Contactless ultrasound is a novel, easily implemented, technique for the Ultrasonic Treatment (UST) of liquid metals. Instead of using a vibrating sonotrode probe inside the melt, which leads to contamination, we consider a high AC frequency electromagnetic coil placed close to the metal free surface. The coil induces a rapidly changing Lorentz force, which in turn excites sound waves. To reach the necessary pressure amplitude for cavitation with the minimum electrical energy use, it was found necessary to achieve acoustic resonance in the liquid volume, by finely tuning the coil AC supply frequency. The appearance of cavitation was then detected experimentally with an externally placed ultrasonic microphone and confirmed by the reduction in grain size of the solidified metal. To predict the appearance of various resonant modes numerically, the exact dimensions of the melt volume, the holding crucible, surrounding structures and their sound properties are required. As cavitation progresses the speed of sound in the melt changes, which in practice means resonance becomes intermittent. Given the complexity of the situation, two competing numerical models are used to compute the soundfield. A high order time-domain method focusing on a particular forcing frequency and a Helmholtz frequency domain method scanning the full frequency range of the power supply. A good agreement is achieved between the two methods and experiments which means the optimal setup for the process can be predicted with some accuracy.

中文翻译：

---

用于合金熔体中非接触式超声空化的声共振。

非接触超声是一种用于液态金属超声处理（UST）的新颖，易于实施的技术。与其在熔体内部使用振动超声波探头，而不是导致污染的超声波探头，我们考虑将高交流频率电磁线圈放置在靠近无金属表面的位置。线圈感应出快速变化的洛伦兹力，进而激发声波。为了以最小的电能使用达到空化所需的压力幅度，发现有必要通过微调线圈的交流电源频率来实现液体体积中的声共振。然后使用外部放置的超声波麦克风通过实验检测到气蚀现象，并通过减小固化金属的晶粒尺寸来确认。为了用数字预测各种共振模式的出现，需要熔体体积，保持坩埚，周围结构及其声音特性的确切尺寸。随着空化的进行，熔体中的声速发生变化，这实际上意味着共振会断断续续。考虑到情况的复杂性，可以使用两个相互竞争的数值模型来计算声场。专注于特定强制频率的高阶时域方法和扫描电源整个频率范围的亥姆霍兹频域方法。两种方法和实验之间达成了良好的协议，这意味着可以以一定的精度预测该过程的最佳设置。实际上，这意味着共振会断断续续。考虑到情况的复杂性，可以使用两个相互竞争的数值模型来计算声场。专注于特定强制频率的高阶时域方法和扫描电源整个频率范围的亥姆霍兹频域方法。两种方法和实验之间达成了良好的协议，这意味着可以以一定的精度预测该过程的最佳设置。实际上，这意味着共振会断断续续。考虑到情况的复杂性，可以使用两个相互竞争的数值模型来计算声场。专注于特定强制频率的高阶时域方法和扫描电源整个频率范围的亥姆霍兹频域方法。两种方法和实验之间达成了良好的协议，这意味着可以以一定的精度预测该过程的最佳设置。