Considerations when Designing an Optically Accessible High-Pressure Combustion Test Article

Paper from the AFRC 2018 conference titled Considerations when Designing an Optically Accessible High-Pressure Combustion Test Article

Experimental high pressure combustion testing is necessary to evaluate computational models and better understand flame dynamics in conjunction with flame behavior in modern combustion systems. In order to accurately replicate conditions that might be found in a combustor, high pressure and high temperature test articles are required. There are many technical challenges associated with making such harsh environments while providing optical access to perform advanced diagnostic techniques. This paper will detail some of the greatest challenges and best practices associated with rig design that accommodates diagnostic techniques such as PIV and Chemiluminescence while maintaining relevant physics and conditions of a modern combustor.

Considerations when Designing an Optically Accessible High-Pressure Combustion Test Article David Wu Department of Aerospace Engineering, Georgia Institute of Technology, Atlanta, Georgia, 30332 Abstract Experimental high pressure combustion testing is necessary to evaluate computational models and better understand flame dynamics in conjunction with flame behavior in modern combustion systems. In order to accurately replicate conditions that might be found in a combustor, high pressure and high temperature test articles are required. There are many technical challenges associated with making such harsh environments while providing optical access to perform advanced diagnostic techniques. This paper will detail some of the greatest challenges and best practices associated with rig design that accommodates diagnostic techniques such as PIV and Chemiluminescence while maintaining relevant physics and conditions of a modern combustor. Introduction Combustion systems are getting ever more complex as there is a push for efficiency and lower emissions. These systems have quite a few technical hurdles, in particular, damaging combustion instabilities, as well as, higher heat loads that require more unique cooling methods or exotic materials. As combustion systems become more sophisticated, more intricate methods of design are needed to mitigate against these challenges. The primary tool to resolve these issues is computer assisted design with physics based models. These models need to be validated with data from engine tests or experiments, which can prove quite difficult at the higher temperature and pressure conditions of a modern combustor. This paper will be addressing some of the considerations necessary when designing an optically accessible test article for the purpose of simulation validation and/or investigating fundamental combustion phenomena. This will be a general overview of some of the biggest challenges with an optical test article from my personal experience. That being said, these will vary from different experimental setups with different goals and might not be relevant for some. Overall, though, most test articles have three primary data collection goals. The first being collecting accurate boundary conditions for the simulation where possible. The second being flame properties such as flame shape, velocities, and heat release distribution from the main combustor region to validate the predictive capability of the model. The last being exhaust properties, such as, emissions and thermal properties from distribution to overall maximum numbers for survivability of downstream equipment; which also validates the accuracy that the model can simulate an actual combustion system. Reverse Flow vs. Direct Flow Pressure Vessel Before I discuss some of the measurement techniques and challenges, let me introduce you to two common pressure vessel designs we employ at the Ben T. Zinn Combustion Lab at Georgia Institute of Technology. The first is a direct flow pressure vessel that does not use process air for cooling purposes, as can be seen below in Figure 1. The second is a reverse flow pressure vessel that uses process air to help cool the test article and increase the preheat for process air, which is shown below in Figure 2. There are many other designs and techniques for pressure vessel design, but the afore mentioned are very common for continuous combustion testing. Figure 1 Direct Flow Pressure Vessel with Cooling Flow and Water Jacket Figure 2 Reverse Flow Pressure Vessel with Water Jacket In Figure 1, above, you can see one of our direct flow pressure vessels, which is the simplest, but has its drawbacks. Overall, the cooling flow can be mixed with the process flow post combustion or vented separately in the pressure vessel. This experimental setup, though, has difficulty reaching higher preheat temperatures and has higher air flow requirements due to the process air and cooling air are supplied separately. This pressure vessel configuration is easier to control than the reverse flow configuration due to the independently controlled air sources. In Figure 2, above, is one of our larger reverse flow pressure vessels. In order to get to higher preheat values, the process air partially cools the combustion chamber while elevating the preheat temperature. Generally, this setup usually gives an extra one-hundred degrees for preheat, depending on which flow condition and global equivalence ratio is being tested. The major downside to this setup is if there is a specified preheat temperature, it can be very difficult to hold a stable preheat temperature; especially over a variety of flow and firing conditions. Both experimental setups above implement water jacket exhausts. Exhaust orifices can be changed between runs to influence pressure and nozzle velocity operating conditions. These water jacketed exhausts have to be serviced regularly due to the huge thermal gradients they experience cracks and warping over time. The warping and crack effects can be mitigated through monitoring the water temperatures and keeping them stable throughout a test campaign. There is a boost pump attached when necessary to increase flowrates. Measuring Boundary Conditions Depending on the type of test article being built and how the simulation model is simplified, will determine what boundary conditions are most important. Overall, inlet conditions are easiest to characterize during a test with a combination of thermocouples, pressure transducers (static and dp), and acoustic pressure transducers. Other values of interest for inlet conditions are flow conditions, such as, turbulence and overall velocities. At Georgia Tech we have made those measurements either with Particle Image Velocimetry (PIV), Hot Wire, and Pitot Probes. Some of these measurements can also be taken on a flow bench, such as, effective blockage ratio for unique swirler geometries and/or full nozzle assemblies. The more difficult measurements to take are for the boundary conditions of the combustion walls. Due to the high temperatures of flame conditions, directly attaching thermocouples to combustor walls is infeasible. Thermocouples we have attempted to place in combustor liner walls typically will last one or two tests, at most, then require replacement. Other techniques that have given limited success are temperature sensitive paints, temperature sensitive crystals, and optical techniques with both short wave and long wave IR. The issue here is that these techniques are generally not any more accurate than an estimated wall temperature and have other shortcomings, depending how they are deployed. Flame Property Measurements Flame measurements are the most important function of optically accessible high pressure combustion experiments. Although, there are many considerations to be mindful of when designing such a test article. Depending on the diagnostic technique desired and the combustion test's conditions, will determine any special considerations you might need to take. This paper will examine some of the more common issues that need to be addressed, particularly concerning PIV measurements. The first consideration when designing a new test article, is to determine which fuel type will be used in the test campaign. The major problem being soot generation, which will either block or alter an optical measurement. For instance, soot, under certain conditions, will fluoresce; giving a stronger signal than what is actually occurring. The most common problem with soot, though, is that it obscures the optical view, as can be seen in Figure 3 below. Figure 3 Liquid Testing no Window Purge The test showcased above, in Figure 3, was a swirl stabilized liquid Jet A test. Commonly, during combustor tests, especially at higher equivalence ratios we would see a large amount of buildup on the quartz windows; which we speculated to either be fuel coking or soot deposits forming on the windows. In order to combat the material build up, we developed a window purge system using a series of small jets along the windows. Figure 4 Dump Plane Liquid Swirl Experiment Each side of the combustion chamber has its own set of jets independently supplied. These jets can be supplied with any gas, from air to inert gases such as nitrogen. In figure 5, below, (which was taken under the same conditions as figure 3) is with the window purge system in use and using nitrogen as the supply gas. The window purge does not completely solve the issue, but does allow us to have a better field of view. Also, what cannot be shown in these images, is that at a certain test conditions, we are able to test longer with minimal amounts of the window being obscured. Figure 5 Liquid Testing with Window Purge Optical issues are not the only problem when performing optical diagnostics. PIV is quite susceptible to reflections ruining data quality. In figure 6, below, I have showcased a jet in cross flow experiment with a vitiated flow. This image is from an early attempt to take some PIV data that had several reflections present. The bright line toward the bottom of the image is a reflection of the bottom of the test article. Figure 6 Jet In Crossflow with Large Reflection To diminish this reflection, there are a number of paths you can take. The best path would be to replace the bottom of the test article with a quartz plate to allow the laser sheet to pass straight through your field of interest. In this particular experiment that was not possible due to the need for a secondary fuel inlet for the jet in crossflow study. As can be seen below in Figure 7, the reflections have been decreased to allow us to see the jet at the lower right corner. The reflections were not completely eliminated, but were reduced by painting the bottom plate of the test article with a light absorbing paint. The rest of the reflections were removed with calibration images and correcting the data. Figure 7 Jet in Crossflow PIV Reduced Reflections PIV issues are not only caused by reflections, though. Other issues might be seeding density or distortion from beam steering and quartz windows clouding during testing. Beam steering can be particularly troublesome, although, with careful calibration and forethought, it can be overcome. For example, in a direct flow experiment, as can be seen in Figure 1, there are three different temperature regimes that a laser would have to pass through. The first being ambient air temperature and then the cooling flow temperature; which might be only slightly higher than ambient or a much different temperature depending on the application. The last zone would be the combustion region which will be orders of magnitude higher than the other two zones. This can cause some substantial beam steering which will be difficult to calibrate for, as these conditions would only exist during testing. The other major cause of beam steering is the quartz itself. So, considerations should be taken in the geometry of the quartz combustion liners. The use of square quartz tubes is possible with minimal beam steering, but can prove fragile with much longer lead times than circular quartz tubes. Circular quartz tubes have a lensing effect on laser sheets which become much worse the smaller the quartz tubes are. Cost and difficulty in operating increases the larger the quartz tubes installed are. So, careful thought should be used when choosing a combustor geometry. Figure 8 PIV Measurement Insufficient Seed Density and Window Fogging As eluded previously, seed density and quartz optical quality are imperative to a good PIV measurement. Figure 8 is an image of a project we had that had major difficulties with window fogging and difficulty maintaining seed density. This was due to the high temperatures that the quartz reached. During a PIV campaign in a reacting flow, the quartz temperatures became high and seed particulate in swirl flows were impacting the glass at great speed. This causes the seed to either attach, imbed, or score the quartz in a short period of time. A large portion of the fogging you see is due to seed adhering to the quartz during the test. Care should be taken to design either a cleaning method or ease of access to the interior of the combustion vessel to facilitate cleaning due to poor visibility and reduce downtime. Overall, though, due to scoring and seed imbedding in the quartz, the amount of tests per piece of quartz can be quite limited. Figure 9, below, shows a data case with good seed density to observe large flow structures and minimal reflections. Figure 9 PIV Data with Large Flow Structures Just to show an example of what occurs when poor PIV data is processed, Figure 10 shows stereographic PIV data processed with inadequate seed density. The exact cause is difficult to determine without looking at a flame location technique, but likely, either the swirl strength of the vortexes is centrifuging the seed out of the desired region or the seed is being accelerated by the flame sheet; which is causing low seed density in the desired area. Figure 10 Processed Stereo PIV data with Insuffecient Seed Density The data in Figure 10 was recaptured with better seed distribution to produce the image in Figure 11. As you can tell, there are no holes in the new processed data in the flow field and gives you a more complete picture of what has happened. Also, the velocities in and out of the plane are more consistent with what you would expect in a swirl stabilized flame with a precessing vortex core. Figure 11 Processed Stereo PIV data Overall, careful consideration is necessary when choosing a combustor geometry and how you plan to seed your test article if PIV is to be a major part of your test campaign. One last thing to consider, is how to layout your test article. An intense test campaign with multiple optical techniques, can take up quite a bit of space and might block each other or obscure the field of view. Figure 12 Direct Flow Test Article Wrapped with Simultaneous Stereo PIV/ PLIF/ High Speed Chemiluminescence The images above, in Figure 12, are two different views of our direct flow test article instrumented for simultaneous stereo PIV, high speed planar laser-induced fluorescence (PLIF), and high speed chemiluminescence. This setup was particularly difficult due to the number optical techniques being employed in a high pressure liquid spray test article. In addition to the difficulties caused by the limited space around the experiment, was the harsh conditions that the experiment caused. Due to vibration and noise from the experiment, both the PIV and PLIF laser had to either be shielded in protective box enclosures or ported in from outside our high pressure test cell. Where the optics for the laser sheets was placed had been determined by both the field of view desired and spacing to allow the crane to get in for rig disassembly without disturbing the optics setup. So, when designing a new facility, caution should be taken when considering which optical techniques might be employed on your test article and how you will work around or with the necessary optics when doing repairs and maintenance on your test article. Conclusions The challenges and solutions discussed in this paper only scratch the surface of what an experimentalist would encounter during an experimental test campaign while investigating combustion phenomena. Nevertheless, there will be unique problems and challenges an experimentalist will have to overcome. When facing any particular challenge or when designing a new test article, keep in consideration every way that the test article might be used and, most importantly, try to keep it as simple as possible to allow room for improvisation in the event of an unexpected problem. Acknowledgements This research was partially funded by the Air Force Office of Scientific Research (Dr C. Li, Program Officer, 14RQ06COR), the University Turbine Systems Research (Contract Nos. DEFE0025344 and DE-FE0031288), contract monitors Dr. Mark Freeman and Omer Bakshi, the National Science Foundation (Contract No. 1705649), contract monitor, Dr. Song-Charng Kong, and the and the US Federal Aviation Administration (FAA) Office of Environmental and Energy as part of ASCENT Project 27A under FAA Award Number 130C0AJFE-GIT-008. Any opinions, findings, and conclusions or recommendations expressed in this material are those of the authors and do not necessarily reflect the views of the FAA or other ASCENT Sponsors.