This invention relates to the lubrication of screw machines such as screw expanders, which may be used to generate power using, for example, steam as the working fluid.

Positive-displacement expanders are increasingly popular for use in power generation. One of the most successful positive-displacement machines is the p!ural-screw machine, most commonly embodied as a twin-screw machine. Such machines are disclosed in UK Patent Nos. GB 1197432. GB 1503488 and GB 2092676 to Svenska Rotor Maskiner (SRM).

Screw machines can be used as compressors or as expanders. The broadest concept of this invention is concerned with both types of screw machines, but the invention has particular benefits in the context of expanders. This specification is therefore primarily concerned with, and describes the invention in relation to, the use of screw machines as expanders. Screw machines for use as expanders will be referred to in this specification simply as screw expanders.

A key advantage of a screw expander over a turbine expander is the ability to handle a wet working fluid (i.e. a fluid containing both gaseous and liquid phases) with little risk of damage. This is because the fluid velocities within screw machines are approximately an order of magnitude lower than the fluid velocities typically encountered in turbine machines. Thus, screw expanders can admit fluids of any composition from pure liquid to dry vapour, while maintaining thermodynamic equilibrium between the phases. Turbine expanders, in contrast, are vulnerable to blade erosion if any significant fraction of liquid phase is entrained in the working fluid.

Screw expanders comprise a casing having at least two overlapping but non-intersecting bores. The bores accommodate respective meshing helical lobed rotors, which contra- rotate within the fixed casing. The casing encloses the rotors totally, in an extremely close fit. The central longitudinal axes of the bores are coplanar in pairs and are usually parallel, A male (or 'main') rotor and a female (or 'gate') rotor are mounted to the casing on bearings for rotation about their respective axes, each of which coincides with a respective one of the bore axes in the casing. The rotors are normally made of metal such as mild steel but they may be made of highspeed steel. It is also possible for the rotors to be made of ceramic materials. Normally, if of metal, they are machined but alternatively they can be ground or cast. The rotors each have helical lands, which mesh with helical grooves between the lands of at least one other rotor. The meshing rotors effectively form one or more pairs of helical gear wheels, with their lobes acting as teeth. Viewed in cross-section, the or each male rotor has a set of lobes corresponding to the lands and projecting outwardly from its pitch circle. Similarly viewed in cross-section, the or each female rotor has a set of depressions extending inwardly from its pitch circle and corresponding to the grooves of the female rotor(s). The number of lands and grooves of the male rotor(s) is different to the number of lands and grooves of the female rotor(s).

Prior art examples of rotor profiles are illustrated in Figures 1 (a) to 1 (d) and 2(a) to 2(d) of the accompanying drawings and will be described in more detail later.

The principle of operation of a screw expander is based on volumetric changes in three dimensions. The space between any two successive lobes of each rotor and the surrounding casing forms a separate working chamber. The volume of this chamber varies as rotation proceeds due to displacement of the line of contact between the two rotors. The volume of the chamber is a maximum where the entire length between the lobes is unobstructed by meshing contact between the rotors. Conversely the volume of the chamber is a minimum, with a value of nearly zero, where there is full meshing contact between the rotors at the end face.

Fluid to be expanded enters the screw expander through an opening that forms a high- pressure or inlet port, situated mainly in a front plane of the casing. The fluid thus admitted fills the chambers defined between the lobes. The trapped volume in each chamber increases as rotation proceeds and the contact line between the rotors recedes. At the point where the inlet port is cut off, the filling or admission process terminates and further rotation causes the fluid to expand as it moves downstream through the screw expander.

Further downstream, at the point where the male and female rotor lobes start to reengage, a low-pressure or discharge port in the casing is exposed. That port opens further as further rotation reduces the volume of fluid trapped between the lobes and the casing. This causes the fluid to be discharged through the discharge port at approximately constant pressure. The process continues until the trapped volume is reduced to virtually zero and substantially all of the fluid trapped between the lobes has been expelled.

The process is then repeated for each chamber. Thus, there is a succession of filling, expansion and discharge processes achieved in each rotation, dependent on the number of lobes in the male and female rotors and hence the number of chambers between the lobes.

As the rotors rotate, the meshing action of the lobes is essentially the same as that of helical gears. In addition, however, the shape of the lobes must be such that at any contact position, a sealing line is formed between the rotors and between the rotors and the casing in order to prevent internal leakage between successive chambers. A further requirement is that the chambers between the lobes should be as large as possible, in order to maximise fluid displacement per revolution. Also, the contact forces between the rotors should be low in order to minimise internal friction losses and to minimise wear.

As manufacturing limitations dictate that there will be small clearances between the rotors and between the rotors and the casing, the rotor profile is the most important feature in determining the flow rate and efficiency of a screw expander. Several rotor profiles have been tried over the years, with varying degrees of success.

The earliest screw expanders used a very simple symmetric rotor profile, as shown in Figure 1 (a). Viewed in cross-section, the male rotor 10 comprises part-circular lobes 12 equi-angularly spaced around the pitch circle, whose centres of radius are positioned on the pitch circle 14. The profile of the female rotor 6 simply mirrors this with an equivalent set of part-circular depressions 18. Symmetric rotor profiles such as this have a very large blow-hole area, which creates significant internal leakage. This excludes symmetric rotor profiles from any applications involving a high pressure ratio or even a moderate pressure ratio.

To solve this problem, SRM introduced its 'A' profile, shown in Figure 1 (b) and disclosed in various forms in the aforementioned UK Patent Nos. GB 1197432, GB 1503488 and GB 2092676. The 'A' profile greatly reduced internal leakage and thereby enabled screw compressors to attain efficiencies of the same order as reciprocating machines. The Cyclon profile shown in Figure 1(c) reduced leakage even further but at the expense of weakening the lobes of the female rotors 16. This risks distortion of the female rotors 16 at high pressure differences, and makes them difficult to manufacture. The Hyper profile shown in Figure 1 (d) attempted to overcome this by strengthening the female rotors 16.

In all of the above prior art rotor profiles, the relative motion between the meshed rotors is a combination of rotation and sliding.

Against this background, the Applicant developed the 'NT rotor profile as disclosed in its International Patent Application No. PCT/GB97/01333 published as WO 97/43550. The content of WO 97/43550 is incorporated herein by reference. References in this specification to the 'N' rotor profile refer to the profile of the invention that is described and defined in WO 97/43550.

The 'NT rotor profile is characterised in that, as seen in cross section, the profiles of at least those parts of the lobes projecting outwardly of the pitch circle of the male rotor(s) and the profiles of at least the depressions extending inwardly of the pitch circle of the female rotor(s) are generated by the same rack formation. The latter is curved in one direction about the axis of the male rotor(s) and in the opposite direction about the axis of the female rotor(s), the portion of the rack which generates the higher pressure flanks of the rotors being generated by rotor conjugate action between the rotors.

Advantageously, a portion of the rack, preferably that portion which forms the higher pressure flanks of the rotor lobes, has the shape of a cycloid. Alternatively, this portion may be shaped as a generalized parabola, for example of the form: ax + by ^q = 1. Normally, the bottoms of the grooves of the male rotor(s) lie inwardly of the pitch circle as 'dedendum' portions and the tips of the lands of the female rotor(s) extend outwardly of its pitch circle as 'addendum' portions. Preferably, these dedendum and addendum portions are also generated by the rack formation. The main or male rotor 1 and gate or female rotor 2 shown in the diagrammatic cross section of a twin-screw machine of Figure 2(a) roll on their pitch circles, P ₁ , P ₂ about their centres O,, and 0 ₂ through respective angles

The pitch circles P have radii proportional to the number of lands and grooves on the respective rotors.

If an arc is defined on either main or gate rotor as an arbitrary function of an angular parameter φ and denoted by subscript d:

the corresponding arc on the other rotor is a function of both φ and ψ:

ψ is the rotation angle of the main rotor for which the primary and secondary arcs have a contact point. This angle meets the conjugate condition described by Sakun in Vintovie kompressori, Mashgiz Leningrad, 1960: which is the differential equation of an envelope of all 'd' curves. Its expanded form

This can be expressed as a quadratic equation of sin ψ. Although it can be solved analytically, its numerical solution is recommended due to its mixed roots. Once determined, ψ is inserted in (3) and (4) to obtain conjugate curves on the opposite rotor. This procedure requires the definition of only one given arc. The other arc is always found by a general procedure.

These equations are valid even if their coordinate system is defined independently of the rotors. Thus, it is possible to specify all 'd' curves without reference to the rotors. Such an arrangement enables some curves to be expressed in a more simple mathematical form and, in addition, can simplify the curve generating procedure.

A special coordinate system of this type is a rack (rotor of infinite radius) coordinate system, indicated at R in Fig. 2(b), which shows one unit of a rack for generating the profiles of the rotors shown in Figure 2(a). An arc on the rack is then defined as an arbitrary function of a parameter :

Secondary arcs on the rotors are derived from this as a function of both φ and ψ

ψ represents a rotation angle of the rotor where a given arc is projected, defining a contact point. This angle satisfies the condition (5) which is:

The explicit solution ψ is then inserted into (9) and (10) to find conjugate arcs on rotors.

Fig. 2(c) shows the relationship of the rack formation of Figure 2(b) to the rotors shown in Figure 2(a), and shows the rack and rotors generated by the rack. Figure 2(d) shows the outlines of the rotors shown in Figure 2(c) superimposed on a prior art rotor pair by way of comparison.

Wherever curves are given, their convenient form may be: which is a 'general circle' curve. For p = q = 2 and a = b = 11r it is a circle. Unequal a and b will give ellipses; a and b of opposite sign will give hyperbolae; and p = 1 and q = 2 will give parabolae.

In addition to the convenience of defining all given curves with one coordinate system, rack generation offers two advantages compared with rotor coordinate systems: a) a rack profile represents the shortest contact path in comparison with other rotors, which means that points from the rack will be projected onto the rotors without any overlaps or other imperfections; b) a straight line on the rack will be projected onto the rotors as involutes.

In order to minimize the blow hole area on the high pressure side of a rotor profile, the profile is usually produced by a conjugate action of both rotors, which undercuts the high pressure side of them. The practice is widely used: in GB 1197432, singular points on main and gate rotors are used; in GB 2092676 and GB 21 12460 circles were used; in GB 2106186 ellipses were used; and in EP 0166531 parabolae were used. An appropriate undercut was not previously achievable directly from a rack. It was found that there exists only one analytical curve on a rack which can exactly replace the conjugate action of rotors. This is preferably a cycloid, which is undercut as an epicycloid on the main rotor and as a hypocycloid on the gate rotor. This is in contrast to the undercut produced by singular points which produces epicycloids on both rotors. The deficiency of this is usually minimized by a considerable reduction in the outer diameter of the gate rotor within its pitch circle. This reduces the blow-hole area, but also reduces the throughput.

A conjugate action is a process when a point (or points on a curve) on one rotor during a rotation cuts its (or their) path(s) on another rotor. An undercut occurs if there exist two or more common contact points at the same time, which produces 'pockets' in the profile. It usually happens if small curve portions (or a point) generate long curve portions, when considerable sliding occurs.

The 'N' rotor profile overcomes this deficiency because the high pressure part of a rack is generated by a rotor conjugate action which undercuts an appropriate curve on the rack. This rack is later used for the profiling of both the main and gate rotors by the usual rack generation procedure.

The following is a detailed description of a simple rotor lobe shape of a rack generated profile family designed for the efficient compression of air, common refrigerants and a number of process gases, obtained by the combined procedure. This profile contains almost all the elements of modern screw rotor profiles given in the open literature, but its features offer a sound basis for additional refinement and optimisation.

The coordinates of all primary arcs on the rack are summarised here relative to the rack coordinate system.

The lobe of this profile is divided into several arcs. The divisions between the profile arcs are denoted by capital letters and each arc is defined separately, as shown in Fig. 2(c). Segment A-B is a general arc of the type on the rack with p = 0.43 and q - 1.

Segment B-C is a straight line on the rack, p = q = 1. Segment C-D is a circular arc on the rack, p = q = 2, a = b.

Segment D-E is a straight line on the rack. Segment E-F is a circular arc on the rack, p = q = 2, a = b.

Segment F-G is a straight line.

Segment G-H is an undercut of the arc G ₂ -H ₂ which is a general arc of the type p = 1 , q = 0.75 on the main rotor.

Segment H-A on the rack is an undercut of the arc A, -H-i , which is a general arc of the type , p = 1 , q = 0.25 on the gate rotor.

At each junction A,. ..H, the adjacent segments have a common tangent.

The rack coordinates are obtained through the procedure inverse to equations (7) to (1 1 ).

As a result, the rack curve E-H-A is obtained and shown in Figure 2(c). Figure 2(d) shows the profiles of main and gate rotors 3, 4 generated by this rack procedure superimposed on the well-known profiles 5, 6 of corresponding rotors generated in accordance with GB 2092676, in 5/7 configuration.

With the same distance between centres and the same rotor diameters, the rack- generated profiles give an increase in displacement of 2.7% while the lobes of the female rotor are thicker and thus stronger. in a modification of the rack shown in Figure 2(c), the segments GH and HA are formed by a continuous segment GHA of a cycloid of the form: y = R ₀ cosx-R _p , y = R ₀ sinT-R _p x, where R ₀ is the outer radius of the main rotor (and thus of its bore) and R _p is the pitch circle radius of the main rotor.

The segments AB, BC, CD, DE, EF and FG are all generated by equation (12) above. For AB, a = b, p = 0.43, q = 1. For the other segments, a = b = 1/r, and p = q - 2. The values of p and q may vary by ±10%. For the segments BC, DE and FG r is greater than the pitch circle radius of the main rotor, and is preferably infinite so that each such segment is a straight line. The segments CD and EF are circular arcs when p = q = 2, of curvature a = b.

The 'N' rotor profile described above is based on the mathematical theory of gearing. Thus, unlike any of the rotor profiles described previously with reference to Figures 1 (a) to 1 (d), the relative motion between the rotors is very nearly pure rolling: the contact band between the rotors lies very close to their pitch circles.

The 'N' rotor profile has many additional advantages over other rotor profiles, which include low torque transmission and hence small contact forces between the rotors, strong female rotors, large displacement and a short sealing line that results in low leakage.

Overall its use raises the adiabatic efficiencies of screw expander machines, especially at lower tip speeds, where gains of up to 10% over other rotor profiles in current use have been recorded. Accepted wisdom in the art is that if the helical formations of the rotors are not to be lubricated, external meshed 'timing' gears must be provided to govern and synchronise relative movement of the rotors. Transmission of synchronising torque between the rotors is effected via the timing gears, which therefore avoids direct contact between the meshed helical formations of the rotors. In this way, the timing gears allow the helical formations of the rotors to be free of lubricant.

Alternatively, external timing gears may be omitted, such that synchronisation of the rotors is determined solely by their meshed relationship. This necessarily implies some transmission of synchronising torque from one rotor to the other via their meshed helical formations. In that case, the helical formations of the rotors must be lubricated to avoid hard contact between the rotors, with consequent wear and probable seizure. Reflecting these alternative approaches to synchronisation of the rotors and their different lubrication requirements, there are two main types of screw expanders: Oil-flooded' and Oil-free'. An oil-flooded machine relies on oil entrained in the working fluid to lubricate the helical formations of the rotors and their bearings and to seal the gaps between the rotors and between the rotors and the surrounding casing. It requires an external shaft seal but no internal seals and is simple in mechanical design. Consequently, it is cheap to

manufacture, compact and highly efficient.

In contrast, an oil-free machine does not mix the working fluid with oil. Thus, timing gears are provided to avoid contact between the helical formations of the rotors. Each timing gear wheel turns with a respective one of the rotors and those gear wheels mesh outside the casing, where they are lubricated externally with oil. Thus, Oil-free' refers to the interior of the casing rather than to the machine as a whole. To prevent oil entering the casing and becoming entrained in the working fluid, internal seals are required on each shaft between the casing and the gear wheels, as well as an external shaft seal. It follows that oil-free machines are considerably bulkier and much more expensive to manufacture than oil-flooded machines. However, the rotors of oil-free machines can rotate at higher speeds, without excessive viscous drag. Hence, the flow capacity per unit volume of oil- free machines is higher than for oil-flooded machines.

Both oil-flooded and oil-free machines require an external heat exchanger for the lubricating oil before it is readmitted to the machine. In expander applications, the purpose of the heat exchanger differs between oil-flooded and oil-free machines.

An oil-free machine uses the heat exchanger to cool the oil. To complete the circuit, an oil tank, oil filters and a circulating pump are required to return the oil to the bearings and the timing gears. Conversely, an oil-flooded machine requires a separator downstream of the expander to remove the oil from the discharged working fluid. The separated oil then has to be re-pressurised in a pump and the heat exchanger has to heat the oil before it is returned to the high-pressure end of the casing. This is to avoid chilling the working fluid entering the casing, which otherwise would reduce the efficiency of the expander.

These lubrication systems increase the total cost of both types of expander but the add-on cost is much greater for an oil-free expander. Indeed the total cost of an oil-free expander is typically an order of magnitude greater than the cost of an oil-flooded expander of equal capacity.

Cost aside, both oil-free and oil-flooded lubrication systems have other deficiencies.

Attempts to produce oii-free expanders have led to difficulties in that the internal shaft seals may not completely separate the working fluid and the oil that lubricates the timing gears. This problem is particularly acute where the working fluid is a refrigerant or a hydrocarbon, which are highly soluble in lubricating oil.

In the case of oil-flooded expanders, it is practically impossible to separate and remove the oil completely from the working fluid after expansion. This leads to a gradual accumulation of oil in other parts of the system, which creates operating problems. Of course, the same problem arises in oil-free expanders whose sealing arrangements are unable to keep the oil completely separate from the working fluid. However, oil-free expanders are generally better in this respect than oil-flooded expanders; thus, applications that are particularly sensitive to oil contamination may necessitate the adoption of oil-free expanders despite their bulk, complexity and much higher cost. As has been mentioned, the rotors of a screw expander are mounted to the casing on bearings for rotation about their respective axes. Various types of bearings may be used. Lubrication is also an issue here of course.

The vast majority of screw machines use rolling-element bearings to support the rotors. This allows bearing tolerances to be very small, which in turn minimises the clearances that must be maintained between the rotors and between the rotors and the casing. This therefore minimises internal leakage and hence maximises efficiency.

Rolling-element bearings include ball bearings and roller bearings. They function by maintaining rolling contact through a set of spherical balls or cylindrical or frusto-conical rollers that separate two surfaces. If suitably arranged, rolling-element bearings may sustain both radial and axial loads. Despite the predominantly rolling motion between the rolling elements and the tracks on which they run, a boundary film of oil must be maintained between those parts to minimise wear and frictional heating.

The Applicant's International Patent Application No. PCT/GB2006/02148 published as WO 2006/131759 shows that it is possible to lubricate rolling-element bearings in a screw expander even where a liquid component of the working fluid contains only a low concentration of dissolved oil. If that liquid component is supplied to the bearings, the working fluid will evaporate due to frictional heating and will leave enough oil in the bearing housing to supply the boundary film needed to keep the bearings operating effectively. However, the lubricating principles of WO 2006/131759 cannot be applied when using a screw expander for expanding fluids such as steam, or any other fluids in which oil cannot be dissolved in its liquid phase, or where the presence of even small traces of lubricating oil in the working fluid is not permissible. US 6217304 to Shaw discloses a screw compressor for refrigeration apparatus that, in theory, can use droplets of liquid refrigerant entrained in the gaseous phase of the refrigerant to seal, cool and lubricate the rotors. There is provision to inject droplets of liquid refrigerant into the refrigerant flow if necessary, with the penalty of complication. However, US 6217304 does not provide an enabling disclosure of how such a compressor may be run without oil in the working fluid. It merely refers to the use of a thermoplastic or other suitable composite material for the male rotor that allows a small clearance between the rotors.

US 6217304 notwithstanding, no refrigeration compressor has ever been operated successfully to the Applicant's knowledge without copious oil being dissolved in or entrained by the refrigerant. This is not an option if using steam as the working fluid in an open-circuit expander as opposed to a closed-circuit compressor. Moreover, using liquid refrigerant to lubricate a compressor implies incomplete evaporation of the refrigerant and hence a very poor Coefficient of Performance (i.e. refrigeration efficiency). Consequently the design choices proposed in US 6217304 are difficult to justify in practical machines.

It is against this background that the present invention has been made.

From one aspect, the invention resides in a screw machine for use with a working fluid containing a liquid phase, the machine comprising two or more rotors having meshed, lubricated helical formations, wherein the rotors have an 'N' profile as defined herein and, in use, lubrication of the helical formations of the rotors is achieved exclusively - or at least substantially so - with the liquid phase of the working fluid. The invention may also be expressed as a method of lubricating a screw machine when using a working fluid containing a liquid phase, the machine comprising two or more rotors having meshed, lubricated helical formations, the rotors having an 'N' profile as defined herein, wherein the method comprises lubricating the helical formations of the rotors substantially exclusively with the liquid phase of the working fluid.

References to 'substantially exclusively' in this specification are intended to reflect that minor or trace amounts of various other fluids could be entrained in the working fluid, even if not deliberately added to it, and could have some very slight lubricating effect.

However, effective lubrication still depends wholly or predominantly on the presence of the liquid phase of the specified working fluid, such that its absence would render lubrication ineffective.

Essentially all of the lubrication duty is therefore performed by the liquid phase of the specified working fluid. Moreover, the lubricating liquid entrained in the working fluid is preferably derived from the working fluid entering the expander, substantially without prior addition of further liquid to the working fluid. This is an advantageously simple arrangement. However, prior addition of further liquid is possible if necessary and is not excluded from the invention in its broadest sense.

Using the liquid phase of the working fluid as the lubricant for the helical formations obviates expensive lubrication systems for delivering oil lubricant to those formations. It also avoids oil contamination of the working fluid and the need for separating oil from the working fluid after the working fluid has passed through the machine.

The rotors may be manufactured of any suitable material. To minimise whatever wear may occur, their helical formations may be coated with a low-friction coating, for example Balinit C2 (trade mark) as offered by Oerlikon Balzers. Balinit C2 is a "WC/C" coating that is deposited by physical vapour deposition (PVD) and comprises a carbide phase and a carbon phase. The use of a low-friction coating on at least the helical formations of the rotors is preferred where wear characteristics make such a coating desirable. However, for reasons of cost, it is preferred that the helical formations of the rotors are left uncoated if possible; it is a potential benefit of the invention to allow this by virtue of the use of 'N'- profile rotors.

To be sure of the rotors being satisfactorily lubricated, it is safest to use coated 'N'-profile rotors. However, the Applicant's tests also suggest that rotors with other profiles could conceivably be used in some applications if coated with a low-friction coating such as Balinit C2. Thus, from another aspect, the invention also resides in a screw machine for use with a working fluid containing a liquid phase, the machine comprising two or more rotors having meshed, lubricated helical formations coated with a low-friction coating, wherein, in use, lubrication of the helical formations of the rotors is achieved substantially exclusively with the liquid phase of the working fluid. That aspect of the invention may also be expressed as a method of lubricating a screw machine when using a working fluid containing a liquid phase, the machine comprising two or more rotors having meshed, lubricated helical formations coated with a low-friction coating, wherein the method comprises lubricating the helical formations of the rotors substantially exclusively with the liquid phase of the working fluid. In all aspects of the invention, the rotors are preferably supported by bearings that are also substantially exclusively lubricated, in use, by the liquid phase of the working fluid. The use of the liquid phase to lubricate the bearings obviates expensive lubrication systems for delivering oil lubricant to those bearings. It also avoids oil contamination of the working fluid and, again, the need for separating oil from the working fluid after the working fluid has passed through the machine.

Bearings lubricated by the liquid phase of the working fluid are preferably hydrodynamic for simplicity, but they may be hydrostatic. Whilst the larger clearances required for hydrodynamic bearings and consequent whirling of the shaft within the bearing may make the machine slightly less efficient than an equivalent machine using rolling-element bearings, screw steam expanders can be produced using water in the liquid phase as the bearing lubricant.

Hydrodynamic bearings operate by maintaining a film of lubricant between the rotating or sliding parts and their static casing, so that there is no contact between them except at start-up and shutdown. The basic principle of operation is that the film is not of uniform thickness. In the case of journal bearings, this happens where the centre of rotation of the shaft is displaced from the centre of radius of the surrounding casing. This creates a non-uniform film of lubricant around the shaft, which results in a huge increase of pressure in the lubricant in the region where the film is thinnest. The difference in pressure around the film is sufficient to urge the shaft back into alignment with the surrounding casing, hence preventing the shaft and the casing from coming into contact.

The pressures created in the film of lubricant depend on the viscosity of the lubricant and the reduction of thickness achieved in the lubricant film. Normally such bearings are oil- lubricated. However, some specialist firms such as Waukesha Bearings of Wisconsin, USA have developed hydrodynamic bearings that employ liquids of very low viscosity, such as water and light hydrocarbons. The key to the success of these bearings is the ability to operate with very fine minimum film thicknesses and also the use of bearing materials that do not readily seize or abrade during start-up and shutdown. Hydrostatic bearings are also possible within the broad inventive concept but they are less preferred as they need an external pump and circulating system to implement them. Specifically, such bearings keep the rotor shaft from making contact with the surrounding casing by a series of pads in the casing through which high-pressure gas or liquid is admitted. The balance of forces from the pressure between the rotor shaft and the pads prevents contact being made between them and the rotor shaft is supported by the fluid as it revolves.

Rolling element bearings, such as those of the ball or roller type, when suitably designed, can also be used to support the rotors.

It would be possible for the liquid that lubricates the bearings to be sourced separately from the working fluid. However, it is elegant and therefore preferred for the bearings to be lubricated, like the helical formations of the rotors, by liquid phase derived from the working fluid.

In expander applications, the working fluid is, most advantageously, water or wet steam such that the liquid phase used to lubricate the helical formations, and optionally also the bearings, is liquid water such as may be entrained in the flow of steam. However, screw machines could also be similarly designed for any working fluid such as hydrocarbons or refrigerants, provided that the liquid phase of the same fluid is readily available for use in lubricating the helical formations of the rotor and optionally also the bearings.

The use of a low-viscosity liquid such as water to lubricate the helical formations of the rotors has the further advantage that the lubricant gives rise to less viscous drag than would oil. Thus, the rotors can rotate at higher speed than in an equivalent oil-flooded machine, to the benefit of flow capacity.

For simplicity, compactness and low cost, and particularly to avoid unnecessary oil lubrication, it is much preferred that the rotors are not linked by separate timing gears. Thus, like an oil-flooded machine, it is advantageous for synchronisation of the rotors to rely simply upon cooperation of the meshed helical formations. This means that synchronising torque is transmitted from one rotor to another substantially exclusively via their meshed helical formations.

Steam screw expanders have been built and tested on some occasions in the past, but there is no record of any of them being process-fluid lubricated, nor of any of them operating successfully.

By means of the invention, the entire oil lubrication system and the additional components required either for oil-free or oil-flooded machines can be eliminated if the working fluid can also be used to lubricate the bearings.

Test 1

In testing screw machines having an 'N' rotor profile, the Applicant experimented with an air compressor of the oil-flooded type, thus being without timing gears. Hence, the compressor relied on lubricated contact between the helical formations of the rotors to synchronise the rotors.

Counter-intuitively, the Applicant experimented with lubricating the helical formations of the rotors only with water injected into the working fluid in place of oil. In that case, the bearings mounting the rotors to the casing were of the rolling-element type and were packed with grease. The rotors were steel coated with Balinit C2 coating.

The compressor was then run for 150 hours with contact between the helical formations of the rotors, lubricated only by water, injected into the working fluid instead of oil. At the end of that period, the rotors were examined and showed no signs of wear or damage, other than a light polish on the contact band.

Test 2

The coated rotors of Example 1 were replaced by a pair of 'N'-profile rotors of uncoated steel and run for a further five hours. At the end of that period, the uncoated rotors were examined and they, too, showed no significant signs of wear.

Test 3 Subsequently, the compressor with the uncoated 'N'-profile rotors was run, in error, for two hours, without any water injection into the working fluid at all. The rotors still sustained no significant damage, although some damage would of course be expected if the machine were run over substantially longer periods without lubrication of the helical formations of the rotors.

The Applicant has concluded that for screw machines with 'N'-profile rotors, the primary function of the lubrication system is to lubricate the bearings. As long as some liquid is present on the helical formations of the rotors - even if that liquid is of low viscosity such as water - there is no need for timing gears to avoid direct rotor contact. The invention exploits this discovery and embodies this principle.

Reference has already been made to Figures 1 (a) to 1 (d) and 2(a) to 2(d) of the accompanying drawings to describe some prior art rotor profiles. In order that the invention may be more readily understood, reference will now be made, by way of example, to the accompanying drawings, in which:

Figure 3 is a schematic sectional view through a steam expander in accordance with a first embodiment of the invention; and

Figure 4 is a schematic sectional view through a steam expander in accordance with a second embodiment of the invention.

Referring firstly to Figure 3 of the drawings, a screw expander 20 comprises a fixed casing 22 containing two meshing helical lobed rotors 10, 16 that contra-rotate within the casing about parallel axes. The rotors 10, 16 are of any suitable material such as steel, optionally coated with a low-friction coating such as the aforementioned Balinit C2. They have an 'N' rotor profile as disclosed by the Applicant in WO 97/43550. Each rotor 10, 6 is mounted to a respective shaft 24, 26 that is mounted, in turn, to the casing 22 by hydrodynamic bearings 28 supporting each end of each shaft. One of the shafts 26 extends out of the casing 22 to drive a generator (not shown) for producing electricity. The working fluid to be expanded (in this example, wet steam) enters the casing 22 at high pressure through an inlet port 30. The steam flows and expands through the interior of the casing 22, causing the rotors 10, 16 to turn at high speed, and exits the casing 22 at lower pressure through a discharge port 32.

There are no timing gears to synchronise rotation of the rotors 10, 16. Instead, the rotors 10, 16 are synchronised by virtue of the meshing engagement of their helical formations. This requires the helical formations to be lubricated, which is ensured by the liquid phase of water entrained in the wet steam feed.

In the embodiment shown, the bearings 28, too, are lubricated by water derived from the wet steam feed. A reservoir 34 communicates with the inlet port 30 and provides water under pressure through feed lines 36 to each of the bearings 28.

In an optional refinement shown in Figure 4 of the drawings, a balance piston 38 at the end of each rotor shaft 24, 26 uses the pressure of the working fluid to oppose axial loading of the shaft, hence reducing the axial loading experienced by the beahngs 28 that support the shafts. Thus, a pressure line 40 connects the balance pistons 38 to the inlet port 30.

The working fluid for expansion may be obtained from various sources, such as steam from a geothermal source. In this respect, it will be recalled that a key advantage of a screw expander over a turbine expander is the ability to handle a wet working fluid (i.e. a fluid containing both gaseous and liquid phases) with little risk of damage. A screw expander is also much better than a turbine expander at handling contaminated or dirty working fluids, such as wet steam containing fine particles of sand from a geothermal source or rust from corroded pipework. Another advantage is that screw expanders are potentially more cost-effective than turbines for relatively small power outputs.

The foregoing description shows that the invention makes it possible to design and manufacture steam expanders that do not require timing gears, internal shaft seals, lubricant storage, lubricant pumps, lubricant filters or heat exchangers - as are required for oil-free machines - or lubricant pumps, heat exchangers and oil separators - as are required for oil-flooded machines. Moreover, the problem of lubricating oil contaminating the working fluid, which is common to both oil-free and oil-flooded machines in the prior art, is completely overcome by the invention.

Industrial steam systems represent a major potential application of screw expanders in accordance with the invention. Many industrial processes require a supply of steam, examples being food preparation, paper-making and chemical processes. Typically, a central boiler generates steam at a moderately high pressure and that steam is distributed around a factory, plant or other industrial installation via a pipe system. Steam is drawn off through a branch pipe at each location where it is required.

As different processes in an industrial installation may require different steam pressures, each branch pipe typically has a control valve that throttles the steam to whatever lower pressure may be required for the process in question. It is possible to use a screw expander instead of a throttle valve to reduce steam pressure. This makes it possible to recover power from the expansion process while still supplying steam at the required lower pressure. The benefits of cost, robustness, compactness, reliability, efficiency and avoidance of oil contamination allowed by the present invention are crucial to the acceptance of steam expanders for such applications, especially where there is scope to replace multiple throttle valves in an industrial installation.