PUMP INTAKE DESIGN

Th Use of Scale Models

By J. S. PATERSON, B.Sc., C. Eng., AM.I.Mech.E., AM.I.Mar.E.,

Chief Engineer, Research and Development, Drysdale and Co. Ltd., Glasgow, Scotland - a member of the Weir Group of Companies.  

Issued by Engineering in Britain Information Services

Considerable attention has been given to the use of scale models as a guide to design, especially in branches of science such as fluid mechanics where theoretical analysis can provide reliable information only in the simplest cases. The design of centrifugal pumps themselves has reached a stage where to improve efficiency, much research is necessary for limited returns; but by applying model techniques to the design of pumping systems in general, and pump intakes in particular, significant cost savings can still be made. The article reviews current practice in this field and shows bow further saving could be made if greater use were made of model investigations.

Even in this age exploration, design on the grand scale is still associated primarily with civil engineering projects, and a large proportion of these involve the flow of water.

The design of hydraulic structures is often dictated to a great extent by site conditions, which may, for example, limit excavation or impose restraints on the layout and inter-relation of the various units of the plant. As a result, hydraulic design sometimes consists simply of 'streamlining' pre-determined shapes to conform with basic text-book rules.

This is seldom adequate, and in any case, despite the great theoretical advances in fluid mechanics made as a result of Prandtl's boundary-layer concept, it is still difficult in all but the simplest cases to predict the behaviour of water by calculation alone. In order to achieve a satisfactory design - either considering a general case or working within limitations of the kind mentioned - the designer has recourse to experimental scale models.

 Model Applications

Each type of hydraulic structure has its own flow requirements for optimum operation, and it is now common for model investigations aimed at defining these requirements to be written into the specifications for major contracts. The cost of such model testing is only a very small fraction of the total cost of the project and is normally more than repaid by freedom from troubles when the plant starts operating, by the avoidance of subsequent structural modifications, which are expensive in money and time, and by reduced operating costs.

In the power-generation industry, concentration on large units of the plant has resulted in rapid growth to a present unit size of 500 MW and a corresponding cooling-water requirement of about 250 000 gal/min per pump. Likewise, the world's ever-increasing demand for oil, and the resulting increase in the size of crude-oil tankers to about 150 000 tons deadweight at present, has not only raised problems in maintaining a full flow of oil when the cargo is pumped from the ships but has also necessitated large de-watering plant for the huge dry-docks built to service them. Throughout heavy engineering, the trend is towards bigger and better. 

The performance of pumps - particularly that of large-capacity low-head units, i.e. pumps of high specific speed - is influenced vary considerably by approach-flow conditions. The troubles that can arise from unsatisfactory intake-flow are illustrated in Figures 1 to 3 and can be summarised as follows:

a)      Mass swirl around the suction of the pump can result in severe overloading of the prime mover, serious erosion of pump parts, or both, depending on the direction of rotation (Fig. 1).

b)      Surface swirls in the neighbourhood of the pump inlet can develop into air-entraining vortices (Figs. 1 and 2) with consequent reduction of water output, unstable performance, increased mechanical wear and, of course, the introduction of air into the system.

c)      Bad distribution (Fig. 3), unnecessary acceleration and deceleration, and turbulence can all disturb the flow and cause unwarranted losses, so affecting the system head.

All these troubles can usefully be investigated by means of scale models, and so too, sometimes, can other troublesome aspects of pumping systems, such as silting (Fig. 4).

Principles of Scaling

When using a scale model to study flow, it is not possible to achieve conditions which are exactly similar in all respects to those on-site, but close similitude can be obtained by recognising the most significant factors in each case and applying scientific principles. 

Most pump intake designs are such that a free water surface exists over the greater part of the intake and gravity flow has the predominant influence. This is known as the Froude criterion. For surface form, overall flow conditions and depths or heads to be similar, the velocities in the model should then be reduced as the square root of the linear scale of the model. Thus for a 1/16th scale model, fluid velocities should be 1/4 of those on-site.

The effects of boundary walls and fluid viscosity on flow conditions cannot be ignored, however. If these were the predominant factors (the Reynolds criterion), and assuming water to be used both in the model and on-site, the similarity would be achieved by increasing velocities in direct proportion to the linear scale of the model, or the 1/16th scale model, fluid velocities should be 16 times those on site.

Clearly these two conditions are incompatible, and moreover, for the latter to be physically possible, very large scale models would be necessary. However, since full-scale flow conditions are invariably in the turbulent region, any scaling effect from viscous forces can be minimised by ensuring that the model, using Froudescale velocities, is large enough to maintain fully turbulent flow at all critical sections. These critical sections are determined not solely from the actual Reynolds Number, but by taking into account also the configuration of the channel. Surface roughness is also a factor, and to replace the concrete and steel finishes usual on-site, smooth walls of wood and perspex are commonly adopted in the model.

 Eliminating Vortices

Much has been written on the controversial question of which velocity scaling would ensure accurate prediction of the formation of vortices. Some authorities advocate equal velocities in the model and on-site, but in most models of practical installations, increasing velocities towards this target results in considerable distortion of flow conditions. In addition, when such features as submerged ports or mesh screens are present, the inherent gravity flow makes attainment of the target virtually impossible unless very large models are used.

A procedure that has proved reliable during the past fifteen years is to operate the model initially at Froude-scale velocities in order to establish the basic flow conditions. If swirling is evident at any of the critical sections, it can then be further explored by increasing velocities up to the point where flow conditions became distorted by wave effects, exaggerated losses, etc. Any increase in the severity of swirling during this exploration is used, in conjunction with analysis of the flow environment, to estimate the likelihood of vortex formation occurring on site.

All potential vortex regions are eliminated by a modification to the geometry of the intake or by other flow-governing devices.

 Ideal Intake Conditions

The application of model techniques to a particular design problem indicates the best approach to the problem and guides the designer s thoughts along the right lines. Over a period, the experience of similar problems is accumulated, so that eventually experience alone can provide a sound basis for designing new schemes of the same type. Actual model testing to investigate flow conditions in detail mayor may not be necessary, depending on the nature of the project and the extent to which it differs from previous schemes.

Experience has shown that, for pump intakes, ideal conditions would comprise individual approach flow to each pump, with a reasonably uniform distribution of velocity across the full width of the approach, and with no rapid expansion inflow area for a considerable distance upstream of the transition from the free water surface to pump inlet. Velocity should Increase progressively from source to pump suction, and the pump itself, if suspended 10 the flow channel, should no shed surface eddies in its wake. Submergence alone is not a universal solution to problems: Its effects are closely related to approach flow conditions.

Power-Station Cooling Systems

Each type of design has its own. special requirements, so that design is seldom straightforward. In circulating systems for power-station condensers, for example, the necessity for the plant to have high availability, and for the time required for maintenance or repair to cause the least possible disruption, introduces a 'flexibility factor' for which It can be extremely difficult to cater.

Most inland power stations now have to rely exclusively on the re-circulation of water through cooling towers. The possible permutations of condensers, cooling towers, and pumps can be appreciated. In planning such circulating systems, it is usual to design first of all for all likely operating conditions. The model test is then used in addition, to indicate any limitations on a combination of the various units under less likely operating conditions (where, normally, fewer pumps, towers, and condensers are in use and a choice is possible).

Power stations situated on the coast are not troubled to the same extent by this flexibility factor since the source of water is Independent of the discharge. However, screening of the supply water creates an extra problem in these intakes, because current designs of the travelling-mesh screen are not in sympathy with approach flow requirements.

 Return for Investment

The total savings resulting from the use of model investigations into design naturally vary with the application. However, the rectification of hydraulically unsound designs before construction work has started can save thousands of pounds' worth of actual reconstruction work and frequently even more by avoiding the associated delays.

Power consumption is of considerable importance in power-station work, where large-capacity pumps have reached efficiencies of 90% and can each absorb over 5 000 HP. Great emphasis is always laid on pump efficiency, but pump head requirements are often less closely defined and governed. A recent estimate that the capitalised value of savings in pumping head can be of the order of tens of thousands of pounds per foot of head explains the growing interest in system losses.

Hydraulic model techniques can be applied with advantage to a great variety of projects; the accompanying illustrations, selected from investigations undertaken by Drysdale and Co. Ltd. in connection with pumping projects, give some indication of this variety. There is no doubt that, in the past, shortcomings in the design of pumping systems have often caused much unnecessary expense through plant failures or inefficient operation. The increasing number of customers who make use of the experience and testing facilities available from the large pump manufacturers indicates a growing recognition of this fact. Today more than ever, those who design on the grand scale are reaping benefits from investment on the model scale.