m s-1 was chosen as this value was similar the velocities used by Dean et al to produce the deposition plots shown in Figure XX.
2.2.2 First Approximation of High Temperature Ash Deformation
As was stated in Section 1.6, it was planned, in the latter stages of the project, to build a numerical model of the deformation behaviour of VA at high temperatures and infer the key properties affecting deposition damage.
Cast lead pellets were used as the first approximation to the behaviour of ash deformation at high temperatures. As a soft metal it was thought that the behaviour of lead may in some way replicate that of VA at high temperatures. A numerical model was built using the ABAQUS FE Modelling software in order to predict the behaviour of impact. The recorded deformation behaviour and corresponding modelling prediction of the lead impact trial are shown in Figures 12.1 - 12.4 and 13.1 – 13.4 respectively.
Figures 12 & 13.
It was shown that the model could be used to predict the shape of deformation and approximate rebound velocities. As such, it was planned that this model could be modified in the future to reflect the VA behaviour.
2.2.3 Ash Pellet formation and Room Temperature trials
The initial method of producing ash projectiles involved the pressing of ash powder in a cylindrical mould to ~ XXX Pa for 10 secs followed by a sintering process of heating to XXX for XX. This was successful in forming dense cylindrical pellets of length ~ 1 mm ?? and diameter ~ 2 mm ??. A number of these pellets were tested in impact at room temperature to further refine the camera and software settings. The experimental arrangement used to perform these initial trials is shown in Figure 14 . As was expected all four ashes exhibited catastrophic failure on impact and as such there are no observable differences in the behaviour. (Figures 15.1 – 15.4).
Fig 14
Fig 15
It was noted at this point that the angle that the cylindrical pellets struck the substrate was varying significantly between trials. In fact, rolling of the pellet after exiting the GG barrel meant that the pellets were very rarely striking the surface of the target substrate normally. Figures 16.1 – 16.4 shows how this is the case. This loss of axial symmetry would provide a large cost to the 3D numerical modelling, and would make it more difficult to compare the behaviour of the different ashes when they were taken to high temperatures. To resolve this issue, a new method for producing spherical ash projectile was devised.
Figs 16xxx
Near spherical pellets were formed by placing ash powder (< 50 µm) into specially made silicone moulds. The silicone moulds took the shape of a hollow sphere, with an internal radius of ~4 mm. The external diameter and varied between the different moulds but was largely unimportant. The filled silicone moulds were placed in a Stansted Fluid Power Press and subjected to pressures of 1500 bar for 5 mins. The near spherical VA agglomerates were removed from the silicone moulds and sintered in air at 850°C for 30 mins (DO I NEED TO REFERENCE THIS??). A reduction in volume of the pellets was observed. The spherical shape was retained with pellets having measured diameters of ~ 6.5 mm.
2.2.4 Heating the Pellets in Situ
Having successfully recorded images of ash deformation at room temperature and having obtained dense, and spherical ash projectiles for each of the four ashes, the high temperature testing could begin. As was highlighted in Section 2.1.4 a portable induction heater was the device used to achieve VA temperatures in excess of 1500°C.
There were a number of challenges in modifying the existing experimental arrangement used to produce the room temperature impact images shown in Figures 15 & 16. The first stage was to replace a section of the steel GG barrel with an Alumina tube of approximately equal internal and external diameter. A ceramic was needed to replace the steel, which would otherwise have melted or suffered severe creep at the temperatures experienced during the trial.
The Alumina tube was manufactured with a thin graphite sleeve with an internal radius just larger than that of the spherical ash pellets. It was found that the induction heater coupled particularly well with the graphite sleeve which in turn heated the inserted VA pellet through a radiative process. The target was brought much closer to the firing chamber to prevent the VA pellet, now at high temperatures, breaking as it moved through the chamber. The induction coil was placed around the Alumina tube, ensuring the position of the inserted VA pellet would fall roughly in the middle of the coil. A picture and corresponding schematic of the completed experimental arrangement is shown in Figures 17 & 18 respectively.
Figure 17
Figure 18
2.2.5 Brief Summary of Experimental Method (IS THIS REQUIRED??)

- Spherical VA pellets were produced using isostatic pressing and sintering
- An alumina tube was used to replace a section of the GG, and contained a thin graphite sleeve
- The VA pellets were placed within the graphite sleeve
- An induction heater coil was placed around the ceramic tube covering the pellets in situ
- Pellets of all four ashes were heated to a range of different temperatures and fired at a stainless steel target using the GG
- High speed photography was used to record the impact behaviour.

2.2.6 Method of Projectile Temperature Calibration
One of the more significant experimental challenges faced when designing the high temperature trials was the accurate determination of the temperature of the VA pellet on impact. A direct measurement of the temperature of the pellet just prior to impact was extremely difficult. Direct, mechanical contact with a thermocouple would be impossible and a measurement of temperature using optical techniques would be far too inaccurate at the velocities used.
Even a measurement of the temperature of the pellet just prior to firing was difficult to obtain. There was no way of arranging the thermocouple such that it could be in good thermal contact with a pellet that was about to be fired. 
It was decided that the most appropriate method would be to run a series of calibration tests. A small hole was drilled into the back of a sample pellet and a K-Type thermocouple fixed in place using XXXX adhesive. This pellet was placed in the ceramic tube and heated at different induction voltages for APPROX 10 minutes. The change in temperature of the pellet over this time period was recorded using the PicoLog software. Figure 19 shows an example of the temperature profile measured for XX mins at XX V. It was observed that after ~10 mins at each voltage, the temperature of the pellet had stabilised so that it was reasonable t