Over the years, various people have claimed that cavitation produces excess heat. This is also known as sonofusion. Examples include the work of Roger Stringham and the hydrodynamics gadget (https://www.hydrodynamics.com/). (Look up Stringham in the LENR-CANR.org index, https://lenr-canr.org/wordpress/?page_id=1081)
We assume this is caused by cold fusion. I do not know of any theoretical reason for that assumption. It does not seem to have much to do with hydrogen or deuterium loaded into metals. Except that in many cases the collapsing bubbles impinge on metal and inject hydrogen and oxygen into the metal. Based on McKubre's law of the conservation of miracles, I suppose it is cold fusion. Frankly, I don't care whether it is or is not, as long as it works.
In any case, I would like to draw your attention to an ICCF23 presentation about this approach. The title does not indicate that's what it is about:
Excess Energy from Heat-Exchange Systems
In ICCF22, we presented a vapor compression machine (VCS-1) using a 2.75RT freon compressor (Figure 1) which can produce excess energy . The hot refrigerant vapor from the compressor (around 150°C) is used to heat the water flowing through a tiny passage of a triple-pipe heat exchanger. This may cause a violent cavitation of water. The machine was modified furthermore and tested for two years since then. The calorimetric method for COP measurement was improved. The COP inside the steam generator is defined as the heat carried away by water (Qwnet) divided by the net heat input (Wt - QL), denoted as COPx . This is used as the criterion to determine the possibility of excess energy generation. If the measured COPx was greater than 1, then the cavitation-induced low-energy nuclear reaction (LENR) might occur. The test shows that the maximum COPx reaches 1.97 (Figure 1) and COPx increases with decreasing inlet water temperature. . . .
In the video lecture, the COP is shown as high as 1.97 (minute 10). Excess heat, when present, ranges from 2.15 to 4.18 kW (minute 16:30). With one reactor, there were 4 months with no heat, which can be taken as a baseline calibration, followed by 2 months of excess heat (minute 16).
An upcoming JCMNS paper has more details.