EARLY WORK ON FLUID MECHANICS IN THE INTERNAL COMBUSTION ENGINE John L Lumley Annual Review of Fluid Mechanics Vol. 33 pp. 319-338 Jeff Hanna April 26, 2006
OVERVIEW • Stimulus for understanding effects of turbulence • Engine “knocking” • Turbulence plays a significant role • Ricardo – Early 1900’s • Understand turbulence and its effect on knock • Early fuel had very low octane • Either limit compression ratio or increase turbulence • Investigate overhead valve engines and flat-head engines • National Advisory Committee for Aeronautics – Mid 1900’s • Performed two research projects on a simulated cylinder of an aircraft engine to measure internal turbulence • Obukhov – 1970’s • Significant research on swirling motions in ellipsoids • Found two types of instabilities
STIMULUS • Main reason to investigate turbulence effects: “knock” • Auto-ignition of gasoline-air mixture that occurs above a certain temperature and pressure. If the mixture ignites before it is supposed to, the engine cannot function properly. • This auto-ignition reaction takes time, and must not be completed before the spark induced flame reaches all of the gases • Turbulence increases the flame speed, thereby decreasing the amount of time that the end gases must wait. -Desire to induce turbulence Tumble is a rotational motion about an axis perpendicular to that of the cylinder
RICARDO – EARLY 1900’s • Very low octane content rating – prone to knocking • Keep compression ratio low – sacrifices performance • Use turbulence to increase flame speed • Alter shape of combustion chamber • Through lots of testing, Ricardo became convinced that the higher efficiency of overhead valve engines (compared to flat-head valve) was due to much greater turbulence, shorter flame travel, and was thus less prone to detonate
RICARDO – EARLY 1900’s Overhead Valve Engine Ricardo’s Flat-head Valve Engine De K Dykes et al (1965) Lee (1939)
RICARDO – EARLY 1900’s Ricardo’s Flat-head Valve Engine • Concentrate main volume of chamber over the valves, leaving minimum clearance between piston and cylinder head • Chilled portion of charge trapped in laminum so it could not detonate (squish) • Shortened flame travel by moving sparkplug to the center of the chamber 3 1 2 De K Dykes et al (1965)
RICARDO – EARLY 1900’s • Ricardo was able to obtain the same power output as an overhead valve engine of the same dimensions. • Therefore, turbulence levels can be assumed to be comparable • In general, turbulence levels in an engine cylinder scale with the mean piston speed. • An overhead valve engine with tumble reaches a RMS turbulent velocity of scale mean piston speed • An engine without squish reaches RMS turbulent velocity of ½ mean piston speed • Estimating the RMS turbulent velocity for squish reveals ½ mean piston speed • Combine this with the ½ residual to obtain a RMS turbulent velocity of scale mean piston speed – just like an overhead valve engine.
NACA –1938 In 1938, the National Advisory Committee for Aeronautics performed two experiments on a simulated aircraft engine cylinder. Using a glass cylinder and high speed camera, they were able to calculate speeds of chopped goose down in an overhead valve cylinder with 4 valves. Determined RMS turbulent velocity to be approximately 1.6 times the mean piston speed, with small amounts of tumble. Lee (1939)
NACA –1938 Using shrouds on the valves placed in various positions as shown here, NACA determined that the RMS turbulence velocities increased to about 2.6 times the mean piston speed, with much more tumble. Lee (1939)
NACA –1938 In their second experiment, they removed the glass cylinder and instead put a glass window in place of the exhaust valves. Observed that the highest turbulence level during early combustion was from shrouding arrangements D, G and F, which were expected to produce the highest levels of turbulence. The conclusion from these experiments was that the higher levels of turbulence were directly proportional to the gas velocities flowing through the valves. Lee (1939)
OBUKHOV – 1970’s Analyses of dynamical behavior of tumbling motion in ellipsoids that can be utilized for flow in an engine cylinder. Obukhov et al considered an incompressible, inviscid fluid system and found that the simplest non-trivial system is a triplet which can be written in canonical form shown here, and which are the same as Euler’s equations for force-free motion of a rigid body.
OBUKHOV – 1970’s • From the rigid body equations, we know there is a second integral of motion which corresponds to angular momentum. • In a fluid case, this corresponds to the sum of the squares of circulations about the principal sections • Spin of a rigid body about the middle axis is unstable, while spin around the other two axes is stable (various textbooks on mechanics). • Related to fluid mechanics, rotation about the middle axis of an ellipsoid is unstable and will overturn.
OBUKHOV – 1970’s • Obukhov et. al experimented with transparent spinning ellipsoids (filled with water) to look at the instabilities associated with the flow. • The ellipsoid was rotated for a long period of time to ensure solid body rotation, and then quickly stopped. • The flow satisfied the force-free motion of a rigid body equations until the boundary layers became too large, which took approximately 5 fluid revolutions. • If initial rotation was about the short axis, the motion was stable and continued. • If rotation was about the intermediate axis, it flipped over and rotated about the shorter axis within about one fluid revolution
OBUKHOV – 1970’s Overturning process for rotation of a fluid about the intermediate axis of an ellipsoid (Obukhov 2000) Obukhov (2000)
OBUKHOV – 1970’s Stability of rotation about the long axis can be demonstrated if the long axis is less than 2x the short axis. As the long axis length reaches 2x the short axis, the flow flips and forms two vortices, parallel to the short axis. As the long axis is increased, the motion becomes stable again, and then unstable etc. Obukhov (2000)
APPLICATION TO AUTOMOBILE ENGINE Ricardo proved that increasing the turbulence in the combustion chamber increased flame speed, making engines more reliable. NACA demonstrated that valve arrangements make it possible to introduce tumble in a cylinder We expect conservation of angular momentum to amplify the tumble during the compression stroke, as the vortices get smaller. Obukhov showed how rotational flow about the intermediate axis was unstable and would turnover. 2 Problems – Ellipsoid is not a cylinder – Cylinder is symmetric
APPLICATION TO AUTOMOBILE ENGINE • Problems with Obukhov: • The rotation in a cylinder will be of higher order. • Truncating the system at least allows for a qualitative idea of what is happening, even though it is not exact. • Because the the cylinder is symmetric, two of the axes will be the same length. If rotation is about the long axis, I1=I2, corresponding to r=0 from the rigid body equations. This results in a table situation, and no overturning would be present.
APPLICATION TO AUTOMOBILE ENGINE • Multi-vortex instability • Stabilities change as the piston moves up the cylinder. • Initially the long axes is the axis of the cylinder, but once the piston moves half way up the cylinder, it becomes the smallest axis, with two equal longer axes perpendicular to it. • The tumble will break up into a number of smaller cortices with axes at right angles to the axis of initial tumble. • As the piston moves more and more, the number of vortices becomes greater and the individual vortices smaller in diameter. • Gledzer & Ponomarev (1992) indicated that when the piston is half-way up the cylinder, the tumble becomes unstable to half-size vortices at right angles to the original axis, further backing this.