Home | IFS Content | History| Purchase

Measuring Class 1 Loss Coefficients

Objective

• To provide information on the experimental studies that generated much of the Class 1 loss coefficient data in Internal Flow Systems.

Introduction

Major experimental programmes were undertaken in the 1960s and 1970s at the British Hydromechanics Research Association (BHRA now BHR Group Ltd) to provide design data for large flow systems. The programme built on 20 years of prior work at BHRA and was followed in the 1980s by further studies related to compressor stations for the US and Canadian gas distribution systems.

The programme started with literature reviews (1-3). These covered bends, diffusers and combining and dividing junctions. Although not exhaustive, over 700 references were found related to these components. Conclusions from the literature reviews included:

.

Figure 1 Bend Loss Coefficients

Figure 1 taken from a review in 1945 illustrates the wide variation in published loss coefficients for 90 degree bends - the most prevalent component in piping systems. Since 1945 many more studies of 90 degree bends have added to the confusion. Clearly, researchers have not tested geometrically similar components under the same flow conditions and measured the same parameters.

Fluid flows are extremely complex with events tens of pipe diameters upstream affecting flow conditions at inlet to a component and, therefore, flow behaviour within a component and after a component. Pipe surface roughness has a significant effect on velocity distribution entering a component and the pressure loss caused by a component. The internal dimensions of commercial piping components, such as bends and T junctions, are nominal dimensions. Components with the same description but from different manufactures can have substantially different internal geometries. . Many published loss coefficients relate to tests on commercial components whose internal geometry was not measured and recorded. Steps and discontinuities between pipes and components are common when systems are assembled from commercial components so it is highly likely that many published loss coefficients include losses due to steps and discontinuities.

Minor surface faults on the flow surface of a pipe or duct at the location of a static pressure tapping can cause static pressure errors that are as large as the static pressure drop caused by a component. Unless a pipe wall is hydraulically smooth and a pressure tapping is expertly made errors of plus or minus 5 to 25% of the dynamic pressure are not unusual. Pipe diameters in many studies did not allow access to where pressure tappings broke through the pipe wall so it is likely that burrs or other faults existed.

Accurately establishing flow rates is extremely demanding. Few experimenters meet the internationally accepted standards for flow meter instalations. One of the major failures is insufficient inlet pipe length. Most experimenters do not provide sufficient information to make a judgment as to the validity of the flow measurement method.

A few research groups around the world, that had been active for five or more years, developed the expertise and assembled the equipment and systems to generate reliable loss coefficient data for particular components. Loss coefficients were mainly for Reynolds numbers up to 0.5 x 10E6. BHRA test Reynolds numbers typically started at 0.5 x 10E6 allowing loss coefficient comparison with results from these reliable sources. Good agreement between loss coefficients from the two sources provided confidence in classifying lost coefficients as Class.

.

Definitive Loss Coefficients

In today's terminology BHRA's approach to establishing definitive loss coefficient was one of "due diligence". Many conditions where established that had to be satisfied including:

• Components with geometries that could be uniquely described by a number of non-dimensional parameters such as area ratio, bend radius ratio and cross-sectional shape
• Pipes, ducts and component flow surfaces that were hydraulically smooth - This was not achieved for some pipes at Reynolds numbers above 10E6 as minor departures from smooth pipe friction coefficients were observed
• Pipe and ducts mating with components so there were no steps or other discontinuities
• Testing with a sufficient length of inlet pipe or passage to provide developed flow conditions at a component's inlet
• Testing with pipe or passage lengths after a component to re-establish a developed friction pressure gradient over a sufficient length to accurately establish the pressure gradient
• Debiting components with pressure losses that occurred in an outlet pipe or passage that were over and above developed flow friction losses in the pipe or passage at the flow Reynolds number
• Two independent means of flow measurement with repeated calibrations
• Static pressure tappings individually checked - It was not always possible to establish what caused some tappings to read high or low and these tappings were blanked off
• Static pressures recorded using a scanivalve system with high precision betz manometers used as a cross check on each scan sequence.
• Numerous cross checks and repeat testing of components

Experimental Studies at BHRA

Working Fluid

The BHRA studies were designed to generate definitive loss coefficients over a Reynolds number range of 0.5 x 10E6 to 1.2 x 10E6

• Relatively light weight but rigid pipes and ducts that could be easily configured to test single and combinations of components
• Test components could be manufactured at reasonable cost to close dimensions and with hydraulically smooth flow surfaces
• Pipes, ducts and components sufficiently large to provide access to most pressure tappings.
• Simpler measurement systems than with water
• Extensive available measuring technologies developed for aerodynamic research.
• Ease of carrying out flow visualisation to relate flow behaviour to component pressure losses.
• BHRA had many years of experience in modeling large cooling water, hydroelectric and other internal flow systems using air models.

Cross - sectional Shapes

• Round, square and 2:1 rectangular cross sections
• A hydraulic diameter D of 12 inches (305mm) that allowed testing at Reynolds numbers of 10E6 without running into compressibility problems.
• Flanges drilled to templates to avoid discontinuities between sections.
• 8 and 14 inch diameter pipes when components involved changes in cross-sectional area.

Note: The hydraulic diameter D is the diameter for circular cross sections and for non-circular sections is given 4 x cross sectional area/perimeter. Square and rectangular ducts had 1 inch corner fillets that were representative of the corner fillets in concrete culverts.

.

Figure 2 Test rigs

Figure 2 shows two test rigs assembled, one with circular pipe and a 45 degree bend and the other with 2x1 rectangular duct with a 90 degree, aspect ratio 0.5, bend. Square duct is stored between the test rigs. The rigs as shown are configured to draw air through an elliptical inlet flow meter. The test rigs could be reconfigured to blow air through the test components and to carry out studies with dividing and combining flow.

.

Figure 3 Shows testing two r/d = 3 bends in a 90 degree combination angle

Figure 3 shows that by making the 12 inch pipe from reinforced fibre glass weight the pipes were rigid enough to be easily suspended for investigating component interactions.

.

Figure 4. Selection of test bends

.

Figure 5. Selection of diffusers and transitions

Further details of the test rigs and components tested can be found in references at the end of the Chapters 8 to 13 in Internal Flow Systems.

References

Zanker, K. J. and Brock, T.E., A review of the literature on the fluid flow through closed conduit bends. BHRA, TN 901. July, 1967.

Cockrell, D. J. and King, A. L., A review of the literature on subsonic fluid flow through diffusers. BHRA Report TN 902. November, 1967.

Crow, D. A. and Wharton, A., A review of the literature on the division and combination of flow in closed conduits. BHRA Report TN 937. January 1968.

Return

© D.S. Miller (See Permission to Use)

Page under development