“Swirl is not mixing”, at least in the general role it plays in diesel combustion, was one of my first, hard learned conclusions in diesel combustion. While it is a contributor to the mixing process, the view that “spinning of the air” in the cylinder promotes general mixing of the fuel and air to promote improved combustion is wrong. As a young engineer, my early inerrant view was that increased mixing was a general path to improved diesel combustion. I had viewed swirl as mixing and therefore more must be better. Just how wrong this was was to be taught to me in an experiment that dramatically increased the swirl.
In the 1970’s I worked with a special 8V-71N engine (uniflow 2-stroke) that had been constructed using a 8V-92 block using thick walled wet liners. The normal 71 liner thickness was about 4.6 mm thick but this special engine had liners thickness of about 15.6 mm in the port area. The liners were also machined with 35 degree vs the normal 30 degree ports which would further increase the swirl. The result was liners that induced significantly higher levels of swirl than the production engine primarily due to the longer length to width ratio of the ports imparting better directional control of the air entering the cylinder. While I do not have the exact measured or calculated swirl, I” would estimate it was at least doubled if not tripled increasing the swirl ratio from around 2 to perhaps the 6 to 8 range!
At 2100 rated state point, where bsfc in the range of 245 g/kw-hr and smoke of about 1 FSN would be considered normal, the bsfc and associated smoke increased to above 480 g/kw-hr and the smoke to above 7 FSN in the high swirl configuration using the standard 8 hole nozzle and the production piston bowl! Part load and lower speed operation provided similarly poor results. Obviously, improved combustion by assumed increased mixing not only hadn’t been achieved but resulted in dramatic changes in the opposite direction.
First attempts to correct bad combustion were made by testing varied nozzle hole number nozzles with the same total flow area and thus the same injection pressure. 8, 7, 6, 5 and 4 hole nozzles were tested and showed progressive improvements in combustion with fewer holes However, results with the 4 hole nozzle were still dismal at rated speed and fueling with fuel consumption the 335 g/kw-hr range and smoke above 5 FSN.
Subsequently, 4 hole nozzles with larger holes (67% more flow area and subsequently lower injection pressure) were tested but showed poorer relative results. Smaller diameter piston bowls were also tested but provided almost identical results. Higher levels of injection pressure achieved with larger diameter injector plungers directionally improved combustion and reduced fuel consumption to the 290 g/kw-hr level and smoke to below 4 FSN.
The next step in experiment was to remove the liners and incrementally cut down the liner port belts and retest with each of the nozzles of varied hole number (8 to 4). This was done in 4 steps with the final step being a port thickness, and swirl level, similar to the production liner. No noted change occurred until the ports were reduced to length to width ratio of less than 1, below which it would be expected that swirl levels would decrease due to poorer guidance of the air through the ports.
The resulting data once plotted showed a pattern that each nozzle would have an optimum level of swirl at a state point with poorer combustion occurring at both higher and lower swirl levels. These similar patterns were observed at 3 engine speeds and both a high and intermediate fueling levels and at both the normal and higher injection pressure levels. Because of the limited and discrete levels of swirl tested only a limited number of state points tested were optimum, yet by plotting all the data, the trends and the optimums could be estimated.
I was never able to conclude from this experimental data if an incrementally better optimum could be achieved with a higher or lower level of swirl, partly because each nozzle was not tested with it’s exact optimum swirl level but also because swirl in 2-stroke can affect scavenging. Piston bowl diameter and other variable would likely also need to be adjusted. At best I could conclude that reasonably good combustion levels are attainable over a wide range of swirl. The trends that were exhibited when extrapolated would indicate that the highest levels of swirl tested needed perhaps a 2 hole nozzle to achieve an optimum. Considering that several opposed piston engines have successfully utilized either one single hole nozzle with swirl ratios in the range of 12 to 17 range or 2 diametrically opposed single hole nozzles with a lesser level of swirl, this seems logical.
General trends also indicated that part load conditions benefit from either a higher level of swirl or a higher hole number, most likely because of the shorter injection duration. Increased injection pressure, which shortens duration, also consistently showed better results if matched with higher swirl. Use of variable swirl (swirl flaps) to increase swirl at part load while maintaining lower high load swirl is a known example demonstrating this in engines today.
Conclusion: Swirl is a situationally important combustion parameter for optimized combustion. There is no general direction that more swirl provides more mixing or general benefits akin to mixing a can of paint. An optimum level of swirl assists in more uniform distribution of the fuel in air. “Overswirl” results in oxygen starved regions resulting from re-entrainment of oxygen depleted gases and thus poor combustion. It may also result in decreased penetration of the charge and failure to utilize air deep in the bowl. “Underswirl” results in failure to utilize all the air in the chamber resulting in richer combustion. It may also contribute to rich areas due to over penetration of the spray and formation of rich areas at the bowl periphery.