TECHNICAL FIELD

A torque converter is typically used tc connect =. ccr- stant rotative speed power source to a load requiring vari¬ able speed drive. The operating characteristic most desired is a large utility ratio. This is defined as the ratio c: the upper and lower output speeds at which the eff ciency cf energy transmission falls off tc a predetermined value, usually 80% of maximum.

A conventional machine has a fundamental limit tc its utility ratio imposed by the so called "hydraulic shock" loss. The machines of the invention utilize structure v.πi r. eliminates this source cf loss and thereby allows for ar advance m the magnitude of the attainable utility ratio.

A machine according to the invention using air and water as the working medium can be operated as an automatic transmission by changing the relative proportions of air and water.

SUBSTITUTESHEET

OBJECTS OF THE INVENTION

The principal object of the invention is to provide a torque converter of extended utility ratio.

Another obiect of the invention is tc provide novel means of control of the improved machine.

Yet another object of the invention is to provide an improve torque converter which is effective when only par¬ tially filled with hydraulic fluid.

A further object of the invention is tc provide ε torque converter which can be operated as an automatic transmission.

Other objects of the invention will in part be obvious and will in part appear hereinafter.

For a fuller understanding of the nature and objects of the invention, reference should be had to the following description and drawings.

BRIEF DESCRIPTION OF THE DRAWINGS

Figure 1 is a partial side cross sectional viev. ci :. torque converter according to my invention.

Figures 2 and 8 are front views of impeller clades which would be used if the machine were a conventional one.

Figures 3 and 7 are side views of turbine blades which would be used if the machine were a conventional one.

Figures 4, 5 and 6 are side views of impeller blades according to the invention.

Figures 9, 10 and 11 are front views of turbine blades according to the invention.

Figures 12 through 17 are sections through a passaαewa- according to the invention taken along section lines indi¬ cated in Figure 3.

Figures 18 through 23 are sections according to the in¬ vention taken along the section lines shown in Figure 7.

Figure 24 is a side view of a stator blade according to the invention.

Figures 25 through 30 are sections taken along the section lines in Figure 24.

Figures 31, 32, 33 and 34 are the top, side, rear and perspective views of another embodiment of a stator blade according to the invention.

Figures 35 and 36 are side views of movable elements tc be used with certain passageways in a machine according tc the invention.

Figure 37 shows yet another modification to the sectior. cf Figure 14.

Figure 38 shows an array cf cc-rotatmg vortices.

Figures 39 through 42 are schematic representations c z arrangements of vortices .

Figure 43 is a perspective view cf a novel arrancer.er.' cf vorticity generators.

Figures 44, 4Ξ and 46 are the top, side and front viev cf a novel leading edge vorticity generator.

Figure 4 ^~ s Ξ. sice view cf ar.ctner odif cat cr. t_ tr.-,

SUBSTITUTE SHEET

- ----

The same reference characters refer to the same ele¬ ments throughout Figures 1 through 47 of the drawings.

Figures 2.1.1 through Figure 5.2.1 are illustrations forming part of a technical description of the invention.

The same reference characters refer to the same ele¬ ments throughout the several views of Figures 2.1.1 through Figure 5.2.1.

Figures 48 through 57 are photographs of flow visual¬ ization studies illustrating the principles of the inven¬ tion.

Figure 58 is a partial side cross-sectional view simi¬ lar to Figure 1 of a torque converter according to my inven¬ tion especially designed to utilize two working fluids c: differing densities.

Figure 59 is a partial cross-sectional view taken along the line 70-71.

Figure 60 is a partial cross-sectional view taken along the intersecting cylinder 60-60 shown in Figure 59.

Figure 61 is a partial cross-sectional view taken along the cone generated by rotating line 72-73 of Figure 58 around the axis of rotation cf the torque converter cf Fig¬ ure 58.

Figure 62 is a view along the cone generated by line 86 when rotated around the axis of the machine shown in Figure 58; line 86 being perpendicular to line 72-73. This view is of the stator 90 with the impeller 88 removed.

Figure 63 is a view similar to Figure 62 of the impel¬ ler with the stator removed.

Figure 64 is a cross-sectional view taken alone the cylinder generated by rotating line 74-75 of Fiσure 5_ around the axis of the machine.

Figure 65 is a partial top view cf the cross sect c:: c: Figure 64.

Figure 66 is a partial cross-sectional view taker. a_cr.r the plane 78-79.

Figure 67 is a diagrammatic partial top view if tne turbine cf Fiσure 6 .

Figure 68 is a diagrammatic partial bottom v ew cf the impeller of the Figure 58.

Figure 69 is a partial cross-sectional view taken alone the cylinder generated by rotating line 82-83 of Figure 5£ about the axis of the machine.

Figure 70 is a diagrammatic view perpendicular tc Fig¬ ure 69.

Figure 71 is a cross-sectional view taken along tne cone generated by rotating line 84-85 of Figure 58 about the axis of the machine.

Figure 72 is a view taken on a cone generated DY a perpendicular line bisecting line 84-85 rotated aoo tne axis of the machine, and is similar to Figure 62 and snows the turbine only.

Figure 73 is a view, similar to Figures 63 and 72, cf the stator only.

Figure 74 is a partial cross-sectional viev: illus¬ trating a modification of the invention.

Figure 75 is a partial cross-sectional view illus¬ trating another modification of the invention.

Figure 76 is a partial cross-sectional viev: illus¬ trating a further modification of the invention.

Figure 77 is a partial cross-sectional view illus¬ trating a still further modification of the invention.

Figure 78 is a partial cross-sectional diagrammatic view of a still another modification of the invention.

Figure 79 is a diagrammatic view of the torque con¬ verter according to my invention operated as an automatic transmission.

The same reference cnaracters refer to the same ele¬ ments throughout the several views of Figures 58 tnrougr ^~ i .

Ail of the views herein, and the use of front and s:α-; tc describe certain views, conform tc the usage cf SAΞ pub¬ lication number AE-Ξ, Chapter 21, Design cf Single Stac- Three-element Torque Converter bv V. J. Jandasek.

SUBSTITUTE SHEET

BEST MODE FOR CARRYING OUT THE INVENTION

A torque converter according to the invention is shown m Figure 1. The input shaft is denoted by 1 and the output shaft by 2. The input shaft drives a centrifugal pump cr impeller consisting of a housing 3, a shell member 9, and an inner core member 8. The hydrodynamic blading array is mounted between member 8 and 9, forming the passageway of the impeller. The blades 10 and 10' are shown in Figure 1 and various forms of blades are shown in Figures 2 through 6. The output runner or turbine consists cf a hub 4 supporting a shell 12 and an inner core 11. Turbine blades 13 and 13' are shown in Figure I and various forms for them, are shown in Figures 7 through 11. The stator is supported by its hub 5 and consists of a core 14 and shell 15 with blades 16 mounted therebetween. Forms of blades are shown in Figures 24 through 30 and Figures 31 through 34. The stator is connected to a fixed frame 7 through a one-way clutch shown schematically at 6.

The operation of a conventional torque converter can be understood by referring to Figure I. The flow of the hy¬ draulic fluid contained within the machine is directed out¬ wardly away from the shaft and energized by the blading IC within the rotating impeller. The fluid then enters the radial turbine runner where the energy of the fluid is ex¬ tracted by the blades 13. Next, the fluid enters the stator stage of the turbine where the torque reaction takes place. The general direction of flow is shown by the curved arrow.

This flow is controlled by the conventional blades If and 13 shown in Figures 1, 3, 7 and £. These blades are designed to minimize the effects of the so-called secondary flows. These flows are ones which contain ε component z : vcrticity in the mean stream direction within the passages formed by the core, shell and biaces. In a conventional machine the energy of the secondary flows or streamwisf vertices is lost and serves only to heat the working fluid ε: that the efficiency of the machine s reduced.

In a torque converter there is only one speed ratio at which the flow enters the blades at their design point. At all other speed ratios there is a non-optimum relative angle of attack between the flow direction and the direction 01 tne blade camber line at its nose. This results in flov losses that are known in the literature as "hydraulic snock" losses or as "nose stall" losses. The approaching flov: αoeε not stay attached to the suction side of the blade. Ratner , it forms a separation bubble with part of the flow recircu¬ lating in a direction which is reversed with respect to tne mean flow. Under most conditions, the separation region or bubble is unstable and the energy of the flow is disεipatec by the resulting turbulence. Since the energy associatec with the turbulence resulting from this stalled condition, s lost to the machine, its efficiency as an energy transmis¬ sion device is reduced. In most practical calculations in¬ volving torque converters , the energy in the flow normal to the thin blade nose camber line is assumed to be entirely lost.

In a conventional prior art machine the thin blades shown in Figures 2, 3, 7 and 8 can be replaced with airfoil shaped blades with well-rounded noses. This reduces the nose stall loss. However, to reduce the nose stall com¬ pletely requires the use of such thick blades that the flov. passages become restrictively small and friction between the flowing fluid and these walls becomes excessively high. Thus, the design of the conventional machine is forced tc compromise the peak efficiency obtained at the design point in order to obtain efficient operation over a wide range cf speed ratios; i.e. tne utility ratio is severely limited.

The present invention greatly reduces tne severity cf tnis compromise sc that a more useful range cf niσr. eft_- ciency can oe achieved. This is accomplished _r celioerate- i'_ introducing secondary flows or εtreamwise vertices t' tne flov: within tne passageways of tne machine. In a con¬ ventional machine the energy cf tnese flows would ne _.o£t, however, m tne present invention a novel struct r-. is

SUBSTITUTESHEET

introduced in the following stages of the machine so that the energy of the vortices can be recovered. Furthermore., the structures which produce and recover the energy of these vortices are not subject to the compromises resulting from the nose stall phenomena of the conventional machine and an increased utility ratio is achieved.

A simple way to obtain the results of the invention is illustrated in Figures 1, 4, 5, 9 and 10. Every other im¬ peller blade 10 and 10' is cut away alternating between the forms of Figures 4 and 5 to form the salient edges 17 and 18. These edges are disposed along the mean flow path. When fluid enters the impeller at a relative angle of attach to the blades, vortex sheets springing from these edgeε roll up into streamwise vortices which exist within the passage¬ ways of the impeller. Thus, nose separation stall and re¬ versed flow is avoided. The energy in the component of the flow which would otherwise be lost is now largely contained within the secondary vortex flow. The blade 11* ' of Figure 6 is an alternative embodiment of the blade 11' of Figure 5.

In a similar fashion, turbine blade 13, 13' is cut away alternating between the forms of Figures 9 and 10 tc form salient edges 19 and 20 which are disposed along the mean streamwise direction. When the streamwise vortices impact this structure, the energy contained in them is transferred in part to the turbine. Nose stall at the turbine entrance- is also avoided and the object of the invention is achieved. The blade 13 ' ' of Figure 11 is an alternative embodiment of the blade 13' of Figure 10.

The stator of the machine is modified as shown r. Figures 24 through 30. Salient edges 21, 22, 22, and 24 c: blade 16 is disposed along the mean streamline to centre! the streamwise vortices. Nose stall is avoided and the flov iε fed back to the impeller in an efficient fashion.

Further alternative embodiments are shown in Figures 11 through 30. Figure 12 is a section normal to the mean flov taken along the line designated as A-A in Figure 3. Tne alternate salient edσes 1 ^" and 15 are snart eάσeε located :

the same place as described for the thin blades cf Figures - and 5. However, the blades now have a fillet regie:. 12 added en all sides to produce a cusp like shape. This avoids a square corner where the thin blade meets the snel_ or core walls and a more efficient flow is achieved.

In Figure 12 the vortex sheets springing from tne cusoε are denoted by the arrows 26 and 27. These sheets are i: the process of rolling up around the vortex cores denoted c the small circles in the figure. These vortices grov. in strength and move into the region shown in Figure 13 which s a section along B-B of Figure 3. Thus the staggered rov of vortices of Figure 12 is oemg guided mtc a single ro; cf Figures 13 through 17.

Figures 6, 14, and 15 show an additional modif cation. Since the vortices are stable structures, not as much trans¬ verse blade surface is required to guide the flow as in e. conventional machine. Thus, the blade of Figure 5 is fur¬ ther cut away as shown in Figure 6. The salient edge 28 stabilizes and guides the flow. Fillets such as 29 form a cusp like structure for this edge. The walls of the pass¬ ageway such as 30 where the blades are not cut away nave oeen similarly rounded with fillets to avoid square ccrnerε . Indeed, the preferred shape of a passageway conducting a streamwise vortex approaches a circular section. The sali¬ ent edge 28 is almost gone by the time the flow reacneε section D-D shown in Figure 15. It has vanished at station E-E shown in Figure 16. Thus a contra-rotating pa r of linear streamwise vortices are conducted by the oval channel 29 to the flatter channel 30 shown in Figure 17. This fiσ¬ ure is a view of the exit of the impeller. Tne pcrticr. cr the exit region 30 is rounded to allow the vortices t: aα- -uεt tc the entrance conditions cf the turbine with sa__.__.ent edges 19 and 20.

The same type of modifications to the turbine r nne are shown in Figures 18 through 23. Figure 23 snows tr._ entrance region to the turbine taken at the location "-- ^' snown in Figure ~. Fillets such as 21 are addec t: ma.-:-

SUBSTITUTESHEET

cusp like sections leading up to salient edges 19 and 2C. The vortices are being generated as indicated by the arrows 32 and 33. The vortices are guided by the turbine passage¬ ways those sections are denoted H-H through L-L. Again some of the core blades are removed entirely as shown in Figure 11 so that a salient edge 34 separates the vortex pair in Figures 21 and 22. By section J-J in Figure 20, the edge has vanished. The passageway continuing each pair is nov: rotated aε shown in Figures 19 and 20 so that the linear array of Figure 23 becomes a double staggered row of stream- wise vortices in section L-L of Figure 18.

The vortices exiting the turbine runner impinge on the stator shown in Figures 24 through 30. A conventional sta¬ tor would have a straight leading edge parallel to the exit edge of the conventional blade of Figure 7 and a trailing edge parallel to the leading edge of the conventional impel¬ ler blade of Figure 3. A section taken along -M of Figure

24 of a conventional blade would be airfoil shaped. In contrast to this the novel stator shown in Figure 24 has leading edges 21 and 22 and trailing edges 23 and 24 dis¬ posed along the mean streamline direction. In addition to this the airfoil section has modifications shown in Figures

25 through 30. The purpose of these modifications is tc control the streamwise vorticity in an efficient manner.

The leading edge of the stator uses salient edges 21 and 22 shown in Figure 24 that function in a manner similar to that described in connection with the impeller. This is shown in Figure 27 which is a section N-N taken aε indicated in Figure 24. The arrows 35 and 36 indicate the generation cf the streamwise vorticity. In this case a non-staggered double row of vortices is being generated. The pair cf vortices is conducted into the section 0-0 of Figure 26.

As this vortex pair enters the section P-P of Figure _._ tne transfer of energy intc the vortices occurs along the trailing salient edges 23 and 24. These edges are disposed along the mean streamwise direction and cut away aε shown ir A share salient edσs is net necessar" t: acccrr-

cliεh this energy transfer as is shown by the modifieα sec¬ tion P'-P' of Figure 29.

Figure 25 showε the modification to the airfoil secticr M-M of Figure 24. The modification allows the vortex pair to form m a stable oval shaped region while the cross sec¬ tional area of the passageway still follows the same pattern as would a passageway built of conventional airfoil members. An additional modification to this section is shown in Fig¬ ure 30 and is designated M-M. This modification rounds tne trailing edge of the stator blade εo that the flov conditions around the impeller salient edges 17 and 18 s accommodated. Notice that the exit portion 23 and 24 of tne stator blades overlap the entrance region edges 1 ^" anc I , c: the impeller. Such an overlap is not possible in a conven¬ tional machine.

A modification to the stator blaαe is shown m Figures 31 through 34 which are the top, side, rear and perspective views of the novel blade. This blade is pivotally mounted on bearings 69 and 70 and a bell crank mechanism 71 is provided to adjust the pitch angle of the blade through eccentric 71.

This embodiment of the inventive concept yields a vor¬ ticity controlling salient edge without modifying tne plan- form of the blade. The blade 16 is formed with a crescent shaped mound with salient edge 37. This mound resembles a crescent shaped sand dune aε is sometimes found in nature. The edge 37 generates streamwise vorticity as indicated b^ the arrows 38 and 39. The other side of the blade 16 has ? similar edge 40 which generates streamwise vorticity denoted DV arrows 41 and 42. The vorticity is generates preαor: - nately on tne suction side of the blade so that different vortex patterns form depending on the speed ratio c: tn- machme and the pitch angle of the blade.

The exploitation of streamwise vcrticir allows :cr novel variable geometry control elements. Two examples are given in Figures 25 and 36. In Figure 35 a paεsagewa_ cer - taming streamwise vortices is denoteα r*_ 42. Tne eassaσ---

SUBSTITUTESHEET

way contains salient edges 44 and 45. A movable element 46 is provided which pivots about 47. When a suitable means causes 46 to move to its upward position as indicated by arrow 48, the effect of the edgeε 44 and 45 is removed.

In Figure 36 the passageway 43 contains a movable plate 49 with salient edge 50. This edge acts like a thin airfoil of small aspect ratio. The flow iε similar tc that described in connection with Figure 6. The plate 49 is connected by 51 to a mechanism which translates it in the direction of the arrow 52. When the plate is translated to the right, the edge 50 is removed from the flow and the operation of the machine iε modified.

Figure 37 shows an additional modification to the pas¬ sageway of Figure 14. Unlike conventional machines there is no requirement on the passageways to be composed of surfaces swept out by axial generatrix. This requirement is only encountered at the regions where the flow passes from one cascade of blades to the next. Consequently, the circular portions of the passageway are staggered in an axial direc¬ tion. The shell side of the impeller is 53; the core side 54; the suction side 55; and the pressure side 56. This configuration iε beneficial in suppressing other secondary flowε which are not desirable.

Various arrays of vortices are found to be stable m nature. This suggests that such configurations can also be used in the present invention as shown in Figures 38 through 42. Figure 38 shows a single row of vortices. These vor¬ tices are co-rotating as opposed to the contra-rotating single row discussed previously. This row iε generated ey tne edges 57 which are located on one wall only.

Figures 39 and 40 show a double row of non-staggerec ^" vertices. The vortices in each row rotate in the cppoε t-. sense. In Figure 39 the salient edges 58 protrude frcrr. tr._ wall, while in Figure 40 the salient edgeε 59 are carried or a centerbody.

Figure 43 shows an array of vortex generators 61. f.z which exploit the abilit" cf a followinσ array cf element/

64 to roll up the streamwise vorticity generated by a pre- ceeding array 62. This would be especially useful in the overlap region that occurs in the space between edges 21 and 17 in Figure 1. In Figure 43 the initial array s composed cf a series of protrusions 62 which are located on the snell surface 9. These shapeε are curved, cuεped airfcils with salient edges 63 generating an array of εtreamwiεe vortices. They are followed by a similar row of bodies 64 cf iarger size and spacing. The vorticity generated by row 62 is rolled up by row 64. The flow on the wall 9 iε laminarized in part and its friction loss reduced. The energy wound ue in the streamwise vortices iε recovered by the turbine εtructureε 19 and 20.

The location of the streamwise vortex within the pas¬ sageway can be utilized to reduce or prevent unwanted separa¬ tions and turbulence of the mean flow. For example the vortex core should be located nearer the suction side rather than the pressure side of the passageway. Also the vortex core should be nearer the core side than the shell side.

Figures 44 through 46 show a modification to a cusp like streamwise vorticity generator which positions the core of the vortex away from the following wall at . low angles cf attack. The salient edge is denoted by 66 and the arrow 67 shows the vorticity generation.

Previously a description was given of the use cf twe vortices within a single channel. More than two vortices can be used to advantage, especially in designs which must energize the central region of the passageway to avoid vorte: breakdown. Such an arrangement is shown in Figure 47 which is a modi ication cf the blade 10 of Figure 4. The salient edge 17 is now cut in a saw tooth fashion as shown by 6S. This sawtooth generates a triple helical vortex structure within, the passageway.

It is possible to operate the present machine when cr. y partially filled with hydraulic fluid, the remaining space being filled with vapor. The operation cf the machine n this mode is cuite different than the conventional cr.e . 1:

SUBSTITUTESHEET

the novel machine the lowest pressure is located at the vortex cores. Thus, the novel machine can operate when partially filled without the difficulties a conventional machine experienceε when a free surface is pierced by a blade.

Machines Specifically Designed to Operate Utilizing Fluids of .Differinσ— Densities

The machines illustrated in Figures 58 through 78 are specifically designed to operate utilizing a liquid and a gas. The liquid, preferably, is water with or without anti¬ freeze, lubricants, etc. and the gas is air. This machine- utilizes an extensive overlap between successive stages 88, 89, and 90 such that vorticies emanating from one stage having an air core emerge as a sheet and are curled up into vorticies in the successive stage by cusp-shaped slots in the leading edgeε thereof. Preferably, each vortex ^" iε con¬ fined to a single closed channel or passage in each stage. Thus, the stages may be conceived of as comprising multiple tubes; one for each vortex having vortex generating means at the inlet of the tube and vortex into sheet flow forming means at the exit end of the tube. The exits from the tubes or passages are also slot-like and cusp-shaped. The flow cf the vortices as in the machines of Figures 1 through 57, is along the mean flow path of a conventional machine. Thus, the tubeε follow spiral paths.

As seen in Figure 61, the counterclockwise vortex flow within the tubes or channels of one stage 90 exit as a sheet flov: and are re-directed into counterclockwise flowing ver¬ tices in the channels of the next stage 88, then pass- through the tubes in the next stage and emerge from εict- like nozzles at the exits cf the tubes or channels. Thus, the use of a single tube tc guide each vertex through each stage allowε for differing fluid dynamic contrcl at the input and output of each stage and minimum flow losses in the tube, passage, cr channel.

I provide auxiliary blades 116 rotating with the impel¬ ler. These blades intrude into the turbine and aic r εtarting the flow and preventing stalling.

In the alternative embodiment of the invention illus¬ trated in Figure 74, guide vanes 125 are provided to αei guide the flow into a εtage and to help form each vortex. In a modification cf thiε illustrated m Figure "5, cuεps are employed at the leading edges of these vaneε to ne_t form the vortex within the passage of the stage.

In another modification of the invention, the input slots of two or more sets of tubes within each stage ma- pe offset along the leading edge cf the mteractior regie s-- tnat when the machine is not completely filled with flu c, tne radial flow of the fluid will be intercepted rr tne slots extending farthest forward on the leading edge first, and this will capture the entire flow. As the machine has additional fluid added to it, the flow will progreεs farther along the leading edge and then be intercepted by the suc¬ cessive sets of input slots so that as the machine is filled with more liquid, more torque is converted. Thiε modifica¬ tion of the invention provideε for accumulation εuch tnat tne machine may be operated as an automatic transmission aε illustrated in Figure 79. This staggered input slot desiσr can be utilizeα as illustrated in Figure 78, such tnat air can be selectively injected into differing sets of slots ir order to change the balance in the machine to counteract vibration.

Torque converters designed according to my inventicr illustrated in Figures 1 through 57 are designed to operate witn a conventional transmission, fcr exampxe; in a car \. t. three gear sets. however, these machines may a sc be oper¬ ated as an automatic transmisεior. witnout gear εetε : tv _ working fluids are employed, fcr example; a r and ware: -* air and hydraulic fluid. This is pecauεe tne caeacif c: the machine may pe varied by increasing the liquid gaε ratι ^r cf working fluid ir the machine thus cnan mc the mas _; c: tr.-r vertices and tr.erec varvmc tne efrective σear ι_ .

SUBSTITUTESHEET

What iε required for an automatic transmission, iε that the torque converter have a very large utility ratio, that is the maximum over the minimum operating rpm must approach 5 to 1 with an efficiency of greater than 75% and means must be provided to vary the capacity of the torque converter to thereby vary the effective gear ratio.

The use of two fluids having differing densities will achieve this result in torque converters as illustrated in Figures 1 through 57 and iε even more effective in the torque converters illustrated in Figures 58 through 78.

My improved two fluid torque converters exploit the remarkable stability of a Streamwise Vortex when it contains a fluid core of lower density than the vortex itself. I have found, for example, that an air core, water vortex (formed in a U-tube by a crude rotating member which pro¬ duces a very turbulent Streamwise Vortex) produces only about 5 to 10% of the loss expected from flow in a similar passage in a conventional torque converter.

Thus, operating torque converters according to my in¬ vention with Streamwise Vortices having lower density fluid cores, reduces energy losses in the converters. That iε, more energy comes out of the torque converter as useful rotational torque over a wider speed range.

According to the present invention, the gear ratio of the torque converter is changed by changing the ratio of air to liquid or low density fluid to high density fluid in the torque converter. By introducing more air, which carries negligible energy, the capacity of the torque converter may be reduced and the engine driving it will speed up. Thus, the effective gear ratio is reduced, but since the efficien¬ cy remains high, I can effectively change this gear ratio. Varying the ratio of high density fluid to low density fluid causes an accumulation effect.

In the modification of my invention illustrated in Figure 77, an additional accumulation effect is caused by providing sets of inlets to each stage of the torque con¬ verter which inlets of each set are offset along the leading edge of the interaction region, such that, at low flow rates

oniy the first set will carry fluid and at higher ratios cf water to air, successive setε will begin to carry fluid through the device; thus, effectively changing tne gear ratio of the device.

For example, suppose only one εet of stater passages contains liquid. The impeller passages are in relative motion as the impeller rotates with respect to the stater. Thus, the liquid emanting from one set of εtator passages, will be equally divided among all impeller passages and this fluid will accumulate back into streamline vortices m the impeller stage. Thiε process is greatly aided by the en¬ hanced stability of the Streamwise Vortex flov: wner. tw fluids of differing density are employed.

It should be noted that the torque converters cf Fig¬ ures 58 through 78 operate without reversing the direction of the vortices employed, whereas the torque converters cf Figures 1 through 57 do reverse the direction of vortices from stage to stage or within a stage. The torque convert¬ ers of Figures 58 through 78 have extensive overlap between stageε. And the design of the passages within each stage is such that the inner lower density core is kept as active aε possible throughout the circuit.

Passive vibration control is achieved by having tne passages of each staggered set of passages rotationaliy symmetric, as illustrated in Figure 77. That iε, each inner passage would have an inner passage located diametrically opposite to it on the machine or three inner passages would be located at equal angles around the machine. Passages cf an outer set would be located with the same symmetry half way between those of the first set.

Active vibration control may be achieved aε illuεtrater ir. Figure 78. Here the torque converter (cr a ventilate- ^" fluid coupling) , may be used as an active vibration αampe: ry having one or more sensors 142 connected to impute 141 :: computer 144. The sensors 142 sense an impending npaiance and the computer 144 contains a program controlling valves 12^ and 13?" to sets cf pasεageε with their inlets cfϊset radialiv as snowr..

SUBSTITUTESHEET

TECHNICAL DESCRIPTION OF THE INVENTION

I. Introduction

This document discloses a new invention relating to hydrodyna ic transmissions. The purpose o- the invention is to provide an improvement in the overall range— eak efficiency of the conventional torque converter. This is accomplished by a change in the structure o-f the flow passages within the device. The novel structure provides the means -for energy trans-fere utilizing streamwise vortices. A streamwise vorte:. is a rotating -flow which has its a;:is o-f rotation aligned with the mean stream direction.

Streamwise vortices euist in all conventional turbomachines in various ^■ forms and are termed "secondary ^■ flows". Such -flows ordinarily contribute to energy losses and a good conventional design suppresses them as much as possible. However, the dissipation of the energy of a streamwise vorte:: is not a rapid process. Indeed, such a vortei: is lcnown to be a very stable structure capable of suppressing turbulent energy dissipation. But, a conventional turbomachine is not equipped to recover the energy transported by a streamwise vortex and so this energy is eventually lost.

In contrast to this, the new invention provides a turbomachine with fluid passages and a structure which confines and stabilizes streamwise vortices within these passages. Further, the passages are provided with a structure which forces separation along salient edges deposed along the average direction of flow. These edges generate, enhance, or reverse

streamwise vortices so that energy can be transferred into or out of the machine efficiently over a wide range of operating conditions. Thus, the energy loss normally associated witn secondary flow does not occur. Instead, the energy of this flow is put to wαrl. to increase the efficiency of the machine.

In part II the basic principles of the application of streamwise vortices is explained, Also, the results of experimental work to date is summarised. Part III discusses the nature of the improvement in the basic turbulent flow environment produced by the use of the invention. In Part IV the detailed construction of a machine employing the invention is given. Part V considers the improvement in the range—peal: efficiency which can be expected from the application of the invention.

SUBSTITUTESHEET

II. Basic . Principles of Streamwise Vorte>: Applications I I-l. Limitations of Conventional Torque Converters

A hydrodyna ic transmission consists of a pump and a turbine. The pump is driven by an internal combustion engine which is essentially a constant speed device. The turoine ι _→ connected to a load whose speed may vary from zero to a valun¬ equal to that of the ennine. The purpose of the transmission is to transmit the energy from the engine to the load as efficiently as possible over the entire range of output speed.

A pump and turbine arranged as a hydrodynamic transmission rύ θmfΛ/ { ι ^r tg. ,2, J.1. In this scheme a major part of the fi-x?rgy dmXx erad by the engine is converted to inetic energy of the fluid by the runner of the pump 2.1.1. This lcinetic energy is then transformed into potential energy by tne diffuser, 2.1.2. The potential energy is then transformed Dad; into lcinetic energy by the nozzle 2.1.3 and finally this energy is extracted by the turbine runner 2.1.4. The fluid is tnen led back to the entrance of the pump to complete the cycle.

The arrangement shown in Fig. 2.1.1. is very inefficient because of the flow energy losses which occur in the dlffu≤ r. These losses were eliminated by an invention due to Foettinger who placed the turbine runner in close proximity to the pump runner without an intervening diffuser. Because of tni '- invention made at the turn of the century, hydrodynamic

transmi ssions achieved eff1 ci enci es which m=,de them competitive with other forms of transmissions. To this day Foettinger' = invention is employed in all commercial hy__rQ_.yn_.mic trans issi ons.

The nature of the flow energy losses which occur in a diffuser are illustrated in Fig. 2.1.2-a. Before discussing these losses, it is necessary to establish the relative importance of the inertia effects and the viscous effects which operate on the fluid used in a practical machine. This rati is the Reynold's number and is cf order 10 . Thu≥, inertia effects must predominate everywhere in the fluid except in regions where the fluid shear is very high. In the unstalled diffuser of Fig. 2.1.2, the effects of the viscous forces are confined tα a thin boundary layer on the wall of the device. The flow within the boundary l yer s turbulent.

Suppose the flow enters the diffuser with a uniform specific kinetic energy - . As the flow progresses through the device, the kinetic energy in the boundary layer is dissipated by viscous forces. This limits its ability to progress in the direction __> against the increasing specific potential energy yr

If the gradient becomes too large, the flow will separate and the entire flow pattern will change. The diffuser is said to be stalled and the energy losses will greatly increase, see Fig. 2.1.2-b. An efficient design must limit ^S y/J to a value which can be supported by the turbulent tranfere of energy into the boundary layer. This is a fundamental limitation on the efficiency of diffuser. The efficiency is

SUBSTITUTESHEET

further compromised by unsteadiness and non-uni αrmities in the inetric energy of tha approaching flow.

Similiar considerations apply to passages within the blade arrays of hydrodynamic transmissions. There is a definite limit to the divergence of the passage which can be tolerated before low separation occurs. This is Illustrated in Fig. 2.1.3 where the passage is denoted 2.3.1. ->

In the off-design condition the local angle of attack c becomes large. imposes an adverse pressure gradient %- ^> ° in the vicinity of the nose of the blade, sao Fig 2.1.4. Here the flow must accelerate to turn the corner 2.4.1 so that the pressure drops to a very low value. After the corner iε turned, the kinetic energy must then be converted bad: into potential energy and the problem of the adverse gradient threatens separation near the nose at position 2.4.1.

Thus, the fundamental design compromise for airfoil shaped blades is as follows, see Fig. 2.1.5-a and 2.1.5-b. The thickness and solidity of the blade must be made large enough to avoid separation at the nose in order to minimize the hydraulic shock loss. But, the passage between the blades must not diverge more then some optimum amount. Both requirements can not be satisfied at the same time. Thus, a compromise results in range—peak efficiency performance which can not be avoided by conventional designs. These limitations arise from the same problem of the adverse pres¬ sure gradient in a diffuser which was overcome in part by Foettinger.

In the present invention the conversion αf kinetic energy

tD potential energy in the direction s in the presence of a boundary layer with »Ois also eliminated. This is accomplished by providing the blades with a salient edge deposed al onπ a streamline. As the angle of attac of the blades is increased, the kinetic energy of the tangantial component of the flow is converted into streamwise vorticity, not kinetic energy in the flow direction. As with the Foettinger invention, it is not necessary to convert the kinetic energy of the streamwise vorticity into potential energy with a diverging passage. Instead, the vorte:: is led directly into the next stage of the machine which is also provided with a salient edge deposed along a streamline. Here, the vorticity is reversed and a transfere of energy is accomplished with very high efficiency. The details of this newly discovered phenomena are explained in the ne.:t section.

SUBSTITUTE SHEET

II-2. Free Diffusion and Ψorte>ι Reversal

In the absence of & boundary layer, the conversion o-. the kinetic energy of the flow to potential energy can take place at very high efficiency. Indeed, this is eisactly what happens in the flow which approaches the nosa pi the blade as shown in Fig. 2.2.1 Ringlβb recognised this and labeled this process "free diffusion". He attempted to utilise the phenomena in a cusp shaped diffuser, Rβf. 2. The proposed device lε shown in Fig. 2.2.2. The boundary layer approaching tho salient edge 2.2.1 of the cusp is not subjected to a strong adverse pressure gradient. Hence, flow separation is avoided. The free diffusion process occurs along the stagnation streamline 2.2.1. Ringleb gives a two dimensional design method which provides for a trapped, stable vorte:. at position 2.2.3. The device does not operate as designed. Fluid leaving tha salient edge 2.2.1 has lost energy to viscous forces along the boundary. This does not affect the free diffusion greatly, but upon reaching the stagnation point, a part of the low energy fluid 2.2.4 is returned to the interior of the vortex. Thus, low energy fluid accumulates in the trapped flow region and destroys the stability of the vorteji. A large perturbation of the flow occurs and this low energy fluid is expelled downstream. A new vortei. is formed from the high energy -flow and the total energy loss is large. This process continues in

an irregular fashion and the flow over the cusp aittuaer su-ffers large turbulent lasses similiar ta a duct with a sudden expansion or tα a stalled diffuser.

Under special circumstances not considered by Ringled, ._. trapped vorte:: can be utilised efficiently. This requires . special geometry which was first investigated by Saunderi, Ret 3. and confirmed by Rαshkα, Ref -I. Thi'-_ very simple. ιαea formed the basis for the roo mounted wind deflector which initiated the truck, aerodynamic device industry.

Returning to Ringleb's idea, it can be seen that efficient free diffusion is possible with a cusp-like device provided that the low energy fluid is removed. An easy way to accomplish this is to have a component of the flow normal to the plane of the paper in Fig 2.2.2. Thus, the low energy fluid is swept downstream, and a stable vorte:: is generated in the lee of the cusp. However, this is no longer a two dimensional flow and the vorte:: is a streamwise vorte:. since its aι:is is parallel to the normal component of the flow. Ξuch a low is truly three dimensional since the vorte:. must have a beginning and an end. But, it is very helpful to conceive of the flow aβ having a two dimensional section given by _> picturu such as that of Fig. 2.2.2 cut from a more general low with a component normal to this section.

The new phenomena of vortex reversal is seen tc involve free diffusion. In order to transfere energy into the blaαε array, pressure forces are required. Thus, a conversion of l-.inetic energy to potential energy must take place. But, this

SUBSTITUTE SHEET

transfere is now done by free diffusion which brea s the requirement ai a diffuser duct wlth*// S opposing the boundary layer along its wall. Thus, the new phenomena yields very high efficiency of energy transfere. Experimental data illustrating this new phenomena are given in the next two sections.

II-3. Experiments with Two Dimensional Vortex Reversal

A3 explained in section II-2, it is very helpful to think of a streamwise vortex as a two dimensional flow in a plane normal tα the axis of the vortex. Accordingly, experiment_> were undertaken in <ι water table to investigate the key phenomena of vorte:ι enhancement and reversal. If the flow is unsteady, accumulation of low energy particles in the cusp vortex can be avoided and the phenomena is easily demonstrated. The first series of experiments to be described concern vortex reversal in a linear array of co-rotating vortices. Tho array is generated and reversed in the water table with an arrangement shown schematically In Fig. 2.3.1. The cusps are formed of circular sections and their dimensions are given in Fig. 2.3.2. The cuops are carried by a towing cart and 3rQ moved through the water parallel to a line connecting their salient edges. The motion consists of a constant speed translation in one direction followed by a sudden reversal to a simlliar motion in the opposite direction. After several repetitions of this motion, a cyclical state Is achieved. The resulting flow of the fluid Is made vlsable by the use of hydrogen bubbles emitteα from a wire stretched between two cusps.

During each cycle of the eκperiπ.βnt, each vortεx of the array is observed to reverse direction without any indication of instabili ies or turbulence. The core of each vortex moves

SUBSTITUTESHEET

aiong a line connecting the edges of the cusps suggesting that the vortex is stable along this line. In the photo¬ graph of Fig. 48, a counterclockwise vortex is observed when the motion of the cart is to the right. The right hand side of cusp is at a higher pressure than the left hand side. When the motion is reversed, the counterclockwise vortex s observed to unwind and then to wind up again into a clock¬ wise vortex to the right of the salient edge. Similarly, the pressure and suction sides of the cusps are reversed. During this motion, the exchange of pressure and kinetic energy as free diffusion occurs away from the developing boundary layer and so is virtually loss free.

To investigate the effects of turbulence on the flows shown in Fig. 48, a grid was used to generate a homogeneous turbulent flow in the tank before the motion was begun. Even large amounts of initial turbulence failed tc change the basic vortex reversal phenomena.

The next series of experiments to be described concern a linear array of contra-rotating vortices, see Fig. 2.2.2. The photograph of Fig. 49 shows the flow which results. In this case the vortices do not change their sense cf rota¬ tion. However, when the cart is accelerated, the vortices are energized and translated with the cart, while then- cores remain centered aε shown. When the motion is such that the vortex strength is increased, the array of vortices is said to be enhanced by the motion. In Fig. 49 the vortex sheet leaving the salient edge is seen to be bending to the right hand side for all cusps. When the motion is reversed these sheets bend to the left. A shadow graph of this = snown in the photograph of Fig. 5C. The shadow graph tech¬ nique is described in Ref. 5 and is useful for picking out details of the flov;. In this case the spiral roll-up c tr_-- vorte : sheets easily seen. If a uniform flow perpendicular tc the plane cf Fig. 50 is imagined as convecting this spi¬ ral sheet downstream, a useful idea about the roll-up cf tne three dimensional sheet can obtained.

The photograph cf Fig. 51 is a shadow graph cf tne same set up as Fig. 50. However, in Fie. 51 the cart s driver so fast that an intense series of surface wavelets are gen¬ erated at the cusp edges. These wavelets perturr. tne flov in a manner similar to turbulent eddies. It car. oe seer, from Fig. 51 that the line of eddies is rolled up into tne larger vortex in a way that is similar to the roll up cf tne vortex sheet of Fig. 50. This illustrates the tendancy toward stabilization of transverse turbulence DV a vorte:.. It is an illustration only and not a demonstration cf the real effect because the turbulence is artificially con¬ strained bv the presence cf the free surface.

SUBSTITUTESHEET

II-4 Streamwise Vortex Flow Visualization

A cascade wind tunnel as shown in Fig. 2.4.1 is used tc study the three dimensional generation and reversal cf streamwise vortices. The first experiment tc be described concerns an array of tubes which generate a stable contra- rotating set of vortices similar to the two dimensional flov; of Fig. 49. The cascade is formed cf 3 inch ID thinwall tubes cut as shown in Fig. 2.4.2. The inlet portion of the array has a rounded nose constructed with clay. The flow iε visualized using a smoke wire.

The photographs of Figs. 52 and 53 show the generation cf a streamwise vortex which is seen to persist downstream inside the tube. The two photos differ only in a slight change of the camera angle. In Fig. 52 the flow along the lee side of the cusp is visable. It is seen that the flow rounds the nose of the cusp and spirals up to the salient edge. No hydraulic shock loss occurs in the flow even though the flow is approaching at a local angle of attack of C , = 40 degrees.

In Fig. 53 the camera views the cusp from a position in line with the salient edge. This viev; allows the position of the vortex core to be observed. Comparing the core loca¬ tion in this photo with that of Fig. 49 shows that the three dimensional vortex tends to form further down behind the edge of the cusp. However, the core of the vortex does move toward the stable position at the center of the tube as it progresses downstream. The initial portion of the stream¬ wise vortex is more elliptical then the two dimensional one. But, its shape also approaches the two dimensional circular shape as it moves downstream.

Looking to the left of the center cf Fig. 53, the stag¬ nation line terminating on the top wall cf the array can be observed. This line separates the counterclockwise rotatmr vortex cn the left from the clockwise rotating vorte:: or; the right. The stagnation zone terminating this line lies along the end wall cf the cascade in a streamwise direction. The

free diffusion process is seen to be operating along the stagnation streamline undisturbed by a low energy bounάar'- layer or shear flow. Thus, the process of generation cf tne vortices is virtually loss free except for the thin vorte:. sheet which leaves the salient edge of the cusp. The dis¬ turbances from this sheet are so small that they are not visable in the photograph.

As discussed in section 3.1 an investigation by Smg , Ref. 6 suggests that a streamwise vortex may be unstable when the core of the vortex contains an axial jet in the streamwise direction. Accordingly, an experiment was per¬ formed with an axial net issuing from a nozzle αirected r.tc the cascade. The results are given in the pnotograons cf Fig. 56 wnere the jet nozzle is seen to the left. Smoκe wires are located directly m front of the et and on a strut placed in front of the cascade. The total head of the jet was twice that of the free stream. Smoke m the more turbulent regions of the jet is not visable. However, a time exposure photograph is able to capture enough average streamlines to show that the streamwise vortex still forms in the tube in a stable fashion. Neither the intense turbu¬ lent mixing nor the axial velocity of the jet changes tne essential features of the flow.

Similar experiments were made with arrays of tuoes tc generate all of the variations of streamwise vortices shown m Fig. 2.4.4. In all of these cases stable vertices are easily generated. An interesting variation in the structure of the cusp of a co-rotating array is shown in the photo¬ graphs cf Figs. 54 and 55. The planform of the salient edσc of the cusp has been modified to a sawtooth snape , see F .. 55. This generates an interlaced spiral of three hel ca_ streamwise vortices all of which are stable.

Streamwise vortex reversal has also beer, ooserve. : the cascade wind tunnel. The set up is shown m F e. _ . - .1. An initial set cf tubes is used to generate the vortex arrε; which is then reversed by a second set of tupeε. Ir. add_- ticr. tc the stead'- state reversal, uπsteac- reversal r.a;

SUBSTITUTESHEET

been observed when the first cascade is translated oast the second cascade. Since the smoke must be observed inside the curved walls of the model, photographs of the flow are dif¬ ficult to interpret and are not included. Ir. all cases tested, however, it is possible to observe by eye the gener¬ ation of stable streamwise vortices in the second cascade of tubes .

In addition to these observations a number of experi¬ ments have been carried out which demonstrate streamwise vortex generation from salient edges which are curved with respect to the approaching flow. Also, streamwise vortices have been observed which are generated by a salient edge at the rear of a cascade when the edge is deposed at an angle with respect to the stream and is loaded by a cross flov;.

The smoke wire technique is limited to low Reynolds number experiments. However, it is easy to extend the ob¬ servation of the flow pattern to higher Reynolds number by using the wool tuft technique and the stethoscope probe technique. The Reynolds number is increased to the value existing in the full scale torque converter by increasing the tunnel speed. The flow pattern is observed not tc change with speed. Thus, these results are independent cf Reynolds as expected. Since the flow pattern does not alter with Reynolds number for stationary arrays, it is net ex¬ pected to vary when moving cascades are used.

Additional experiments were performed in an axial flow rig shown schematically in Fig. 2.4.6. These experiments were concerned primarily with the effects of centrifical and Coriolis forces on the generation and reversal of streamwise vertices. In addition this rig does not have an artif cal boundary condition at the ends of the array as does the cascade tunnel. Thus, the cascade data car. be verifier. Experiments on tube deformations, nested tubes, bent faces, and cresent shaped εtatore were all carried out with tms rig as indicated in Figs. 2.4.6-a through 2.4.6-c. Vorte:: breakdown was observed with the model shown with the label

A.11 cf the experiments performed wit:

this rig support the contention that the present mvent cr will be successful if designed within reasonable limits.

The general impression of all of the opservat ons αe- scribed m this section is that tne generation cf streamwise vortices is a very powerful mechanism which w l_ operate over a wide range of approaching flow angles, vorticit cistπbutions, and turbulence levels.

SUBSTITUTESHEET

III. Modificatαns to the Flow Caused by Streamwise Vortices

The nature of the turbulent flow in turbomachinery is not well understood. Nevertheless, it is necessary to consider this complicated phenomena if real progress is to be made. This section discusses the modi ications to the flow expected from the introduction of streamwise vortices and the implications for the design of the machine. 11I-l. Persistance of Energy of Streamwise Vortex

The streamwise vortices shed by the wings of large aircraft persist for miles. This poses a seriours hazard for a light aircraft attempting to land following the arrival of a larger commercial jet. Consequently, the fluid dynamics of this problem has attracted a great deal of attention in recent years. Various devices have been proposed to break up these vortices and dissipate their energy rβpidly. So far this problem has resisted a practical solution. The reason is that a streamwise vortex is remarkably stable and it iε just this stability that is exploited by the present invention.

Quantitative measurements on a streamwise vortex are given by Singh, Ref. ά. The vortex is generated by an airfoil spanning one half of a wind tunnel. The arrangement is sketched isometrically in Fig 3.1.1. The coordinate system used is shown in Fig 3.1.2. Singh reports the results of hot wire measurements of the flow variables at various downstream stations, 2, up to a value of z/c » BO where c is the length of the chord of the wing. The Reynolds number of the experiment is comparable to that found in a torque converter.

In order to establish a basis of comparison of the data o Singh with that of the proposed device, it is necessary^ to restrict our attention to _. the flow in the neighborhood of the tip of the airfoil. A control surface for this purpose is sketched in Fig 3.1.1 and consists of a. ylinder whose axis Is centered on the trailing vortex core. The flow through the circular sections at 2/c ^β O.B and s/c_-5._> are given by Singh. For these two stations the length to diameter ratio ot the control cylinder is 3.S. In the impeller of Fig. -1.1.4 the streamwise vortex is generated by salient edge 4.1 beginning at staion 0 and Is conducted by a circular channel to the e.it of the impeller at station 10. The length to diameter ratio of this channel is also about 3.5.

If a substantial percentage of the rotational kinetic energy entering the control surface at station :/c=0.Q is dissipated by the time the flow arrives at station z /c- Z . ύ , then there iε no hope for the invention. To examine this question it is necessary to perform some calculations on the Singh data ιand this is explained in the fallowing paragraphs.

Let N be an arbitray extensive property of the fluid with a distribution per unit mass of y ,w» i _{t r} -

<___.✓. where the integration is taken over the volume enclosed by the control surface. The total time derivative of N is given by

the control When N i s the mass of the f l ui d η = 1 and ^- ;θ θi ^v e- the

SUBSTITUTE SHEET

equation of continuity.

The quantity of primary interest is the rotational kinetic «-. Ϊ- energy s. . . But first it is necessary to consider the angular momentum of the fluid 2/ =. j^-- -. ^ _s torque on the

_____ control surface, denoted by r is given by

C.S. c.v.

Using the data of Figs. 2 and 3 of Singh ^• the total angular momentum entering C.Ξ. are

span of the wing for both the wing root side and the far wall side of the tunnel. These curves are not symmetrical. _H ence the total angular momentum associated with the two curves are different and lead to the different curves of Fig. 3.1.4. However, the total angular momentum associated wi _t h either curve does not vary between the two stations within the limits of the resolution of the data.

An interpretation of the above results follows from the observation that no significant shear stresses act on the cylindrical portion of the control surface. Thus, total angular momentum must be conserved. On the other hand, pressure forces on this surface are active and are responsible for the enduring change from one 17^. profile to the other. These forces result from the fact that the varte.s core is net centered in the tunnel. The image of the vorte;; in tne winr

root wal 1 has a stronger effect and induces the l rger value o.

The streamwise vorten core is stable within the confine- of the duct. When the flow rotates toward the far wall, l inetic energy is stored as potential energy by pressure forces e.:erted by the wall of the duct. As the low contiues in its helical path this potential energy is transferred bad, to I inetic energy of rotation. This interchange takes place very efficiently wi h virtually no energy lost in the process. Ξimiliar efficient processes are encountered in the new invention.

Having established that the angular momentum inside thu control surface is conserved, it follows that the vorticity can only change by diffusing outward from the core. This process will continue until the fluid is in solid body rotation where the metic energy of rotation will reach a mimi um. The l.inetic energy of rotation is a component of the total l.metic m m. energy zshi -iM

Thu- second integral on the right hand side measures the rotational lcinetic energy. It iε computed as a function of r/c, and the results are plotted in Fiy 3.1.Ξ as a percentage loss in rotational kinetic energy between stations :/c-- O.B and r/c= Ξ.O. The results are referred to the entering rotational

,.,-_ ^. _= energy _^ K _tmm &' .θ) ~ ^.m < s)

The energy loss occurs principally in the core region, r /c < t /

SUBSTITUTESHEET

0.1. The total loss does not exceed ten per cent. Also plotted in Fig 3.1.3 is the same result for =/c=B0, a value over 20 times further downstream than is of interest for the Invention. Even in this eutreme case the loss in the rotational lanetic energy does not exceed -}0V.. It is claar that there . is very little inherent dissipation of energy in a streamwise vorteu over the distances of interest for the invention.

For an aircraft the energy that trails as a streamwise vortex pair is not recoverable. Thus, it represents a drag loss which must be supplied by the propulsion system. This drag is termed the "induced drag" in the literature. It iε calculated by equating the rate at which wαrlc must be done in overcoming this component of drag to the rate at which rotational lcinetic energy is added to the air. Thus,

In the novel machine the "induced drag" o a finite airfoil does not result in an energy loss. Rather, the rotational kinetic en-ergy is e ficiently -transferred to a following stage of the machine.

The thrust of Singh's paper iε to discover instabilities in the streamwise vorte;;. As pointed out in the discussion, part VII, he is only able to do this with a very special wing of high section L/D > 40. According to Singh, instabilities are not found in streamwise vαrte;< systems with large initial turbulence. Yet, this is the operating environment of the invention. Even when Singh manages to produce an instability

it does not lead to violent vorte.: breakdown.

In contrast to this the designer of the novel machine can choose parameters to enhance vortex stability. For example, the planform of the salient edge need not bo rectangular. The experiments reported herein use an elliptical leading edge planform. The corresponding modification of the airfoil o-f Singh is shown in Fig 3.1.3. A careful study of these and similiar photos show no signs of instabili ies developing in the streamwise vortices springing from these elliptical leading edges.

SUBSTITUTESHEET

III-2 Flow Stratification and Streamwise Vortices

The previous section examined the suppression of turbulence produced by a streamwise vortex whose a:<is w s a straight line. If such vortices are to be used in a centrifical machine, the effects of curvature must Qo considered. But first it iε nceεεary to consider the turbulence and curvature effects already present in centrifical machines.

As explained by Johnston, fte . 7, both wall curvature and wall rotation affect the turbulent boundary layer. Fig. 3.2.1 shows the conditions under which stabilisation and destabilizatiαn occur. Baljβ, Ref. i, pp. 376-3B9 applies these principles to a description of the flow found in a centrifical rotor. This flow is dominated by the turbulencπ mechanism and is very complicated. It is the subject of ongoing research and a real understanding of the flow does not exist. Nevertheless, enough data does exist to show what will happen when a streamwise vortex iε impressed upon this flow in. accordance with the resent invention.

In Figs. 3.2.2-a through 3.2.2-c a flow through a cen ^¬ trifical rotor is shown schematically similar to that dis ^¬ cussed by Baljε in connection with his Fig. 6.36. Flow stratification effects result in a low energy core being formed in the corner where the suction side and the shroud side of the passage meet. In the torque converter the hub corresponds to the shell and the shroud to the toroidal core. Figs. 3.2.3-a through 3.2.3-c shows the effect of imposing a streamwise vortex on this flow.

The vortex is generated by a salient edge

placed so that the vortex core is closer to the suction side and the shroud side of the passage. The central portion of the streamwise vortex suppresses the turbulence. When the exterior of the streamwise vortex interacts with the wall, the boundary layer turbulence is increased. Thus, the extent of the law energy core is reduced and separation of the flow -is suppressed. Low velocity particles accumulate in the vorte;. core rather than on the walls of the passage. Phenomena similiar to this has been observed in the cascade tunnel. The results are described in section IV-1.

Since the vortex core is further from the hub side and the pressure side, the increase of the turbulence on these walls is not as great as for the opposite side of the passage. Nevertheless, the transport of turbulent energy into these boundary layers is increased and so the blade loading should be increased for optimum results when streamwise vortices are present. One way to increase the blade loa ings to use multiple vortices within a single passage. For example, a contra-rotating pair of vortices within a single passage is shown in Fig. 3.2.4. In this design a cusp is used to further stabilise the pair as they are subjected to the Carious farces operating in the passage. Notice that the total wetted area of the passage is reduced over a passage constructed of two-dimensional elements. This is further discussed in section IV.

SUBSTITUTE SHEET

III-3 Skin Friction Reduction by Laminarination

This section describes an additional inventive step which is not necessary to the operation of the machine. This step also results in an overall improvement in the range—peal: efficiency of a torque converter by reducing the skin friction losses even in the on-deεign condition. It exploits the fact that streamwise vortex energy accumulated in a passage of the machine can be recovered in a successive stage.

The idea is illustrated schematically in Fig. 3.3.1. The hub side and the pressure side of the passage are the generators of the greatest portion of the turbulent slcin friction losses. Hence, a number of small co-rotating streamwise vorteu generators are placed at the entrance to these surfaces. These generators produce a stagnation zone on the wall as discussed in" section 11-2. The flow leaving the stagnation zone is laminar and its skin friction is correspondingly reduced. As the flow progresses over this surface it encounters another line of vortex generators of larger height. These create additional zones of laminarised flow on the wall of the passage. In addition, they roll up the streamwise vortiςes shad by the preceβding row of generators. This process is repeated until a single large streamwise vortex is obtained in a stable region of the passage where its energy can be extracted by the next stage of the' machine. An appreciable region of laminar flow has been achieved on a surface which would otherwise be subject to very intense turbulence. Since the slcin friction produced by laminar flow

-43-

ιs much lower than that for turbulence, a substantial net reduction in energy losses is achieved.

It should also be mentioned that the conventional use of small vortex generators to energized the boundary layer and suppress separation can be used with the invention. With a schema such as in Fig. 3.3.1, the energy consumed by these generators can be partially recovered rather than being last as in conventional applications. Thus, this application is more efficient in the present caεe.

T HEE

II1-4 Vortex Breakdown

Vαrteκ breakdown occurs when a streamwise vortex enters passage which diverges too rapidly. The core of the vortex does not have enough momentum in the £ direction to overcome the pressure gradient. In this it is similiar to diffuser stall mentioned in section II-l. However, the breakdown of the flow into turbulence occurs in the Interior of the fluid away frσm a solid boundary. The subject of vortex breakdown Is currently the topic of much research and is not well understood, see the review by Hall, Ref. Q.

The phenomena is shown schematically by the sequence of Figs. 3.4.1—3.4.3. In Fig. 3.4.1 a Veπturi tube is shown with conical diffuser. The flow approaches from the left and the pressure is plotted below the section of the tube. For a low diffuser angle, the pressure loss J> is small. In Fig. 3.4.2 a streamwise vortex with swirling velocity is shown approaching the tube. As explained previously, the swirl increases the transfere of energy across the wall boundary layer. Hence, a larger angle of diffuser will give the same low value of Λ . If the diffuser angle is made larger still, vortex breakdown occurs. This is illustrated in Fig. 3.4.3 by 3.4.1. This results in a sharp increase in the pressure loss Ap.

The center of the tube is filled with violent turbulent eddies which do not contact a solid surface.

Clearly, vortex breakdown represents a limitation to the efficient application of streamwise vortices. In the experiments conducted to date, vorte;; breakdown has only been

observed in connection with a flow entering a sudden expansion of a duct with an abrupt area increase of two to one. Thus, it does not seem to be operating under the conditions o-f interest for the invention. However, future work must be prepared to observe this phenomena if It occurs.

Several methods of overcoming vortex breakdown are available. First, the divergence of the ducts can be limited as- in the conventional machine. Indeed, the design at any airfoil shaped blade Is essentially an effort to manage the boundary layer pressure gradient for maximum efficiency. If the pressure distributions are optimum for these boundaries, vortex breakdown will probably be avoided. It has o.nly been observed in the present experiments under extreme conditions.

The second method consists of energizing the vorte:: core. In this way the maximum value of--^ can be Increased. An arrangement such as that shown in Figs. 54 and 55 _wl 11 accomplish this. The average core shed by the sawtooth salient edges in these photos contains more energy than that of a single core. Hence, it is more resistant to vorte:. breakdown.

Λ third method is of interest in its own right and consists of operating the torque converter partially flllec with air. The machine Is still charged with sufficient pressure to prevent cavitation, but the air is introduced into the vorte> cores. The viscosity and density of the air are very much lower than that of the transmission fluid and so the vorte:: cores are essentially stable and loss free. The fact that the air will always remain in tne center of the passage-:

SUBSTITUTESHEET

of the machine through-out the circuit points out the essential differences between the present invention and a conventional machine.

The idea is illustrated for the Venturi tube in Fig. 3. ■] . where an air bubble 3.4.Ξ is trapped in the core of thu swirling flow. The bubble is at an approximately constant pressure and so the graph of the pressure is modified as shown in the figure. Since the core of the streamwise vortex i - not greatly affected by the air-oil interface, the kinetic energy of the core can overcome the γ^S in Fig 3.4.4. Notice that the same tube without the air bubble would suffer vortex breakdown, Fig. 3.4.3.

To apply the idea to a practical machine requires that an air charging system be used as well as an oil charging system. The air system could incorporate an air trap at a high point of the system where air escaping from the torus would eventually accumulate. The trap could be provided with a sensor which would indicate when the supply of air was low. The charging pump would, then supply air from the atmosphere. If the air supply were adaquate the pump would supply air from the trap to tha injection point at a vortex core beglning at the stator. If it were desired to vary the capacity of the machine, the vortex cores containing air could be increased or decreased thus providing an efficient means of control.

IV Construction

In this section various constructions cf the machine are given. The purpose is to teach the application of tne invention. Thus, not all permutations and combinations are covered.

The energy transferred m a torque converter is porpcr- tional to the product of the flov/ rate and the change ιr. periphial velocity. When impacting on an appropriate blade surface, the effect of the streamwise vortex is to auσment the change in this periphial velocity. Thus, a good way tc investigate the application of the invention is to start with a proven conventional design with a good flow rate anc modify it to accept streamwise vortices. IV-1 Modification to Jandase ' s Design

Jandasek, Ref. 9 gives explicit details regarding ε contemporary torque converter design intended for automotive use. This design represents a good starting point to ex¬ plain the modifications required by the present invention. Fig. 4.1.1 corresponds to Jandasek ' s Fig. A-l and shows the development of a torus for a 12 inch diameter machine. Fig. 4.1.2 corresponds to Jandasek ' s A-2 detail of the impeller blade. Figs. 4.1.3-0 through 4.1.3-10 shows six sections perpendicular to the flow through the impeller developed onto the plane from frustrums of cones through sections C through 10 of Fig. 4.1.3-a.

Figs. 4.1.4-a through 4.1.4-b show a modification tc the impeller blade of Fig. 4.1.2 in accordance with tne teaching of the invention. The leading edges of alternate impeller blades are cut away to form salient edges 4. i and 4.2 which are deposed partially along tne mean streamwise direction. These edgeε form contra-rotatmσ streamwise vortices n the regions cf station G tc , the sections c z ^' Figs. 4.1.6-0 to 4.1.6-4 which are convect.ee DV the mea: flow in the sections cf Figs. 4.1.6-6 to 4.I.6-1C. A fur¬ ther modification is shown m Fig. 4.1.5 wherein the bladr- connected to the shell of the torus s further reduced.

SUBSTITUTE SHEET

The square sections of Figs. 4.1.3-0 to 4.1.3-10 are now replaced by sections containing cusp-like sections as shown in Figs. 4.1.6-0 to 4.1.6-10. In this figure, the view of station 0 shows the flow approaching a contra-ro¬ tating array of cusps similar to that shown in Figs. 52 and 53. Streamwise vortices are generated by these cusps which extend to station 4. Station 2 shows how the cusps are used to guide and stabilize the roll-up of the vortices. Gener¬ ation of the vortices is complete by station 4 and stations 6 through 10 conduct the fully formed vortices. Since these vortices enhance the stability of the flow, the number of guiding surfaces can be reduced. Thus, the blade in Fig. 4.1.5 has been removed. By station 8 this blade is com¬ pletely gone. The contra-rotating pair is conducted through station 8 and discharged out of station 10 which is a view shown facing the outflow. The radial edges of the passage are rounded so that a sudden expansion is avoided.

The construction shown in Figs. 4.1.4 through 4.1.6 is chosen to be a simple modification to Jandase 's impeller. Thus the modified blades are inserted directly into the original core-to-shell space. A plastic model of this modi¬ fication has been constructed and is shown in Fig. 57. The model has a diameter of 28 inches and comprises a segment of the torus containing four core blades and three shell blades. The salient edges 4.1 and the edges 4.2 are visi¬ ble. A baffle is also shown which extends upwards from the core of the torus. This baffle is added only for experi¬ ments in the wind tunnel.

Experiments with the device shown in Fig. 57 were per¬ formed in the cascade tunnel. Enough time has not been spent to properly adjust the baffles to achieve true radial cascade flow. Nevertheless, certain qualitative observa¬ tions are of interest. First, contra-rotating vortices do form as implied by the linear experiments. Second, these vortices persist through the curvature without any indica¬ tion of breakup or excessive energy dissipation.

An additional series of experiments was run wirr. tr.ii device with the cusps blades replaced by thin snee radial blades very similar to Jandasek ' s design. At only moderate angles of attack, the leading edge cf these blades stailec. The slow moving particles from the stalled region migrated tc the core side of the torus. The flow on the snell sic-: was choked by this accumulation cf slow moving particles. The effect is very similar tc that described m ccr.r.ectior with Fig. 3.2.3. IV-2 Modifications to Jandasek ' s Turbine Runner

Figs. 4.2.1-a and 4.2.1-b correspond to Jandase ' s Fie. A-2 of the turbine runner. The modification reσu re t; recover the energy of the streamwise vortices is snowr. r. Figs. 4.2.2-a and 4.2.2-b. Again the entrance pcrt cr. ci the blades has been cut away to provide salient edges 4.2 and 4.4. The flow approaching these edges contains contra- rotating vortices. These vortices are reversed by either 4.3 or 4.4. Thus, pressure forces act on the cusp-like projections to transfer energy to the runner in the correct direction regardless of the sense of rotation of the ap¬ proaching vortex. An additional modification is also snowr. n Fig. 4.2.3. Here, most of the shell blade has oeen re¬ moved to increase the loading m a manner similar to that discussed m conduction with Fig. 4.1.5.

Additional modifications to Jandasek ' s design are snowr. in Figs. 4.2.1 through 4.2.2-b. Here, the trailing edge cf the blades are also cut away as shown by the dotted line ir. Fig. 4.2.1-a. As well, the blade has been extended to over¬ lap the space originally occupied by the stator, see Fig. -1.2.2. (This modification could also be made tc tr.; trailing edges of the impeller blades but is net sr.owr. fcr simplicity.) These cut away portions form salient eάcee and 4.6. If these blades are curved in accordance Jandasek's desigr. criteria they will enhance tne energy c: tne streamwise vortices present m the turbine cnanr.els.

Fiσs. 4.2.4-0 throuσh 4.2.4-10 snow sections tnro c.

SUBSTITUTE SHEET

various stations of the turbine containing additional modi¬ fications as taught by the invention. Station 10 is a view of the entrance to the turbine as seen by an observer moving inward with the fluid. The cusp-like modifications are similar to those described in connection with Fig. 4.1.2 of the impeller. The additional features shown in Figs. 4.2.4-0 through 4.2.4-10 become apparent upon examination cf station 4. Here the vortex pair iε constrained by a channel which does not fit inside of the core-to-shell space of Jandasek. As the flow proceeds to station C the angle between the confined pair of streamwise vortices is rotated so that the flow exits as a stacked array of vortices similar to that shown in Fig. 2.4.4.

IV-3 Modification to Jandasek ' s Reactor

Figs. 4.3.1-a to 4.3.1-b correspond to Jandasek ' s Fig. A-4, a detail of the reactor blade. Modifications to this reactor are shown in Figs. 4.3.2-a through .3.4.-D. The streamwise vortices leaving the turbine impinge on salient edges 4.6 and 4.7 where they are reversed. Although these edges are in a radial plane, the approaching vortex array iε stacked so that the peak of the vortex generation occurs at opposite times on edges 4.6 and 4.7.

A modification to the trailing edge is shown in Fig. 4.3.4-A, section A-A. The trailing edge is rounded, not sharp. However, this edge is still a salient edge affecting the streamwise vortex. This is because of the close interaction of the trailing edge and the leading edge cf the impeller input blade, Figs. 4.1.4-a and 4.1.4-b, edgeε 4.1 and 4.2. The flow oscillates over these edges in the manner cf the flow leaving a "Thwaiteε Flap", thereby producing efficient streamwise separation without a sharp edge. . ^: . similar effect car. be utilized ir. connection with the impeller or turbine trailing edges, e.g ^" . edges 4.5 and 4.6 cf Fiσs. 4.2.2-a and 4.2.2-b.

IV-4 Variable Geometry

Variable geometry can be provided m the torque con¬ verter cf tne present invention by botn conventional and unconventional means. Besides tne use cf a one-way cuter. (which is essentially a variable geometry device) , tne use of stator blades movable m pitch is widely accepted. Fig. 4.4.1-a shows a blade 4.3 with a pivot 4.1 and bell crank 4.2 similar to reactor blades used in automotive torque converters. However, the blade has been modified by tne addition of salient edges 4.4 which cause the leading edge cf the blade tc function as a streamwise vortex generator such as 4.6 and 4.7 of Fig. 4.3.2. In this case the cusp¬ like surface is formed like a cresent shaped pyramid aε car. be seen in the desert where streamwise vortices are gener¬ ated. Figs. 4.4.1-b through 4.4.1-d are top, front and side views, respectively, of blade 4.3 of Fig. 4.4.1-a.

A more novel approach to variable geometry can be seen m Figs. 4.4.2 and 4.4.3 where the special characteristics of the invention are exploited. In Fig. 4.4.2, the impeller blade 4.10 has a trailing edge cutout similar to the turbine blade 4.5 combined with 4.6. A variable element 4.11 is pivoted about 4.12 and suitable means is provided to move this element from the position marked "out" to that marked "in". When the element 4.11 is "in" streamwise vortices are enhanced. Similar arrangements can be used on the leading edges of arrays.

A similar idea is shown in Fig. 4.4.3 where the blade 4.10 iε now provided with an auxiliary blade 4.14. Wher this blade is "out" the flow is not affected. When suitable means are used to push the blade"in", streamwise vertices can be ennanced. A stator could be constructed using tπ ε idea wherein the blades are completely removed fror tne circuit. Ir. this case the stator need net be provided witr a one-way clutch.

SUBSTITUTESHEET

IV-5 Modification to Cusp-Like Vortex Generators

Figs. 4.5.1-a through 4.5.1-C shows a modification to a cusp-like streamwise vortex generator. In figures 4.5.1-a through 4.5.1-C, the blade is in a cascade utilising salient edge 4.5.2 to generate a leading edge vortex. Suppose that it is desirable to move the core of this vortex further from the following wall. Then the modification shown in Figs. 4.5.1-a through 4.5.1-C can be used. The salient edge 4.5.3 is carried on a crosswise member so that the vortex core iε further from the centerline of the blade at small angleε of attack.

Ref. 12 discusses streamwise vortex generators which are used to energize the boundary layer and prevent separa¬ tion. No provision is made for recovery of the energy of the vortices. A variety of generators is considered in this reference. Any of these generators could be adapted for the purposes of the invention. In some cases modifications would be required to prevent a stalled region developing in the immediate vicinity of the generator.

Mention should also be made of the fact that a finite wing of low aspect ratio makes a fairly efficient vortex generator for the purposes of the invention. For example the first turbine blades of Walker, Ref. 10 could be cut off at half span. This would result in a vortex generator for co-rotating vortices of fair efficiency. Their efficiency could be improved by shaping the cut edges into appropriate sharpened sections .

IV-6 Co-Rotating Designs

Designs based on co-rotating arrays cf streamwise ver¬ tices are also possible. Indeed, a machine can mix co- and contra-rotating vortex arrays. In most cases the co-ro¬ tating designs derive from conventional designs ir. a straight forward fashion and so will not be discussed fur- tner ir. this disclosure.

V. Estimates of Improvement in Range—Peat. Efflclences

In this section an estimate will be given of the order o-f magnitude Improvements which are expected from the use at tne invention. Qf course the actual improvements possible can only be determined from a full scale test of the final design. V-l. Significance of the Shock Loss

In this section, data in Refs. 9 and 10 are utilized to evaluate the importance of the hydraulic shod: loss in a conventional hydrodynamic transmission. The assumption is then made that the energy which iε ordinarily lost is instead utilised by the new invention. Thus an estimate of the magnitude of the improvement can be obtained.

Consider first the paper by Walker concerning a multi-turbine converter. The power input to this machine is

where

c

Rearanglng gives the head as

Now the power input must equal the power output plus the power lost in the fluid -circuit. Thus,

The hydraulic losses are separated by Walker into two parts:

"fijjrt aβ-j iS where H_ is the flow loss proportional to the wetted surface

SUBSTITUTESHEET

of the channels in the machine, and H„ is the shoci: loss which is proportional to the ''instantaneous'' change in the periphial fluid velocity. ,_,

Walker measures the power input and the power output of the machine which determines the total hydraulic loss. He then fits the hydraulic loss data empirically so that the flow loss coefficient remains conctant and the shock loss is proportional to the difference of the velocity squared. The results of this procaedure are given in Walker's figure 19. Figure 5.1.1 reproduces this data with all of the shod: losses added together. The figure shows what happens to the input energy as a function of the speed ratio.

In the invention, the rapid change in periphial velocity is not lost. Instead it is transferred into streamwise vortices which are transported by the main flow. The energy of these vortices is then recovered by the following elements of the machine. Hence, the energy which Is lost to hydraulic shock in <a conventional machine is utilized efficiently in the novel invention.

To estamine the importance of this invention to the overall performance of a torque converter, assume that the sho : loss is z ero and the flow loss coefficient is unaffected. The entire input energy is then divided between output energy and flow loss. These results maybeplotted

Comparison of Figure 5.2.1 shows the effect of this change. The useful range of tne machine is

greatly extended. It is also Instructive to compare these results with Fig Ξ of Pαlak, Ref. 11. Polak is evaluating transmissions for use with the DDE advanced turbine engine. In his Fig 5, he shows that a conventioal three-speed torque converter is competitive with a variable belt drive CVT in the range of vehicle speeds above 20 MPH. The mimimun acceptable efficiency of the transmission is 75'/.. Comparing the 757. point shows that the lower range of useful speed ratios has been extended from a value of 0.5 to 0.3 . This corresponds to the el imlnation of one gear set from a power shift transmission. The savings in cost, weight, and space resulting from this improvement provide an indication of the economic importance of the invention.

The multi-turbine machine described by Walker was produced in the 1950' s for luxury automobiles where smoothness was of major importance. Long before the energy crises of the 1970' s, the multi-turbine device was replaced by the three-speed torque converter primarily because of the better efficiency of the latter device, By applying the principles of the invention, however, the multi-turbine device can be made competitive with the advanced machines considered by Pαlak.

The importance of shock losses for the three element converter can be assessed by examining Figure 15 of Jandasek, Ref 9. The data in this figure is obtained by Jandasek from measurements of the efficiency of the machine in the same way as done by Walker. The details are contained in his Appenαi.,

SUBSTITUTE SHEET

B. Again, tha data depends on empirically fitting various flow coefficients rather than by direct measurement of the flow variables. Nevertheless the strong effect of shod: losses are clear from the magnitude of the change in the inlet angles for the cascades. For example, the stator is subjected to a range of angles of from 20 degrees at a speed ratio of 0.9 to an angle of 150 degrees at a speed ratio of 0.0. The design entrance of the cascade is 90 degrees. Thus, the local angle of attack of this cascade iε 60 degrees at the speed ratio of 0. At the 75"/. efficiency point the local angle is 40 degrees which occurs at a speed ratio of O.S.

Jandasek fits the data by assuming all of the energy associated with the unmatched periphial component of the velocity is lost. This. is the shock loss. But, consider the data given in section II.4 for this same local angle of attack of 40 degrees. Figs. 54 and ->5 _{arβ recorc} j _{5 D} .f actual flows approaching the special construction of the invention. There is no evidence of turbulence loss or reversed mean flow. Instead, the streamlines in the photos clearly show the peripheral component of velocity being wound into a streamwise vortex. This vortex iε seen to persist far down into the channel. Thus, the energy which would ordinarily be lost; to hydraulic shock is now convected downstream in an orderly fashion from whence it can be recovered. This recovery has been demonstrated expe imental ly, as described in section II.4.

This data clearly indicates that the useful range of the

t^ee l _{ement macnine can alsQ bβ β)(tended} ^ . ^ t ^h e ^p eak or _{de5ign pQint e Iciβncy} . _{% maintalnβ£ <}

SUBSTITUTESHEET

V-2 Reduction of Flow Losses

This section considers the effect of the use of streamwise vortices on the flow loss coefficient C_» . Since the shock losses have been completely eliminated, there must be an elimination of the slow moving ^' particles of turbulent fluid which block the channel. This eliminates the chol ng effect on the through-flow thus reducing its s in friction.

A further reduption in the choking of the through-flow follows from the fact that the thickness and solidity of the streamwise vortex generating structure required for off-design performance can be reduced over that of conventional blade design. Due to the stability of the vortex structure, the αn-design channel loading can be increased as explained in section III. This reduces the required solidity s^ill more. Thus, a further reduction in the value of C results.

The presence of the streamwise vortices will tend to suppress turbulence throughout the entire hydraulic circuit. This effect is difficult to assess, but if properly exploited it is bound to have a beneficial result in the lowering of C -

Suppose that all of the above effects result in a reduction of the value of Cj- by SO.. The importance of this can be seen by an inspection of Fig. 5.2.1. Here the data of Walker from Fig. 19 has been replotted with C _< * ^« - 0 and r. - ^{» •} C _JI . The 75'/. efficiency point now occurs at a speed ratio of 0.1, Thus, a torque converter with a single internal gear set

is competitive with the best automotive transmissions using multiple external gear sets. In view of the extremely high losses represented by current values of Cc , this estimate of a 50'/. reduction in Its value does not seem to be an unreasonable goal.

The above discussion does not take into account the more exotic techniques of air filled vortex cores or 1aminan2 tion of portions of the boundary layer.. The practical application of these ideas would eliminate the need for any gears at all in automotive applications.

SUBSTITUTESHEET

Figure 58 shows a torque converter configured to ex¬ ploit a Streamwise Vortex containing an inner core of a fluid lower in density than the main working fluid. A pre¬ ferred working fluid would be water treated with a suitable antifreeze, lubricant, and an anti-rust additive. The pre¬ ferred second fluid is air.

The torque converter comprises input shaft 1 and an output shaft 2 , both of which rotate in the clockwise direc¬ tion as seen from the left. Casing 200 mounted on shaft 1 carries the impeller 88. Casing 202 mounted on shaft 2 carries the turbine 89. Stator 90 is mounted to fixed abut¬ ment 204 which does not rotate. The impeller casing 200 is provided with a rotary seal 206 at abutment 204. The impel¬ ler casing 208 has a similar rotary seal, not shown, at the intersection of the turbine casing 202. The inner seal of the impeller turbine and stator stages and the ends of the interaction regions therebetween are conventional and are not shown.

To facilitate the explanation of the machine shown in Figure 58, a circle is drawn at the mean annular flow path 69, where the helical path 69 has been rotationally pro¬ jected onto the plane as shown. Radial sections normal to 69 are taken at eight positions around the mean annular flow path 69 and are denoted by 70-71;72-73; ...84-85. The view of each section is in the direction of the mean flow 69, that is counterclockwise as seen in Figure 58. To illus¬ trate the flow at right angles to these sections, a standard polyconical approximation is used. In the case of section 70-71 which is shown in Figure 59, the polyconical approxi¬ mation is a planar cross section.

A cylindrical cross section at cylinder 60-60 of Figure 59 is then taken and is shown in Figure 60. Thus, this region approximates that of an axial machine.

The polyconical approximation for the flow in the re¬ gion of section 72-73 shown in Figure 61 is obtained by projecting a line 86 (see Figure 58) perpendicular to 72-73 and tangent to the mean annular flow path 69 so that it intersects the axis of rotation of the machine at 87. This cone is then developed into the plane of Figures 62 and 65.

The outline of the flow passages of the machine intersected by the cone are shown in Figures 62 and 63.

The section 74-75 is shown in Figure 64 where the sec¬ tion is unrolled and developed on a plane. The configura¬ tion normal to this section is approximated by the annular disk shown in Figure 65. The flow in these two figures approximates that in a radial turbomachine.

At the outside of the hydraulic circuit 69 (Figure 58) , the flow is again approximately axial, and a normal section 78-79 is shown in Figure 65. The configuration perpendicu¬ lar to this section is shown in Figure 68.

Section 82-83 is developed in a manner similar to sec¬ tion 74-75. Figure 69 is the developed section 82-83, while Figure 70 approximates the configuration normal to this section by an annular disk.

Section 84-85 is developed in a manner similar to sec¬ tion 72-73. Section 84-85 is shown in Figure 71 developed on a plane, while the configuration normal to this is illus¬ trated in Figures 72 and 73.

Returning now to Figure 58, the machine is generally configured as in Figure 1 with an impeller 88, a turbine 89 and a stator 90. The mean annular flow is in direction 91. The machine has extensive overlap. The stator 90 output 92 is overlapped by impeller input 93, the impeller output 94 is overlapped by turbine input 95; and the turbine output 96 is overlapped by stator input 97. During normal torque multiplying operation, the machine rotates clockwise when viewed from the left in Figure 58. The Streamwise Vortices utilized in this example all rotate in the counterclockwise sense when facing along the flow direction 91.

It is convenient to explain the operation of the ma¬ chine by discussing the flow as it enters the impeller 88 at a typical input 93. See section 72-73, Figure 61, and its associated normal development in Figure 62. At section 72-73 (Figure 61) , the flow consists of a set of twelve Streamwise Vortices 98 with air cores 99. Each Streamwise Vortex proceeds into a passageway 100 which both curves and contracts. The changes of curvature is in the direction

shown so that the correct torque reaction is developed on the stator 90. The passageways contract as working fluid passes out of these passages 100 along edge 92 into the impeller 88 input 93. The flow issues from slots 101 in the stator 90 as indicated by arrow 102 in Figure 61. The im¬ peller 88 has cusp-shaped slots 103 fashioned in section 72-73 to receive flow 102. The flow impacts the cusps as shown so that the torque generated by the impact is in the same direction as the input torque thereby causing torque multiplication. Since the impeller 88 is in relative motion in the clockwise direction, as shown at 104, the impacting flow 102 is spread out. Thus, the point of impact 105 will move along the cusp 103. The flow divides at this free diffusion point. Flow proceeding to the right will move along the cusp 103 and join flow issuing from the next stat¬ or passage slot 106, while flow proceeding to the left will accumulate into a clockwise flowing Streamwise Vortex 107 with an air core.

Figure 63 shows the development of the impeller cusp from its beginning 108 through its section 72-73 to a point just before it becomes the circular tube shown in Figures 64 and 65. The cusp develops in a compound curve 109 which promotes the generation of the Streamwise Vortex in the mean direction 91. The sectional area of the cusp increases in direction 91 so that the fluid from the stator slot 101 is accumulated properly. In general, it is advisable to avoid sharp pressure gradients in the direction 91 so as to pro¬ mote Streamwise Vortex stability. As the end of the input region 93 is approached, the cusp 103 is provided with an edge 110 which catches the Streamwise Vortex with core 107.

As shown in Figure 62, the tubes 210 of the stator taper down in the direction of the flow 91 into the tapered cusp-slots 101 and 106 (see also Figure 61) which unroll the vortices 98 into a sheet-like flow 102 (Figure 61) . Note that edge 212 is located inwardly of the outlet edge 92 of the stator 90 to aid in this unrolling process.

The flow through section 74-75 (Figures 64 and 65) now consists of a set of independent Streamwise Vortices each of

Since the flow in the turbine 89 must proceed inward and maintain an air core as much as possible, low aspect ratio nydrofoils 116 are provided (Figures 58 and 70) . These foils or blades 116 are driver, tne impeller 8£ anc are set at a very large relative angle of attacx so aε re constitute an inward flowing pump. The blades are highly swept tc avoid loss and may superventilate at least fcr a part of the operating cycle. These blades are also useful in starting the flow.

Although not shown, blades similar to 116 may oe in¬ cluded m the stator near section 70-71. Sucn blades arc also driven DV tne impeller and help to start tne macr.me . They may penetrate the air core 99 and be twistec sc n t they act on both sides of Streamwise Vortex 88. Thus, tne-_ push the vortexes along through the passages. This helps to start the machine and prevents stalling.

The turbine 89 to stator 90 interaction is illustrated in Figures 71, 72, and 73. The behavior in this overlap region is similar to that discussed oefore. All cf the Streamwise Vortices are rotating counterclockwise as viewed in the direction of flow, that is into the page. A Stream- wise Vortex carried by the turbine 89 exits aε shown arrow 118 and impacts the stator 90 cusp structure 120 at 121. Again, accumulation is necessary since tne turbine 8 is in relative motion with respect to the stator 90 in tne same direction as the impeller as shown by arrow 104.

Figures 72 and 73 show a cone approximating a plane view of this interaction where both the turbine 89 and tne stator 90 are developed as before. The passageways 121 c: the turiDine 89 curve r>acκward in the opposite d rectio: or 104 as snown. They also contract aε fluid is being trans¬ ferred from the turbine Streamwise Vorte:. tc tne state: tnrouσn slots 123. The stator approximation cf a t_ane ie' s aisc snowr. in Figure 72. The cusps 124 of tne stater 9' dc net curve appreciably, aε the flow from tne turoir.e is reactmc on tne stater thereby producing a torque : t e c ocΞ i d rect cr. tr tnat cf tnf t ni g.

SUBSTITUTESHEET

-63- which has been formed in a manner similar to vortex 107. These Streamwise Vortexes are energized by the centrifugal pumping action of the impeller 88 thereby absorbing input shaft power as desired. These Streamwise Vortexes are con¬ ducted to the interaction region between the impeller output edges 94 and the turbine input edges 95 (Figure 58) .

The impeller turbine interaction is shown in Figures 66 and 68. The impeller 88 is fitted with passageways 111 with slots 112 which resemble those of the stator output. These passageways curve in the forward direction, that is in the direction 104 of the impeller tangential motion so that power continues to flow from the input shaft into the fluid. The passageways 111 contract in cross section at the outlet slots 112 so that sheet flow comes out of the slots 112. Passageways 111 appear elliptical in Figure 66 since the section cuts them at an angle. In fact, these sections approach a circular form when cut normal to the Streamwise Vortex mean flow.

The turbine 89 is fitted with cusps 113 which are simi¬ lar to cusps 103 (of Figure 61) . The impact of the flow 115 loads the turbine so as to transmit torque to it in the same direction as the impeller motion which is desired. Since there is a relative motion of the impeller 88 past the tur¬ bine 89 in the direction 104, the turbine cusps 113 are elongated in the tangential direction to allow accumulation of fluid as before. The turbine passageways 114 curve as shown so that the deflection of the flow transfers torque in the proper direction (see Figure 68) . Inlet slots 114 in¬ crease in cross section in compound curves in a manner simi¬ lar to that described with respect to input slots 109 of Figure 63.

The flow proceeds into passageways of the turbine which are circular in cross section containing discrete Streamwise Vortices as shown in Figures 69 and 70. A typical Stream¬ wise Vortex with air core is shown at 115. This section behaves in a manner similar to the process described with respect to 74-75 (Figures 64-65) except that now the flow is inward toward the machine axis section. Thus, the machine is operating here as a radial flow turbine.

Further αepiction cf the stater 90 is showr ir. Figures

59 and 60. The hydraulic circuit s now complete anc tne flow consists of a set cf discrete Streamwise Vortices at section 70-71 of Figure 59 as discussed previous! . Figure

60 shows that the stator has started to curve m tne direc¬ tion of 104 to prepare the flow for entry into the impeller 88.

A modification to the structure of tne impeller 8£ cf Figure 61 is shown in the corresponding view of Figure 74. The cusp 103 is modified slightly and auxiliary vanes 125 and 126 are added. These vanes are shown m section and are supported by struts so that the flov: car proceed cr cot: sides of each vane as indicated oy arrows 12 ^" ar.c 12c. These vanes aid the cusp m receiving and accumulating tne flow from the stator 90.

Figure 75 shows another modification to the lmpeiiei 8c of Figure 61. Instead of auxiliary vanes 125 and 126, aux¬ iliary cusp members 129 and 130 are used. These cusps es¬ tablish additional Streamwise Vortices which may or may not have air cores that will eventually amalgamate with the principal air core of 107. In any event, the cusps 131 and 132 perform the same overall purpose as vanes 125 anc 126 a? can be seen by arrows 133 and 134. The flow is deflected and accumulated so as to aid transfer of torque tc tne nr- peller.

Figure 76 illustrates the amalgamation of the vorticit- generated by the auxiliary devices . Instead of a more cr less flat surface existing on impeller 88 between cusps at the entrance region, see Figure 62 near 91, the entrance flow now interacts with tne auxiliary devices. As tne _c-. progress in direction 91, these auxiliary channels ceflec tne flov: ir a directior opposite tc directicr 104. Ir.:. transfers additional torque tc impeller 5c. Tne αeflectc. flow is added tc the main cusp flow 1C ^~ εc that amalca atic- cf the Streamwise Vortex is assurec.

Tne concept cf auxiliary vanes cr cusps ciscussi. : ccr.nect cr. w t . Fiσures ^"" _" -76 car D- acclied t^ ιl_ ^~ " tr cascade interactions cf tne Stre m-.;!;

SUBSTITUTESHEET

Vorte : discussed above. The essential principles are the same.

It is well known that the capacity of a torque convert¬ er cr fluid coupling can be varied by varying the percentage of working fluid and air. Although this method has had wiαe spread use with fluid couplings, its use in torque conver¬ ters is limited to situations wherein the machine is either full or empty. In the present invention however, it is possible to continuously vary the capacity of the machine within a limited but useful range. This is made possible by the ability of the Streamwise Vortices to tolerate air cores and also by the very strong anti-foaming action associated with their rotary flows.

Admitting, more or less, working fluid to the machine relative to air described above, will vary its capacity tc absorb torque. When used in conjunction with an automotive engine, thiε will vary the effective gear ratio of the transmission. This action can be extended further as shown in Figure 77. This figure is a modification of Figure 61 with impeller 88 and cusp-like slot 103. In Figure 77, another set of slots, one of which is 135, have been added to the original set. The new set is added in a staggered fashion along the leading edge of the impeller 88 so that the new cusp 135 is further down stream in direction 91 tnar. is 103. Thus, as fluid iε added to increase the capacity cf the machine, the set containing 103 will fill first. The set containing 135 fills next and the range of useful con¬ trol of the machine's capacity is thereby extended. Several staggered sets of slots may be employed.

An important function cf a conventional torque convert¬ er is to absorb vibrations which might arise anywhere ir. t -; drive trail.. This results from the inherent damping capa¬ bilities cf the working fluid even though the conve ticna. achine itself generates significant vibrational impulse; from the "shock" flows at the blade leading edges. In the instant invention, the ability to absorb vibraticr. car. r>-:- greatiy enhanced by utilizing means which centre1 the dis¬ tribution cf working fluid versus the air distribution i:

portions of the hydraulic circuit thereby controlling unbal¬ ance within the machine.

An active means tc accomplish vibration absorption ιε snown in Figure 78. This figure shows the impeller 88 as ir. Figure 77, except that both sets of cusps 103 and 135 have beer, fitted with auxiliary devices 129 and 130 and 136 and 137. A further modification is the inclusion of setε cf air tubes 138 and 139' . These air tubes communicate with tne start of the air core of the corresponding Streamwise Vortex as shown. Auxiliary cusp 129 is now designed so that its deflected flow 133 joins the Streamwise Vortex in 103 and closes the air core 141 m this region. Thus, tne size cf this core and hence the amount of flow this Streamwise Vor¬ tex can accept, is controlled by the air pressure m tuce 138.

If the pressure is low, the Streamwise Vortex within 103 accepts more fluid. On the other hand, a higher air pressure forces some of this working fluid to move toward cusp 135. Since the working fluid is much denser than the secondary fluid (air) , a redistribution of mass occurs thereby allowing for a means to control vibration.

Tube sets 138 and 138' communicate through a suitable rotary seal with control valves 139 and 139' which control the air source 140 in response to a signal from the computer 144. This computer receives a signal from a set of vibra¬ tion sensors shown schematically at 142. The computer is programmed to establish a feedback system. When a particu¬ lar vibration is sensed, the computer causes a correspondmc redistribution of mass so that the vibration is actively suppressed.

I have net shown the cross section 76- ^" cf Figure 5_ since this would essentially, merely show tuυes only, εi i- iar to the cross section at 74-75 shown m Figure 64. Simi¬ larly, I have not shown cross section 80-81 as this wc !_ only show tubes similar to that shown in cross section 82-52 (Figure 69) . Those skilled in the art will understand tnat the tubes cf each cf the sections 85,59 and 9C spiral _cr' tne mean flow path 69 following tne same flov patter: ; i

SUBSTITUTESHEET

a conventional machine. That is, the vortices ^" in my ma¬ chines, as shown in all of the figures herein, follow the same flow path aε the desired laminar flow path in a conven¬ tional torque converter. The tubes are circular in cross section, except at the inlet and outlet ends thereof, so as to guide the vortices with minimum loss. The tubes taper towards the slots in the inlet and outlet ends of each stage in order to capture the flow and form the vortices i the input slots and to form the sheet-like flow from the outlet lots. All of the inlets and outlets are cusp-shaped tc aid in this process.

It will thus be seen that I either provide vorte:: flow between stages aε in Figures 1 through 57 or unrolled sheet flow between stages in Figures 58 through 78 with vortex flov/ within each stage. Machines can be designed according to my invention wherein there is constant vortex flov: in the same direction as in Figures 58 through 78 and vortex flow between stages rather than the unrolled sheet-like flow of Figures 58 through 78.

An automatic transmission according to my invention is shown in Figure 79. Here a torque converter, according to my invention, employing vortex flow in the stages, iε shown at 220. The torque converter is filled with two fluids of differing densities, preferably air and water, as previously described. The torque converter is connected between a power source 222 and a power user 224 which are connected to the torque converter 220 by input shaft 226 and output shaft 228. A computer 230 is provided having inputs from rota¬ tional sensors 232 and 234 sensing the rotations of the input shaft 226 and output shaft 228, respectively. The computer 230 is also responsive to the throttle 236 and controls flow cf the working fluids between a reservoir 235 and the torque converter 220. Under control of the computer 23C, the reservoir 238 injects or extracts working fluid from the torque converter 220 through pipes 240 and 242 tc control the ratio of high density to low density fluid therein and thus control the effective σear ratic cf th torcue converter 22C.

among those made apparent from the preceding description are efficiently attained and since certain changes may ce made m the invention without departing from the scope thereof, it is intended that all matter contained in the acove αe- scription snail be interpreted as illustrative and not m a limiting sense.

Having described my invention what I claim as new and desire to secure by Letters Patent is:

SUBSTITUTESHEET

REFERENCEΞ

(1) Q. E.. BaljB, "Turbo-Machines," John Wiley and Sons, H_.H York, NY, 19Θ1.

(2) F. D. Ringleb, "Separation Control by Trapped Vortices," "BOUNDARY LAYER AND FLOW CONTROL," Perga on Press Inc., New York, NY, 19όl.

(3) Ξ. Saunders, US Patent 3,241,376, I960.

(4) A. Rαshko, "Interaction Effects on the Drag of Bluff Bodies in Tandem," "AERODYNAMIC DRAG MECHANISMS," Plenum Press, New York, NY, 197B.

(Ξ) W. Merslcirch, "Flow Visuώ.1ration, " Academic Press, Inc., Naw York, NY, 1974j

(6) P. Singh, "Physics of Fluids," 19ι 1BΞB ⁽ 1976 ⁾

(7> J. Johnston, "Internal Flows,"

"TOPICS IN APPLIED TURDULENCE, " ΞprInQβr-Vβr1 g, New York, NY

1976.

(B) M. Hall, "Vorteu Breakdown,"

"ANNUAL REVIEWS OF FLUID MECHANICS," Annual Reviews Inc., Palo

Alto, CA, 1972.

(9) V. Jandasek, "Design of Single Stage Three Element Torque Converter," SAE Publication AE-5, 1973.

£10) F. Walker, -"Multiturbine Torque Converters," SAE PuDlicatiαn AE-S, 1973.

(11) J.- Polak, "An Automatic Transmission for Automotive Gas TurbiRΘ Power Plants," SAE Publication ΞDP-Sυ/4άΞ, 1900.

(12) -H. Paarcey, "Shock InJduced Separation and Its Prevention," "BOUNDARY LAYER AND FLOW CONTROL," Pergamon Press Inc., New York, NY, 1961.