IE7

Conceptual Development of Fluttering Oscillating Reciprocating Waving Flapping Blade Fin Foil Aerofoil Airfoil Hydrofoil Wing Vane Windmill Wind Tidal Fluid Kinetic Energy Convertor Flutter Engine Power

Introduction/Background/Prior Art: There are at least four conceptual origins of ideas for oscillating wind energy conversion:

1.The vaguest stems from study of Nature where the ease of bird flight and the speed and leaping power of fish suggest a remarkable efficiency to oscillating propulsion, indeed borne out by metabolic measurements. However living organisms have an infinite capacity for active control so attempting to copy their form in a machine can be disastrous. Man's artificial wings had to be inherently stable to control and as their speed was increased, very rigid as well. Otherwise they were soon found to develop unwanted and dangerous oscillations by themselves.

2. Such dynamic flutter of tailheavy control surfaces on flexible air-surfaces was recognised as a spontaneous oscillating conversion of airstream kinetic energy into vibration energy, but the flutter wind engine concept (Duncan) was entirely academic to aeroelasticians charged with preventing aircraft flutter in its many possible modes.

3. Drawbacks to standard horizontal axis rotary windmills provide another two, but more specific, sources of inspiration. Firstly their motion converters (crankcases, gearboxes, and generators) must be mounted on substantial towers for their rotors to be in roughly uniform wind and to clear the ground which is where we want the power and would prefer to do the installation and maintenance. Semirotary oscillation of a wind surface from upright about a low pivot would seem much more convenient. Especially as the increase in blade velocity with height would roughly match the natural increase of the wind with height for near constant angle of attack along the blade. (This is truest for a laminar flow, favoured by low velocity, smooth terrain, and lack of convection.) These strong practical kinematic reasons for rotation about a ground level pivot and thence for oscillation have been wantonly ignored by (wind) tunnel vision academics whose lab airfoils heave vertically through pathetically small distances with inevitably miniscule power for the size=cost of the airfoil, of the impractical linear guides and of the tower to get to uniform and appreciable wind outside.

4.Finally rotary windmills cranking deep well reciprocating pumps are a very bad combination. Fundamentally the crank connection of a wind rotor to the simplest form of borehole pump is a severe mismatch of speed and torque characteristics that results for a typical wind regime in a net efficiency of only 1/6 of the theoretical limit for an ideal load (angularly constant back torque varying as speed squared) (Dixon) . Perversely the pump torque is angularly varying but constant with speed. The crank backtorque peaks in the middle of the upstroke and is nil on the entire downstroke, so the rotor gets STALLED at the peak in a pulsing light wind. Even more non-ideally speed doesn't increase the torque due to the FIXED STROKE of the crank. So in high winds the rotor turns faster than ideal. My inspiration was that an oscillating prime motion driven by the wind might better match such an oscillating load, which turned out to be very true.

 A simple crank also has an impractically large side swing, requiring too big a yaw bearing. So the side motion has to be reduced by vertical sliding mechanisms running in oil, or large linkages with distant pivots and by keeping the crank radius and stroke small. This small net vertical motion requires a large diameter cylinder to achieve enough stroke volume, and rigid stiff pump rod and droppipe to transmit the motion to the cylinder. Nonetheless the rods are hard to make stiff enough to push the pump piston down faster than all can fall under gravity, so the conventional pump has a serious speed limitation. For human-powering this prevents the use of the strongest muscle group in the body, the legs whose natural walking rhythm is too high. Only weaker armpowering synchronous with the breathing rhythm is slow enough. For small rotary windmills or pedal pumps which have high rpms, this requires reducing the speed with either gears or chain drives, adding considerably to the cost.

The rigid rods and droppipe make installation awkward. Routine maintenance is further complicated by the retention of the water column by the footvalve at the base of the cylinder, which must be very watertight for intermittent pumping with the inherently variable wind. So 'Closed-top' cylinders with a reduced diameter dropipe must be lifted with the weight of the water column too, and then each length of rod and pipe unscrewed as they emerge from the well. Losing the remaining strings down the well is easy and hard and sometimes impossible to recover from, and the well must then be abandonned. With Open-top cylinders, only the pump rod string must be hoisted and unscrewed to maintain the piston. There are a few OpenTop designs with the footvalve in a conical seat which can be extracted by either screwing on or hooking on the piston lowered down to screw onto it or connected to the piston with a lost motion link. However extracting this footvalve against the full weight of the water column requires very hard even hammering pulls. The release of the water column and recovery of the footvalve makes pulling the dropipe easier and less necessary.

The design of an oscillating windmill led to a new approach to the wellpump also that overcame these practical problems of existing reciprocating well pumps.

 Design requirements for oscillating windmills vs prior art

There are many properties that any windmill must have, that have not been met by any previous oscillating design. A windmill must self-start as the wind arises but self-protect when it gets too strong, and must accept the wind from any direction. If the base yaws to follow the wind, continuous yaw rotation as the wind rotates around the compass several times, must not interfere with the power output to terra firma ( though wind turbines often have limited wrap powr cables). A practical use for a windmill must allow easy and low loss storage of the output to absorb its continual fluctuations with the wind and interruptions with calm to be able to meet a requirement on demand. Note that conditions are much less demanding of a run of tide mill and do not seriously apply for a run-of-river mill.

An oscillating windmill must withstand the extra inertial loads of continual acceleration and deceleration. Roughly the ratio of inertial bending moments to the total permissible scales as MATERIAL density( flowspeed squared+/- gL) /fatigue limit stress showing an inherent limit on the DESIGN (optimum) flowspeed irrespective of the FLUID density and the machine size. The stream pendulum configuration largely escapes this constraint because the two body forces nearly cancel everywhere. Otherwise this is one significance of the weight specific fatigue strength having units of speed squared, and note that wood easily beats steel and aluminum in this parameter.

The mill's load connection must absorb highly variable powers because the wind's energy flux varies very strongly with windspeed, yet still allow the oscillation to start (and yaw) easily. Less so for a tidal mill and even less for a river mill.  Previous lab models involve mechanisms to kinematically produce a linear (DeLaurier, Hien, Clayton; and Jones, Platzer & Davids) or semirotary (Bade, Kentfield) oscillating motion of a wind surface for one wind direction. They ignore that any such mechanism will ensure destruction in high winds. For with a prescribed amplitude and so stroke the speed will increase with the wind as much as for the rotary machine but here the inertial reactions will eventually surely overwhelm any mechanism, no matter how ridiculously expensive and inefficient. These are at best concepts for run of river mills where the flow stream is altogether SLOWER, unidirectional and very steady, but all with bearings submerged and very low linear to low semirotary (cross wind axis eg Trapp)swept to blade area.

Payne's unisail oscillating vane ideas included a semirotary resonant oscillation of a wing about a ground level pivot so that the inertial reactions are balanced elastically in a dynamic oscillation of a certain natural frequency but he still imagined a mechanism to twist the wing. How the varying power with windspeed would be absorbed and ultimately limited was simply ignored.

In fact with a dynamic oscillation Bielawa observed that the oscillating vane amplitude can be variable, but failed to exploit the fact. This variability is actually essential for efficiently converting the highly changeable power coming from the wind, at least for the constant force pump load . Then a sufficiently non-linear conversion of amplitude to pump stroke can realise a highly VARIABLE STROKE that 'endstops' the oscillation absorbing whatever power it is generating . The one way (single-acting) pump load non-linearly opposing displacement CANNOT STALL the oscillation if the wind dies abruptly, because then the spring is unopposed in returning the oscillator back to the undamped center. (Note that this makes the pumping frequency twice the oscillation frequency; a stroke and return for each excursion away from equilibruim and that the pump not only non-linearly damps the motion, it non-linearly stiffens it.) So indeed the stalling and fixed stroke problems of the rotary windpump CAN be overcome in principle by a dynamically oscillating wind surface.

In air compressing the ramping of airpressure during the prelimary compressing up to outlet pressure adds a load variation, so the rotary windpump has an even lower mean/peak torque ratio for single acting but twin cylinders to increase it only cost more. Equally the above pump concept suffers from small amplitudes being unable to compress the air to outlet pressure so it then just reexpands on return to vertical acting as an imperfect (and indeed unwanted spring). Also the air inlet is set by the return of the previous swing whereas the power available to compress it is set by the current next swing. A better approach would be for the variation to be in the amount of ambient air admitted into the cylinder and to exhaust all of that, rather than leave a variable amount of a constant charge of air in the cylinder at the end of the compression for it to imperfectly re-expand. Thus the cylinder could be bottomed out at the neutral vertical position driven by a falling weight which is lifted (non-linearly) on the away strokes of the pendulum. Naively since both works are linear with stroke, the weight's acceleration when the inlet valve has just closed and cylinder pressure is low can be balanced by its deceleration when the pressure rises above its equilibruim value to the outlet pressure which would need to be closely regulated. But the windmill's return limits the acceleration at small strokes meaning an unsafe condition at the disproportionate large strokes, when the windmill can return faster.

The only real solution is to convert the oscillating windmill's variable stroke into a variable number of fixed aircompressing strokes,safely averaging out the ramp with speed and the pendulum's inertia. The final incomplete stroke will re-expand but this truncation loss can be made negligible with a high enough conversion ratio. For instance a small motorcycle piston with built in gearbox and replaced head can be stroked many times by the windmill stroke racheting the spocket slowly.

The rotary windmill is more robust against loss of load as it just overspeeds until the apparent wind is so high there is nonet torque though there there will still be downwind load on the rotor as a whole. The oscillating windmill increases its amplitude for higher apparent wind too but unless entire revolutions are permitted not sufficently to zero the torque. This is impractical for any restoring force and defeats the advantage of proximity of the pivot to the ground. Thus an alternating tank water-compressing-air concept is not practical due to its spreading the pressure ramp over multiple cycles. Instead the pressure ramp is gone over many times within a cycle in the geared solution. And this ramp caused the fatal problems in the 1:1 counterweight compressing scheme.

 A broad division of windmills (eg of rotary vertical axis) can be made between crosswind lift and downwind (differential) drag devices, as to the motion and corresponding component of aerodynamic force which generates the power. Lift is generally much superior on account of the amplification of the magnitude of the true wind by the vector addition of crosswind motion, allowing very high power from high speed x high torque from the enhanced wind pressure. The speed ratio of crosswind movement to true windspeed has an optimum increasing with the lift to drag ratio of the blade.

Unfortunately obvious sources of high oscillating lift have high drags as well. For instance Payne suggested using a circular cylinder because the steady separated flow with a vortex on each side of the rear is unstable to quickly and alternately shedding these vortices generating a large oscillating lift. But this von Karman vortex shedding is so rapid relative to the windspeed and diameter that the 0-pk amplitude is resticted to say 2 diameters to keep the crosswind speed to roughly less than 3 times the windspeed when the net crosswind force from lift AND drag turns negative. This small swept area severely restricts the possible power (by rough actuator theory) as well as making the pump stroke awkwardly small.

On wider cross-sections with distinct corners between upstream & downstream faces, the flow separates at the corner binding a more stable vortex immediately downstream. If the wind shifts a bit to one side it separates less at the corner there and more at the opposite corner so producing a net lift towards the near side, and if that produces motion to that side, then the apparent wind will shift even more and so on. (Den Hartog ) This sort of drag into lift sideways divergence can amplify any natural oscillation, for instance generating the wild rolling of sailboats running before a strong wind. (Marchaj).

But the high drag inherent in these vortex-shifting flows limits the benefit from fast crosswind motion which enhances the power most for the highest lift/drag ratios. Even if the surface is a broadside circular arc airfoil like the running sail (Lawson), it is very hard to wash away the established stalled flow (ie both rear vortices) and its high drag with high enough crosswind speed for long enough. Also the high drag, at rest, can become too great to resist in storm winds, so it as well as the oscillating lift have to be avoided somehow in high winds to make a practical oscillating windmill.

Classic binary aircraft flutter shows a way to generate high oscillating lift without high drag, by using the highest lift to drag cross-section, an airfoil, and letting it twist as a second degree-of-freedom. Then with even slight tailheaviness this foil is unstable to a combined oscillation of foil crosswind motion and twisting above a certain 'critical' windspeed. The decisive virtue of this flutter phenomenon as the basis of an oscillating windmill was not recognised by Bielawa or indeed previously in aeroelasticity: that with not too much tailheaviness there may be another higher critical windspeed at which this binary system becomes stable again. This means no oscillation and no lift and only very slight form drag in storm strength winds, ie INHERENT HIGH WIND PROTECTION. I discovered this by conjecture and experiment , and then confirmed it analytically. In fact the flutter twist amplitude decreases continuously with windspeed curtailling the bending monents in strong winds. These stress-limiting features are the decisive advantage of the flutter variable (roll and twist) amplitude approach.

Machine development

The fluttering windmill concept of a wing twisting on a nominally upright axle that can oscillate about a horizontal wind axis near the ground as motivated above had to be translated into practical embodiments which accept winds from all direction and non-linearly connect the oscillator to the pump submerged in the ground or surface water to maintain prime with the intermittency of the wind, and maximize the wind energy capture for the structure needed. The latter immediately means that the configuration should allow as large an oscillation of the wind surface as possible for more power. In linear flutter theory the power capture grows as amplitude squared with an eventual actuator theory limit of power proportional to amplitude. Thus the semirotation possible from vertical should be high, limited only short of contact on the ground.

Bielawa used a flexible composite support for roll stiffness for his wind tunnel model limiting the maximum roll deflection to about 40 deg . In twist cantilever bearings were used but also a torsional restraint as natural in aircraft aeroelasticity. This unnecessary twist stiffness would be highly stressed by the tailwind onslaught of a new breeze and for large amplitude oscillation tailwinding of the wing when its twist transiently exceeds 90 as the wind lulls causing a snap 270 rotation if free. So it is simpler and more practical to have the wing completely free in twist, and incidentally tap only the semi-rotary roll motion relative to the ground for pumping. Note that the decrease in twist/roll amplitude ratio with windspeed means loading twist however non-linearly cannot contain the roll amplitude. Whereas loading roll non-linearly can contain both in normal winds.

To allow a very small perturbation to static equilibruim to develop to a decent amplitude and so power, it is essential that there be effectively no pump stroke=braking back torque at small amplitude. But it must develop continuously at intermediate amplitude because in light winds the speed ratio grows quickly with amplitude and can become excessive so the power must start to decrease due to drag if there is still no load to remove it. Naively to realize the optimum amplitude which varies linearly with windspeed to keep the speed ratio constant, and the optimum power increasing as windspeed cubed, the stroke should vary as (intermediate) amplitude cubed. This should allow self-starting as the very small amplitude power at the linear stability cutin windspeed varies as amplitude squared. For high winds, the optimum amplitude is simply in the ground so the pump stroke must climb even more quickly with large amplitude to hold it (just) below the interference limit. The maximum stroke must exceed the maximum power at this limit vs windspeed which is fortunately limited by the high wind return of the wing to dynamical stability..

 Standard ways of achieving a particular output function are non-circular gearing (eg elliptic) and cams. Here the load is single-acting, suiting a cam and the simple Spanish coil tension convertor invented.

The output motion needs to be substantially vertical suggesting a crank to a linear track which if it crosses center is near cubic for semirotary oscillation of at least +/- 90. The most practical layout is with the crank at the top of the pendulum just beneath the driving wing but this makes the away motion downward which is only good for pressure pumping water at or near the surface or possibly inertia pumping from greater depth near the upper limit of design windspeed. This design requires a stiff low friction sliding bearing which could be provided by a very high quality stainless steel cylinder. Possibly a long swing arm could be used if a yawing support for it can be provided.

 Floating Configuration Pond Pumping Fig's 1 & 2

 The variabilty of the wind direction introduces a third degree of freedom to the basic binary flutter modes of crosswind roll and wing twist. The obvious method is yaw of the roll horizontal axis under a tailvane and this is easiest for a floating unit which could pressure pump the water. (The floating can also easily provide low-loss roll motion with an internally ballasted circular cylinder as hull . However this restricts the roll amplitude to well less than 90, and so the power yield from a given size wing.) The special requirements for the powertakeoff are yaw freedom and non-linearity. Fortunately these can be separated into two distinct components, the first being an ouput pipe swivel incorporated in the anchor line. Actually the mass-machined pipe union fitting just needs a setscrew to lock it into just the right tightness for this . The anchor self adjusts well to the changes in the water depth, and allows easily moving the entire unit and shore pipes.

Then one simply has to devise a non-linear powertakeoff unit that can yaw with the hull but not roll with it. The high roll stiffness needed for this roll reference suggests a catamaran unit . To try the cam the first choice is to which part to assign the cam profile and which the cam follower. Here the wide catamaran base is ideal for profiled rocker arms pivoted outboard so then the roller rotates centrally on the pendulum. As the roll increases glancing contact of the roller far from the rocker pivot changes to near perpendicular contact close to the pivot for a huge increase in geometrical advantage. A virtually straight rocker arm can generate the initial cubic variation of stroke with roll with ultimately even stronger variation near maximum amplitude. The rocker arms can be united to the same pump with a short link that accounts for the variation in their end distance, but this involves moving the two arms together when only one is being actually depressed by the roller. The arm motion and friction is halved and the link friction and lost motion is eliminated if they are connected to separate pumps. The key dynamical advantage of this is that it trebles the time for gravity to return each arm and piston for less flow friction in cylinder filling and 3 times the max frequency.

With the anchor at the bow of a ballasted central hull, the stern is convenient for this pump catamaran which can then also support the tailvanes to yaw the hull into the wind direction. There is the important consideration of how to minimise yaw oscillations in sympathy with the main roll oscillation. The primary source is the net downwind component of aerodynamic force acting on the rolled wing for a first harmonic yaw torque. A torque of opposite phase can be generated by mounting the wing (and conveniently the cam roller) on an upright axle at the stern of the hull. The inertia of the elevated wing then reacts to the roll acceleration by generating a virtual yaw torque about the center of the hull. It might be thought that this stern location of the wing would help in the tracking like on a downwind rotary machine, but here the wing is not paired and completely free in twist so the separate tailvanes are defintely needed.

Alternatively a catamaran unit can support an elevated pendulum pivot between two cross trusses with the pendulum just clearing the water surface, and the swiveling pipe union suspended from the bottom of the front truss. The roller is then just above the counterweight and the rocker beams hang down with the pumprod at their outboard ends rather than counterweight. This simpler configuration has less roll friction and more roll amplitude for alot more power.

Well-pumping Land configurations

For wellpumping on land, yawing the whole machine including roll counterweight in light winds can be critical on friction without good bearings so it was worth considering another approach, two perpendicular horizontal axes instead of yaw. This spherical pendulum idea is more elegant as there is no tailvane and no need to yaw the whole base including the counterweight. On the immediate downside the maximum roll amplitude for a spherical pendulum wing is necessarily less than 90 whereas the yawing design with an elevated horizontal axis can have more, limited only by ground contact. The simplest movement that works equally for all wind directions to non-linearise into pump stroke is the increase in distance of the bottom of the spherical pendulum from the ground point directly below it at equilibruim.

Naively the pendulum would roll about the wind direction with little pitch about the crosswind axis because the binary oscillation reactions are second order, ie. mean and second harmonic in that mode, and wouln't cause resonance. In fact the immediate inspiration for the spherical pendulum was that a satisfactorily working model of the floating unit was measured to actually have a natural pitch frequency less than twice its roll. Ideally the Lissajous figure of eight combining a bit of the right phased pitch second harmonic with the roll fundamental could augment the power by moving the wing downwind slightly in the direction of the large drag component in the large apparent wind at maximum speed ratio as the wing crosses vertical at maximum twist and then recovering upwind again at the end of the roll when the twist and so the induced drag is passing through zero and the wind is less closer to the ground. Detailled calculations showed modest enough (mean and second harmonic) pitch response, but the phasing was not ideal so the predicted power decreased about 10%.

But with the pitch frequency so decreased to equal the roll, another instability shattered all assumptions as soon as a proper pump load was applied at full scale. The center point power takeoff negatively damps a 90 out-of phase fundamental pitch oscillation which grows towards full coning which it doesn't damp at all. Twin pumps to separately damp roll about perpendicular horizontal axes can't give isotropy and non-linearity at the same time, and can't be installed down the same well.

The wing twist responds to the downwind acceleration by increasing on the downwind moving side and decreasing on the upwind moving side. This creates a differential drag that fuels or at least doesn't damp the downwind first harmonic. Usually this peak twist will grow to exceed 90 and the wing eventually flips tail first through the wind on the downwind side which kills most of the oscillation which then rebuilds over several cycles until the same flip recurs and so on.

Because of the inability of a pump connection to tap/damp this coning ROTATION, the spherical pendulum configuration is fundamentally flawed. The best that could be done was to use an offcenter pendulum frame horizontally hinged to an intermediate frame perpendicularly hinged to the ridge of a giant 'sawhorse' frame and copy the motion along the ridge beam with a parallel frame hinged to the pendulum and then orient the copy arm with a small tailvane and only allow it to rotate about this orientation by locking the copy arm yaw at about 10 of roll. This did allow rolls very close to +/- 90 for all wind directions and did avoid having to yaw the heavy pendulum into the wind.

A key component was invented for the non-linearisation of the pendulum rope pull on these machines. From a pulley at the ground center the thick rope was lead to a double drum winch. There the rope was 'Spanish' coiled on itself between two deep sideplates set a rope diameter apart so an Arcimedean-type spiral was formed. The second drum on the same shaft had a similiar coil of pump belt but of opposite hand. As the input coil is unwound and the output coil (and pump plunger) wound up, the torque arms change continuously from high mechanical advantage at first to very low as the input coil becomes completely unwound down to directly pulling on the steel winch shaft whose diameter is comparable with that of the rope. This achieves a very non-linear conversion of roll amplitude to pump stroke with tremendous final amplification at the maximum permissible amplitude, which was further increased by changing to a much stiffer (and more durable) input line in the form of standard leaf chain. The net function is as a non-linear pulley between pulls that can move anywhere in a planar perpendicular to the axle and is conveniently located vertically above a well so the belt directly pulls the pump plunger via intermediate wire. The stretch of the wire is neglible compared to the enormous strokes possible before the belt coil radius exceeds the well diameter 4" or more. Also noteworthy is the tremedous amplification of the pendulum's angular displacement(0-100 ) into the winch axle's (0-1800) with the angular velocity stepup even higher. For aircompressing on the low weight sensitive floating configuration this winch pulled by the pendulum counterweight with an output chain coil ratcheting the motorcycle engine sprocket is the preferred solution with the air being stored in the steel pontoon tanks.

  On land, trying to yaw a planar pendulum with its greater potential amplitude and power was in fact no problem due to the high development and availability of truck taper roller bearings. The entire wing and pendulum to be yawed did weigh far less than a standard fanmill even with the weight for direct aircompressing, but the tailvane would have to be bigger to resist rapid yawing in response to the desired binary oscillation. Alternatively the above yaw locking on roll could be used on the whole machine.

 The machine configuration is defined firstly by the convenience of using the steel well-casing as a pile foundation for the tower and of putting the Spanish belt coil directly over the well at the top of the tower to directly pull the pump plunger. This winch should be as high as possible for the maximum delivery head without needing a seal around the pump wire. So the winch was put above the roll axis with the leaf chain pulled by the pendulum spar immediately below the wing which minimizes the spar bending moment and so its weight which also must be countered. This configuration also gives plenty of vertical distance for a cylindrical weight to slide inside the tower and push below a big piston into a long aircylinder with its bottom and outlet still at ground level.

Then the question is whether to put the pendulum (immediately) downwind or upwind of the tower. Again because of the lateral oscillation of the wing and so its drag there is no possibility of self-steering like downwind rotary windmills. So with the downwind weight of a tailvane, upwind location of the pendulum can balance the net weight about the tower yaw axis for easy turning in light winds without an expensive large diameter rolling element lower bearing encircling the tower. More subtle is that the yaw reaction to the center of mass sway of the pendulum in roll is then again opposed to the oscillating yaw torque of the wing drag (though smaller than for the floating hull).

The wing is more vulnerable in this well version because of the single pump without redundancy and the harder base surface, so protection against overswings due to loss of load is absolutely essential. The first line of defense is a mechanism zeroing twist at extreme roll, but allowing complete 360 twist normally, exemplified by the cam mechanism. This should at least limit the initial oscillation whilst the head rebuilds after a calm long enough for a foot valve leak to empty the drop pipe. The second line of defense for instance if the pump wire fatigues is a hefty lock which latches more extreme roll and holds it. This pendulum locking at horizontal stops the oscillation and the base yaws out of the wind, clearly signalling to the farmer that the pump needs attention. Whereas the rotary windpump keeps on turning and wearing with little sign that it isn't pumping at all.

Companion P5 WellPump development

The Spanish coil winch is a pure tension convertor, which can achieve very large output strokes with little lateral swing through a small yaw bearing. A paramount pump selection consideration was maintaining tension during a rapid downstroke to allow the fast pumping of small sizes of the oscillating windpump in high wind areas.

The solution arose indirectly in attacking the practical difficulty in releasing the water column and retreiving the footvalve that is the bane of deepwell reciprocating pumps. Instead of attaching it to the bottom of the cylinder, consider extending it from the piston to slide in its own smaller bore lower cylinder. It can then be pulled out with the piston from an open-top cylinder, releasing the water column very easily. The sequence of valve opening and closing remains the same as in the standard pump, but the output becomes the difference of the stroke volumes of the piston in the upper cylinder and the foot valve in the lower.

Suppose their ratio is 4:1 from a 2:1 diameter ratio. Then in the descending stroke the water column loses 1/4 of its ascending advance and acts upon the footvalve and so the pistonrod with 1/4 of its fullweight, the remaining 3/4 being borne by the cylinder transition and so the droppipe. The descent of the water in the lower cylinder with the foot valve helps backwash the filter below.

But also the maintenance of some tension on the downstroke allows use of vinyl-coated wire and that in turn the jointed rigid dropipe with continous flexible polyetheylene tube which just arcs down to the ground as it is inserted into the well. These give quantum improvements in ease of both piston or complete droppipe removal.

Not only is the downstroke return force greater than can be transmitted from the wellhead before the normal string of pumprods buckles, it is less degraded by the motion than the dead weight of such rods . For the rods can never fall under their own weight with faster acceleration than g=10m/sec 2. But here with the greater 1/4 the weight of the water column which is only accelerating 1/4 as fast downwards, the piston can easily be accelerated downwards faster than g. This allows longer and faster strokes than the standard pump which means smaller diameter and cost droppipe for the same output. Footpumping is too fast and too unidirectional for normal cylinders but matches the P5 ideally.

In the upstroke the drop-pipe's inner diametrical clearance adds the weight of an annular water column to its own weight to resist the upward lift of friction of the piston seal and of the flow. For pressure pumping to heads much above ground level, a normal bearing oilseal is very effective on the vinyl-coated pumpwire. A plain air tank can be used without an air bladder, as on the downstroke the descending water column can suck in air through a snifter valve below a non-return valve in the tank input pipe. In fact especially with the Wing'd Pump this provides enough flow of compressed air for auxiliary use.

To maintain good flow mechanical efficiency, the P5 must be carefully designed for acceptable flow losses in the upstroke when the net stroke volume must pass through the lower valve, reduced in diameter in the original concept. The losses would have been too high, so the pump concept had to be topologically developed, so the footvalve could operate in the upper cylinder with not much less flow area than in a standard pump.

Battery-charging option

The flutter oscillating wingmill was inspired by the important reciprocating end use in waterpumping and its poor match with rotary windmills. After the development of the Spanish coil winch, it was realised that it also effects a major oscillarotary stepup of about 15 from the slow pendulum angular velocity. This makes the winch shaft much faster than the shaft of a comparable rotary windpump and it alone within reach of a normal rotary single stepup of standard alternator speeds for low cost secondary battery-charging. The oscillation is easily rectified with a roller clutch on the hardenable alternator shaft. Any such windcharger needs a switch to energise the alternator field only when it is spinning fast enough, so switching within each winch cycle costs little if no more.

The choice of stepup was based on severely exposed environment with high impact loadings, and the possibility of generating 24v as +/- 12v with +12v field and a high stepup ratio. Again looking at vehicle parts for cheapness and availability, a ring gear & mating starter pinion (with built-in roller clutch) provided the highest ratio, strength, durability, and efficiency and very low friction in the rewind. The noise penalty was minimised by mounting the ring gear on a sound-absorbing plywheel on the winch shaft. Ideally the pair should be run in an oilbath for smoothness but keeping the oil in and off the rest of the machine and rainwater out requires a complete enclosure with a seal at the alternator at least; but a tarry grease gives reasonably durable open lubrication.