Enriching an environment with oxygen increases the fire hazards if an appropriate ignition mechanism and fuel are present. Serious incidents involving burning metallic materials occur in systems containing oxygen because of a metal's enhanced flammability in such environments. Understanding the relative flammability of metals is crucial for the safe design of oxygen systems. The flammability of these metals is characterised by experimental variables determined from a standard promoted ignition test of a vertically-held metallic rod. One such parameter, which is also essential for modelling, is the Regression Rate of the Melting Interface (RRMI) that propagates upward along the rod as it burns. The conventional method of obtaining the RRMI is based on a time-consuming, expensive and often inaccurate visual analysis of a film recording.
A new and unique method is presented that uses a pulse-echo, ultrasonic system to measure the
RRMI. The initial design used an intricate dual-element transducer and a complicated series of analogue devices for signal processing and acquisition. RRMI values were only obtained for very slow burning tests because of limited acquisition speed, basic processing techniques, and poor signal to noise ratio. The superseding and current system operates a single-element transducer to transmit and receive an ultrasonic structural wave pulse. The transmitted pulse propagates down the rod, reflects at the melting interface and returns for detection. The pulse-signal is recorded by a high-speed digital data acquisition system. A digital signal processing method involving a Hilbert transform accurately determines the pulse's time-of-flight within the rod before being converted into an instantaneous rod length (RL), using the rod's pre-determined sound velocity (SV). The changing RL as a function of test time corresponds to the RRMI and is calculated immediately upon test completion.
The new technique can be applied for use where the initial prototype system did not function and the results obtained compare well with RRMIs determined by the visual analysis technique for relatively slow burning test samples. For pure aluminium rods (3.2-mm diameter) burning at the highest available test pressure (69 MPa), fast RRMIs (up to 220 mm/s) are successfully, accurately and repeatedly measured by the digital ultrasonic method whereas the visual technique and analogue ultrasonic technique fail. Since aluminium burns rapidly compared to most metals and pressure generally increases the RRMI, the limiting factor is currently the physical pressure limits of the test apparatus and independent of the developed ultrasonic measurement system.
For aluminium, the RRMI relationships with test pressure, sample diameter and oxygen concentration are consistent with trends reported in literature. Though published work suggests aluminium burns in a
vapour-phase homogeneous reaction, evidence and modelling presented here support a predominant heterogeneous reaction mechanism. The molten metal oxide ball that surrounds the melted aluminium controls the amount of reactant transfer and produces the unique RRMI relationships observed for aluminium. Because oxygen systems exist in space-based equipment/crafts and the gravity level is known to effect the burning process, additional RRMI values are measured for several metals burning in a reduced gravity environment. Under normal terrestrial gravity, the molten ball cyclically grows and detaches but in reduced gravity the growing molten ball is retained causing faster regression rates due to enhanced heat transfer.
RL versus test time data obtained from the ultrasonic method and used to calculate the RRMI, are discussed in relation to the ignition, burning and extinguishment stages of a burn test in both normal and reduced gravity. For each instantaneous RL the
echo amplitude (EA) is determined from the signal processing and is directly related to the temperature-induced attenuation, providing a qualitative indication of the temperature at the burning end of the rod.
The operation of the ultrasonic system is based on three hypotheses; the echo represents a reflection from the solid/liquid interface (SLI), the effect of a non-planar interface is negligible, and the temperature-related SV change during a test is also negligible. The dissertation investigates these hypotheses using RL, EA and synchronised video results. There is disagreement within literature concerning whether the reflection occurs at the SLI or the molten ball/gas interface and is likely related to the excitation voltage of the ultrasound. However, using theoretical modelling and experimental verification, the reflection is proven to originate from the SLI. Results also validate the second hypothesis by showing the effect of interface curvature on
the echo's time-of-flight measurement is negligible because of the pulse's large wavelength. SV is a function of temperature and a theoretical heat transfer analysis quantitatively confirms that the amount of heat conducted along a burning rod is also negligible. Though the resulting total SV offset is small, which validates the third hypothesis, it is still accounted for in the RRMI calculation.
The work described above has led to a better understanding of the application of structural waves to a fast moving SLI and some additional conclusions reached include:
• The standard ignition configuration of a rod sample (machined grooves and wrapped igniter wire) attenuates the ultrasonic signal more than the temperature-induced attenuation caused once ignition occurs.
• In normal gravity, the exact position of molten ball detachment is determined by synchronising the acquisition of the ultrasonic
measurement system with the video recording. Differences between iron and aluminium results are in agreement with their proposed heterogeneous modes of burning.
• The EA is proportional to the magnitude of the heat-affected length since signal attenuation increases with both temperature and the distance that contains the temperature gradient. A slower RRMI permits more time for heat conduction, which produces a longer heat-affected length and, therefore, a smaller EA. The retention of the molten ball in reduced gravity causes greater signal attenuation than in normal gravity tests because of the longer heat-affected length formed and higher temperatures reached.
• The ignition and extinguishment stages of a burn test do not affect either the calculation of or accuracy associated with the RRMI by the ultrasonic system.
• The EA of the SLI echo recorded during burn tests is much greater
than that predicted using 1-D reflection theory. This anomaly is consistent with the results of several system simulation tests and past literature. Thus, the theory is either inappropriate for the application of structural waves to a SLI and/or the nature of the molten liquid is not characterised correctly.
• Structural waves can propagate in a liquid metal. However, the damping characteristics of the molten ball attenuate the transmitted signal completely.