The Underfall Yard Water Pumps (Bristol Docks): a Mechanical Puzzle.

G4OEP Homepage

Underfall Yard water pumps bristol

Fig 1.  General View of Two of the three Pumping Machines

The three water pumps (two units shown above) in the Underfall Yard form part of a complex of preserved historic features of Bristol's Floating Harbour.  The present electrically driven pumps were installed in 1905, and replaced earlier steam-driven pumps installed in the late nineteenth century.  The 1905 pumps are now beautifully preserved as working exhibits, powered by large three-phase induction motors.  When in operation they form a fascinating mobile display.  Originally the pumps produced high-pressure water at 51 bar which was used as a source of hydraulic power for operating the various lock-gates, swing bridges, cranes and capstans found in various parts of the docks complex.  At present the high-pressure supply produced while the pumps operate is discharged wastefully.  Perhaps a fountain can be installed !

Each of the three pumping machines comprises a pair of reciprocating double-acting force-pumps, one on each side of the central axis which contains a flywheel/gearwheel assembly.  In the following discussion, only one of these six pumps will be considered.

At the site an explanatory sheet is provided which purports to show how the pump operates (Fig 2, right). It seems that the pump is described as a compound system which raises the water pressure to the final high level output in two stages, using first a low pressure (LP) cylinder, and then a high pressure (HP) cylinder.  But this is entirely misconceived.  Each of the six pumps is in fact a double acting ram pump which increases the pressure of the water in one step. The following notes describe the pumps and provide an accurate explanation of their operation.

Each of the six individual pumps has two piston-cylinder combinations, one at each end of a single iron casting which forms a common cylinder block.   The two pistons (or rams) are connected together by side-bars which ensure that the strokes of the pistons are strictly equal: as the inner piston enters its cylinder, so the outer piston is withdrawn from its cylinder by an equal amount.

The piston and cylinder which is nearest to the fly-wheel and crank (the inboard cylinder) has a greater diameter than the outboard piston and cylinder, and it is this difference in size which has lead to the assumption that the cylinders operate at different pressures, and to the idea of referring to the outboard cylinder as the HP cylinder and the inboard cylinder as the LP cylinder.  Compound compressors used for compressing gasses can work on this principle, but water pumps cannot, the reason being that water is very nearly incompressible.  

The significance of this point becomes clear when stage 'D' of the cycle is considered according to the original explanation (Fig 2).  The diameter of the 'LP' piston is greater than that of the 'HP' pump and the strokes of both are the same.  Therefore, as the 'LP' piston moves to the right, the volume delivered by the 'LP' cylinder exceeds that received by the 'HP' cylinder and this excess volume must be accounted for - it must go somewhere.  The explanatory chart (Fig 2) does not acknowledge this difficulty and does not tell us where the excess volume goes.  In fact the explanation is very simple;  the delivery valve (outlet valve 2) opens, and the excess volume of water is delivered directly into the high pressure receiver.  This is illustrated in stage D of Fig 3.  If the volume of the 'LP' cylinder is twice that of the 'HP' cylinder, exactly half of the volume pumped out of the former will be used the fill the latter, and the remaining half will form the output of the pump during this stroke.  Then, returning to stage 'A',  the contents of the 'HP' cylinder will be pumped out during the return stroke, and the result will be equal volumes of high-pressure output on each stroke of the pump.  The machine is thus a double-acting single stage pump.

An interesting aspect of  this type of double-acting pump is that it can be made with just two valves; the non-return valve marked 'Outlet Valve 2' on Fig 3 is actually redundant, and all that is needed is the inlet valve to the inboard cylinder, and the non-return valve in the transfer port between the two cylinders (outlet valve 1).  The latter is sufficient to prevent water from entering the inboard cylinder from the high-pressure receiver when the pistons move to the left.  This economy of parts gives this type of double-acting pump an economic advantage over the more conventional type of double-acting pump which has two identical ram pumps, each with a single cylinder and two valves, operated by cranks set at 180 degrees. The fact that our pump has only one crank represents a further economy of design.  However, in the present installation a non-return valve in the position of 'Outlet Valve 2' might be useful in order to isolate the pump from the high pressure receiver when the pump is switched off.

The pumps and their associated pipework are not exactly rocket science, but are nevertheless enigmatic.  It is not at all easy to trace the flow-path and to identify with certainty the functions of the various parts.  We know that there must be an inlet valve, and also a second valve between the two cylinders, but without an extensive dismantling of the mechanism, or access to historic technical drawings, one must identify the parts and interpret their function using a certain amount of guesswork, assisted by whatever engineering insight one can bring to bear.
Underfall yard pumps bristol

Fig 2.  How the pumps work according to the existing explanation.

Underfall yard water pumps bristol

Fig. 3.  An alternative explanation of how the pumps work.  They are double-acting single stage pumps
  The close-up view (Fig 4,  right) gives a good view of the various parts of one of the pumps - click on it to see the picture full-size.  

An eye-catching feature of each pump is a large and prominent air chamber in the form of a domed cylinder (marked '1' in the lower picture, right).  Air chambers of this kind are usually attached to the discharge pipe of reciprocating pumps in order to produce a more steady flow in the discharge system, and to minimise pressure variations at the outlet of the pump.  Without such a reservoir, the speed of flow of water in the delivery pipe varies according to the movement of the piston, and the corresponding acceleration and deceleration of the mass of water in the delivery pipe causes peaks and troughs of pressure at the outlet of the pump, which  could be potentially damaging.  So it is natural to assume that the air chambers serve this function, and are attached to the final output of the pumps.  But while discussing this with the engineering staff, I was assured that the air-chambers are actually attached to the inlet system, and this seems to be the case when one examines the two pictures (right) closely. The outlet pipe (with a pressure guage attached) and the delivery pipe (horizontally connecting the two sides of each installation) are clearly identifiable, and they do not connect with the single, centrally located air chamber.

Similarly, the two inlet valves, one on each side of the air-chamber (visible in Fig 4) seem to be more complex than a simple non-return valve, and this suggests that they incorporate something more complicated.  Likewise the part where the pressure guage is attached seems to be more than a simple piping elbow, so perhaps it contains a delivery valve after all.  It seems that there are still some aspects of these pumps which need further explanation, but I feel that this discussion might make a useful contribution to identifying the type of pump we are looking at, and understanding the various parts of the mechanism.
Underfall yard water pumps Bristol

Fig 4.  Detail of one pair of Pumps.

Underfall yard water pumps Bristol

Fig 5.  Showing Air Chamber (marked '1') and possible redundant 'Outlet Valve 2' beneath the pressure guage.
Connecting the air-chamber to the inlet of a pair of pumps makes sense according to the present interpretation in which each of a pair of pumps is double-acting.  If the cranks on opposite sides of a pair of pumps are set at 90 degrees, four equally spaced output pulses will occur on each revolution, and successive pulses will overlap to some degree.  The combined output flow will therefore be substantially constant with only minor variations.  However the situation is different at the intake.  Here each pump accepts a double-volume of intake during alternate strokes (i.e. on that stroke where the larger cylinder fills).  These pulsations will also overlap to some degree, adding together to make the combined variation of flow at the inlet extreme.  This will fully justify an air reservoir at the inlet.   Figure 4 shows the rams of the nearer pump at their leftmost position, while those of the further pump are at mid-stroke, confirming that the cranks are indeed at 90 deg.

In view of the uncertain identification of the parts, I am cautious in putting forward this interpretation of the machines, but I definitely prefer it to the interpretation suggested by the explanatory chart provided at the site (Fig 2).  A discussion of reciprocating piston pumps, including the type described here can be found by following this link:

An Engineering Note.  The fact that these pumps have two cylinders of different diameters should not be taken as an indication that the pumps are multistage compressors.  Such compressors are commonly used for compressing gasses, but are rarely if ever used for pumping liquids. Technically these two tasks have distinctly different characteristics, and require different technical approaches. There are strong engineering reasons for using the multistage approach for gasses, but these do not apply in the case of liquids.  The reason is that whereas gasses are readily compressible,  liquids are almost completely incompressible.  Thus in stage 'D' of Fig 1, if the working fluid were a gas, it is completely feasible that the larger volume displaced by the 'LP' piston could be compressed into the smaller volume of the 'HP' cylinder, but when water is being pumped, this degree of compression is impossible, or, if it were possible, would require an extreme pressure way beyond that being considered here.

Gasses are often compressed in two or more stages for two reasons, neither of which applies to liquids.  Firstly, when a gas is compressed its temperature rises (adiabatic heating).  This causes a temporary excess of pressure rise, and this in turn makes it more difficult (consuming more work) to compress the gas further.  If after a certain degree of compression, the gas is cooled, its pressure falls, and this makes further compression easier (requiring less energy).  The loss of pressure during inter-stage cooling might seem to be a disadvantage, but this is not so, since the compressed gas will eventually cool anyway.  This loss of pressure is unavoidable, so it makes sense to use it to reduce the energy cost of the compresser.  This is discussed in greater detail here:

The second reason for compressing gasses in stages concerns the practical design of the pump.  Pumping involves two processes which occur successively as the piston moves into cylinder.  Firstly, at the beginning of the piston's stroke, the volume of the fluid in the cylinder is reduced and the pressure rises towards that of the receiver.  Then, when the pressure inside the cylinder slightly exceeds that of the fluid in the receiver, the outlet valve opens, and compressed fluid is delivered into the receiver at constant pressure.  In the case of liquids, the initial compression (i.e. reduction of volume) phase is insignificant since liquids are almost incompressible.  In this case displacement occurs immediately, and liquid is delivered into the receiver during the entire stroke of the piston.  In the case of gasses, hovever, the compression phase is greatly extended, and displacement occurs only towards the end of the piston stroke, and might not occur at all !  For example, if 1litre of gas were compressed from 1bar to 50bar at constant temperature in a single stage, (approximately the pressures of the underfall pumps), the gas must be compressed to 1/50 of its original volume (i.e. 20ml) before its pressure is sufficient to open the outlet valve.  Under ideal conditions, this volume of gas could be delivered to the receiver, but in a practical case it is possible that no gas would be delivered at all.  The reason for this is that there will inevitably be a certain amount of 'dead space' in the cylinder.  Clearance is needed to ensure that the piston does not hit the end of the cylinder at the end of its stroke, and also a certain volume is needed to accomodate the inlet and delivery valves.  If this combined volume were to amount to 20ml or more, a single-stage pump could not possibly work in the pressure range we are considering.  Of course since water is practically incompressible, dead space can be as large as you like in a water pump, and is irrelevant.

So we see that multistage compression has big advantages, and might even be strictly necessary in gas compressor, but is usually an unnecessary complication in water pumps, and is rarely if ever used when pumping liquids.