Night firing of the SSME at A-3 test stand at the Santa Susana Field Laboratory (SSFL) in California. The orange glow from the clouds is a result of sodium emissions in the exhaust plume
The exhaust species from a highly efficient oxygen/hydrogen propellant engine, such as the SSME, are primarily water vapor (H2O) and a small amount of residual hydrogen (H) and hydroxide (OH). The flame seen is the byproduct of both thermal and chemiluminescent excitation processes occurring in the main combustion chamber and continuing out the nozzle exit.
The radiated energy from the hot plume is predominantly the blackbody radiation from water vapor emission bands throughout the infrared wave-lengths, while the OH molecules radiate strongly in the near ultraviolet. The combustion process is certainly energetic, but it is not sufficiently energetic to excite the hydrogen to radiate in the visible waveband (i.e., there are no Ha or Hb spectral lines in the plume, as are observed in stars). Theoretically, there is nothing to be seen of the plume in the visual wavelengths, and yet the exhaust plume has an orange-like glow and a bright region a few feet below the nozzle lip called the Mach disk. What is it that is actually seen?
Of the numerous shock structures observed in the exhaust plume, the brightest and most visible is the Mach disk, where hot combustion gases exiting the nozzle at supersonic speeds (up to about 3 kilometers per second) slow to subsonic speeds and heat up rapidly. The brightness seen is the continuum blackbody glow of the exhaust species at temperatures above 5000°F.
The plume light from both the Mach disk and the region above it are quantitatively analyzed by real time spectrometer systems to discover which wavelengths are emitted by the exhaust flame. The observed emission spectra from the SSME reveals a wealth of other chemical species whose identification is determined from their emissions at distinct (known) wavelengths.
In the visible and near infrared wavebands, for example, the most prominent spectral lines are those of sodium and potassium, and it is the sodium D-lines that make the orangish glow we see, like the low pressure sodium lamps used for street lighting. Spectral emission bands of calcium hydroxide are also standard features of the SSME spectra, as are less bright emissions due to iron, chromium, nickel, manganese, copper, calcium and other metals. Where do these species come from and why do they appear in the SSME plume?
Additionally, test observers carefully watch the plume during engine hot firings to note the appearance of intermittent colored streaking within the plume or unstable flame structure; excessive or continuous streaking can cause an observer to manually cut off the engine test to minimize engine damage. What causes the streaking, and what significance does this have? The answers to these questions reveal a remarkable opportunity for the health and condition monitoring of rocket engines in real time.
For a reusable rocket engine, such as the SSME, engine component wear or erosion is carefully monitored to establish service life limits. Engine hardware is designed to be surrounded by or immersed in cryogenic liquids, cold and hot reactive gases, propellant combustion products, high pressures, and in the case of turbopump components, they must endure rapid rotation in these environments.
Traditionally, engine performance data is analyzed, deviations from nominal operation are identified and component disassembly and inspection is performed in order to characterize and measure the extent of degradation within the complex structures. This disassembly and inspection approach is costly and time consuming, and afterwards care must be taken to ensure correct reassembly of the engine component. To date, this has been the most direct way to establish average wear rates and operational limits for these complex systems.
Now, however, health and condition monitoring systems are being developed to continually assess engine hardware and operational status in real time. These systems depend on new sensor technologies, multiple sensor data correlations, and advanced analysis techniques applied to conventional engine data for accurate characterization of the engine. Ideally, engine component replacement and disassembly will be performed only when required, thus avoiding the need to disassemble and inspect the engine component at regular intervals. For example, the monitored vibration data from a turbopump during a Space Shuttle flight could serve as the qualifying test of that turbopump for the next flight, circumventing costly disassembly and inspection.
If an engine component wears under the extreme environmental conditions within a rocket engine, where does the worn material go? The amount of material lost during nominal conditions is usually very small, and it is unlikely that this material will remain within the engine. The tremendous propellant flow rates through the engine tend to flush out most of it. Eventually, the propellants and any material released by wear or erosion will reach the main combustion chamber, undergo combustion along with the propellants, and be accelerated rapidly through the nozzle. The fact that this lost material ends up in the exhaust plume opens up a new understanding: spectrometric monitoring of the plume can reveal the presence of small amounts of material lost from the engine as the erosion occurs.
The concept of applying emission spectrometry to rocket engine plumes is derived from the proven chemical analysis technique of flame spectroscopy, where a researcher places a small amount of an unknown material into a flame and observes the flame colors produced as the substance burns. The human eye is extremely color sensitive, but the accuracy of chemical determination by this technique was improved more by experience than anything else, since many substances can produce similar flame colors and thus mimic each other.
The use of prism and grating spectrometer systems enabled the particular color sequences, or spectra, to be recorded on film and quantitatively analyzed and compared. It was found that each element radiates or emits light at particular distinct wavelengths, due to its specific atomic energy level structure. Proper equipment made it possible to distinguish between the subtle color differences of mimicking elements and to identify precisely the component elements within a sample. The relative brightnesses of the spectral lines allowed the researcher to determine the relative amounts of different elements within a mixed sample.
This technique is now applied to the field testing of rocket engines by treating the engine plume as a precisely controlled laboratory flame for chemical analysis. A telescope collects plume light from distances up to 800 feet from the engine and directs the light through a grating spectrometer system and onto a linear array of silicon photodetectors. The optical multichannel analyzer, or OMA, serves as an electronic "film" that can record, display and read out hundreds of complete near-ultraviolet to near-infrared spectra each second. In practice, one to 10 spectra each second are utilized to document the plume spectral characteristics. Now, in a quantitative manner, light from many wavelengths of interest can be compared to identify elements, ratioed to recognize alloys, monitored as a function of time to establish trends, variations, or the onset of significant material erosion, and used to remotely monitor the power level of the engine in real time.
Plume spectrometric equipment set up in preparation for a test firing at SSFL.
Since the first plume emission spectrometry (PES) of the SSME in 1986, plume spectral acquisitions have recorded many nominal tests (engine hot firings without incident), with the baseline spectral features of the plume now well established.
It has since been discovered that the omnipresent sodium and potassium spectral lines result from contamination of the gaseous hydrogen fuel during the drying process prior to liquefaction. Interestingly, the hydrogen-rich start and shutdown sequences are characterized exclusively by the sodium and potassium spectral lines; the OH-band emissions in the near-ultraviolet wavelengths appear only when both the liquid oxygen and liquid hydrogen propellants are reacting together, providing a verification of main propellant combustion. Metals, such as iron, chromium and nickel, are always present in the plume in small amounts, indicating some level of ongoing wear. The detection of large amounts of these metals has been indicative of significant hardware erosion events.
Metallic emissions represent the combustion of wear particles from the alloys used in the SSME. Alloys are composed of a number of elements, and the determination of the exact alloy seen in the engine plume requires a comparison of the relative amounts (brightnesses) of each component elemental species to a known emission standard.
An extensive series of alloy seeding experiments were performed at the Stennis Space Center Diagnostic Test Facility on an oxygen/hydrogen rocket engine dedicated to quantifying the spectral appearances of plume contaminants. These experiments have shown that the SSME alloys can be readily identified in the plume by PES in concentrations as low as 2 to 5 parts per million. (Elements which are more radiative, such as sodium and potassium, can be detected at parts per billion concentrations.) Once the alloy is determined, additional engine data must by utilized to discover which engine component composed of that alloy is wearing.
Significant metallic emissions have been observed in the SSME plume on many occasions. During SSME test number 750-285, bright spectral lines of chromium were recorded during both the startup and shutdown sequences and were found to correspond to preburner faceplate erosion due to misaligned injector posts in the high-pressure oxidizer turbopump (HPOTP). The test sequence from numbers 750-313 to 750-315 was characterized by very bright chromium, iron and nickel spectral lines that quickly diminished to normal levels. This has been attributed to the first hot firings of a new powerhead; the emissions were from metallic dust being flushed out of the powerhead by the hydrogen-rich steam coming from the high-pressure turbopumps. A similar situation occurred on test number 901-651, the first firing after the rebuilding of Engine 0215. A case of copper contamination was clearly demonstrated in test number 750-296, when pieces of copper tape, intentionally left in the main combustion chamber, were burned and flushed from the engine within seconds of engine start. This "seeding experiment" gave considerable insight to the type of spectra that might be expected during erosion of the copper baffle plates of the main injector - a common cause of green streaking within the plume.
There are many conditions of accelerated wear over brief time intervals - most notably during start and shutdown transients, and during thrust level changes - in which wear rates can be an order of magnitude greater than normal. For example, during sudden changes between thrust levels on the SSME, the release of significant amounts of metallic species from the engine have been seen, while slower rampings between thrust levels are marked by the release of quantities that are comparable to average loss rates of metallic species.
An orbital transfer vehicle/integrated component evaluation (OTV/ICE). The plume spectra quantitatively show the changes in spectral signatures during the various events of the high-pressure fuel turbopump incident. Five scans per second were recorded.
(Click image for larger version)
The importance of calcium hydroxide (CaOH) in the engine plume was discovered during hot-fire testing of the Orbital Transfer Vehicle/Integrated Component Evaluator (OTV/ICE) at the Pratt & Whitney Rocketdyne Advanced Propulsion Test Facility at the Santa Susana Field Laboratory (SSFL). This 15,000-pound-thrust, liquid oxygen/liquid hydrogen prototype engine utilizes a high-pressure fuel turbopump that is designed to operate at speeds up to 105,000 rpm. In one test, the fuel pump was destroyed when it seized while rotating at 87,000 rpm, due to an internal seal leak and limited balance piston capability, resulting in a bearing failure. At approximately 2.6 seconds before conventional instrumentation suggested an anomaly, four spectral emission bands of CaOH suddenly brightened and started to fluctuate at high intensities. Spectral evidence for iron also began at this time. After engine cutoff, only the sodium and potassium features remained until the main propellant purge came on, resulti ng in brilliant CaOH emissions.
Post-test analysis revealed that a bearing had undergone severe distress during the anomaly, resulting in a fragmented bearing cage and damaged balls. Turbopump failure was indicated by anomalous wear on the bearing cage (which is approximately 6 percent calcium) and the subsequent release of this material into the main propellant stream, where it reacted with the oxygen and hydrogen to form CaOH. The concurrent presence of iron in the spectra was attributed to ball and race distress occurring at the same time. The final purge flushed out much of the remaining bearing cage material that was eroded away as the pump slowed down.
This preliminary understanding linking CaOH to bearing distress was soon applied to a series of SSME hot firings at the Santa Susana Field Laboratory A-3 test stand. PES data from test number 750-283 was characterized by highly erratic CaOH levels throughout the 300-second test. During test number 750-284, the emissions steadied and decreased to near-normal levels after the first 30 seconds. Based upon the OTV/ICE experience, these observations suggested that a profound cage wear event with corresponding mass loss had occurred, but it was still not known which of the many bearing sets within the engine was degrading.
Conventional strain gage and accelerometer data revealed little cause for alarm in general, and could not isolate the distressed bearing. Post-test analysis of the strain gage data with the Pratt & Whitney Rocketdyne-developed Bearing Signature Analyzer (BSA), which tracks dynamic shaft and bearing frequencies and records their amplitudes, revealed correspondingly large increases in the second cage harmonic amplitude concurrent with the spectrometer data, and localized the event to the (HPOTP).
In the second test, the BSA noted a drop in the synchronous amplitude and a sharp increase in the cage harmonic amplitudes, which suggested that ball and cage wear had occurred and that serious hardware distress and failure was imminent. The disassembly of this turbopump after the second test revealed that the number 2 bearing had failed, fractured and lost its cage, and that the number 1 bearing had taken over the load. If information from either the PES or BSA had been incorporated as redline parameters, it is likely that the first test would have been terminated early with a precautionary shutdown.
In short, there are many immediate understandings that can be gained by the very presence of anomalous species in the plume; deeper understandings of the detailed engine events contributing to the SSME's spectral signatures await correspondingly deeper correlations, linking the timescales of spectral events, intensity profiles, past engine operational histories, known hardware status, data from conventional and other advanced sensors, and engine performance data.
Optical multichannel analyzer systems are in use on the SSME test stands at the Stennis Space Center and at NASA's Marshall Space Flight Center (MSFC) Technology Test Bed test stand. These systems are also in use at the Santa Susana Field Laboratory (SSFL) where they provide key support in monitoring tests of developmental engines and other engines utilizing different propellants, such as LOX/RP-1, LOX/methane, nitrogen tetroxide/ monomethyl hydrazine and propellants with performance additives. The nominal plume spectra from these propellant combinations are markedly different from the SSME's LOX/ hydrogen plume spectrum, but the spectral appearances of plume contaminants occur at their same specific wavelengths and are readily identified against different background spectra. The flexibility of PES systems allows rapid deployment to respond to changing test and test stand configurations and the ability to easily probe different regions of the engine plume by changing the optical field of view and aimpoints. This technology has also been applied to the robotic and laser welding programs to identify spectral regions where other optical sensors may operate without interference from the weld arc, and to vacuum plasma spray process monitoring to verify powder composition and to search for contaminants in the powder.
In 1988, Pratt & Whitney Rocketdyne (now Rocketdyne Propulsion and Power, a part of The Pratt & Whitney Company) performed the Technology Test Bed Plume Spectrometer program for NASA MSFC in which a ruggedized eight-channel radiometer system was designed, fabricated and successfully tested at SSFL's A-3 test stand. Utilizing wideband spectral data from 29 SSME hot firings, eight wavelengths (corresponding to eight elements and compounds) were found to be important for the health and condition monitoring of the SSME. Eight miniature optical detector assemblies, each with a specific narrowband optical filter, served as individual radiometers monitoring the time profiles of the species of interest. A battery-powered data acquisition unit, designed and fabricated at Pratt & Whitney Rocketdyne, stored data from each channel 10 times per second. This self-contained package was mounted on the test stand 30 feet from the SSME and triggered its own acquisition by "watching" for the appearance of the engine plume, and collected data for over 26 minutes upon activation. The data were later downloaded to a desktop PC for analysis of the time profiles of these species in the engine plume.
The significant discoveries made with both wideband and narrowband PES systems have led NASA MSFC to promote the Optical Plume Anomaly Detection (OPAD) program, with a goal of instrumenting each SSME test stand with customized spectrometer systems. One prototype OPAD system is installed at the SSME Technology Test Bed test stand at NASA-MSFC, and an improved model has since been designed.
Combining the best of both worlds, each system consists of a broadband, high-resolution, optical multichannel analyzer spectrometer covering the near-ultraviolet to near-infrared waveband, and a system of narrowband radiometers, which monitor discrete wavelengths of health and condition monitoring species. With all SSME test stands and all engine hot firings to be covered, a vast library of engine spectral data will be obtained to further document the operating characteristics of the engine.
An OTV/ICE engine firing sequence showing(a) the orange hydrogen-rich startup
(b) the mainstage plume
(c) a white oxygen-rich plume after fuel turbopump shutdown
(d) the helium purge, releasing large quantities of calcium and metals into the plume.
With the viability of this sensing technology well established, it is important to extract as much information as possible from the data being collected. The full extent of information buried in the PES data has yet to be revealed; only intriguing hints have actually surfaced to date. Teaming the PES data with data from other conventional and advanced sensor technologies has yielded greater understandings of engine operation, as the synergy between PES and BSA demonstrates. The possibility of extracting a temperature determination from the PES data would yield a plume thermal mapping capability beyond existing techniques, and would have far-reaching results in the area of computational fluid dynamics (CFD) plume code verification. Efforts are now underway to optically probe the harsh combustion environments of the main combustion chamber and the turbopump preburners, this in order to extract new understandings of the combustion process as it occurs and how it affects the en gine hardware adjacent to it.
The large amount of data obtained during spectral monitoring of an engine plume requires significant data processing to identify and track plume species as a function of time. Known correlations between spectral data and engine data are already being incorporated into expert systems capable of assessing species concentrations, identifying alloys and predicting erosion rates for engine hardware, all in real time.
Perhaps the most significant step in realizing the potential for PES is in determining what types of engine wear, erosion or incidents can actually be identified in the PES data. To this end, it is imperative that a rigorous program be initiated to correlate PES data, engine performance data and engine overhaul records. These data bases have never before been linked in a methodical manner, even though there is great potential for identifying numerous engine performance and hardware anomalies as an ongoing characterization of engine health, and certainly well before the anomalies become severe. These data base correlations will help to establish the standard references upon which future health and condition monitoring systems will be built.
The purpose for developing health and condition monitoring systems for rocket engines is to actually fly with them. The incorporation of a spectrometric monitoring capability into man-rated and space-storable rocket engines is the natural pathway in the utilization of the knowledge base gained in hundreds of ground tests. In anticipation of these applications, designs for flightweight optical probes and ruggedized spectrometers that can be vehicle-mounted and operate in the rocket engine hot-firing environment are being developed. When this hardware link is complimented by a knowledge-based data management and adaptive control system that can coordinate real time data acquisition and analysis with engine control actions, plume emission spectrometry will prove itself to be a key ingredient in integrated control and health monitoring systems.