Chapter and verse

Q: When is a guide not a guide?

A: When it is data.

Data, once measured and verified, should not change with time. But hard and verified data should not be confused with guidance. I revised Chapter 4 of the new CIBSE Guide C Reference Data. The revised Chapter 4 now contains both data and guidance. This chapter is used to predict the pressure drop along proposed pipework and ductwork so as to enable the designer to select the most appropriate size and the most appropriate pump and fan.

Let me explain. Much of the information in the revised Chapter 4 “Flow of fluids in pipes and ducts” is new. Most of it is pure data. Where data was unavailable members expect some guidance and so figures have been derived by extrapolation. This has been done purely because it gives “guidance” to users who have a life and do not wish to spend as much time as I have in plotting graphs, thinking about fluid flows, and eventually coming to a conclusion. CIBSE will be only too pleased for someone to verify some of this speculative guidance and convert it into data. In this revision (2007) I have tried to make the distinction between data and guidance clear.

Any observant reader of the guide will have noticed that much of the data in the latest edition has changed. It was, in short, a revolutionary change and the 2007 Chapter 4 bears little resemblance to that of 2001 edition. This process of change is nothing new, however, since those observant readers (with a good memory) will also recall that the 2001 revision to Chapter 4 bore no resemblance to what had gone before it either. Yet when we put together the 2001 edition we every reason to believe this newly published data would stand unchanged for many years.

So why the changes for 2007? Allow me to explain. For the 2001 version, in common with many guides, I used as a primary source Idelchik’s very comprehensive Handbook of Hydraulic Resistance. By the time I had completed the guide in 2001, I had found too many contradictions, misprints, and ambiguities for me to place real confidence in it. A senior researcher in France echoed these worries.

Then a CIBSE member, David Crowder, wrote bringing my attention to an equation by Haaland for pressure loss along pipes, which is so much simpler to use than the previous equations by Colebrook-White and Asthul. David had explored the accuracy of this equation over the whole turbulent domain of the Moody chart and found it to be very good. At a stroke, all the 66 pages of so-called “pipe-sizing tables” became obsolete. No need now for double manual interpolations to obtain approximate answers. Pipework and ductwork calculations are now carried out in an identical manner to one another. With the use of a spreadsheet, all the answers you want for pressure loss along pipes and ducts are instant and accurate.

With the pressure loss along straight pipe and duct runs resolved, the next challenge was pressure loss through pipework components. W Rahmeyer of Utah State University has published several papers since 1999 on proprietary elbows, bends and tees of different materials. He tested several components of malleable iron, forged steel and PVC-U and for each one tested a sample from different manufacturers. His results confirmed that the form and material of pipework bends affect the pressure drop. More importantly he showed that for bends there is a dependence on Reynolds number (which indicates the significance of a fluid’s viscous effect compared with the inertia effect) and for tees a dependence on size. Neither of these phenomena had been reported before.

More data was also available for ductwork components. After seven years of pushing, shoving, and detective work, I at last tracked down a copy of a report, commissioned by the Commission of the European Communities (Brussels), on pressure loss factors of ductwork components: Energy Loss Coefficients of Components Used in Air Distribution Systems. I’d heard it was 800 pages long, so it was worth the effort to find it.

The report was based on studies at four European research centres. Interestingly, i t is the first time, to my knowledge, that cross-checking had been carried out; two and sometimes three centres tested the same component. Only circular components were tested. The European programme was the first occasion where testing had consistently treated Reynolds number as a variable, and found its effects to be significant. Inevitably the data presentation in the new guide (2007) becomes more comprehensive. On the few occasions where the variation with Reynolds number had a small effect, mean values are presented for simplicity.

Soon afterwards I learned of complementary research commissioned in the USA by ASHRAE. Both these studies were for components of circular ductwork and to a degree they are complementary. The European programme tested components of only two sizes (250 and 400mm in diameter), but did so for bends of different angles and over a considerable range of Reynolds number, whereas the US programme covered a very large range of diameters, 150 to 1200mm, but presupposed no dependence on Reynolds number. Thus, much as I would have liked to have kept the guide simple, it has been increasingly difficult to do so for ductwork components. Values of the pressure loss factor z depend upon both Reynolds number (see Figure 1) and size (see Figure 2); making this data available necessitates more tables and larger ones.

Figure 2, derived from US research, shows the smallest duct size tested was 150mm diameter. For diameters less than 150mm, the curves are extrapolations. It is striking to note how tight bends of radius/diameter = 1.0 have appreciably higher pressure loss. It may seem obvious to say that a gentle bend will give less pressure drop than a sharper elbow, but sadly the standard design is for a relative bend radius of r/d = 1.0. The European project tested segmented bends of various bend radiuses from r/d = 0.7 to r/d = 5.0, see Figure 3. This shows that, whatever the Reynolds number, an optimum relative radius of curvature (r/d) of around 2.0 gives a minimum pressure drop. (American practice seems to favour r/d = 1.5.)

Some guides give data for bends of angles other than 90º. Idly doodling one day, I discovered they all gave the same proportion compared with the value for 90°, irrespective of the type of bend. My experience is that life is never so simple and convenient. I found the original source for this simplicity to be Idelchik. Fortunately the European project covered bends/ elbows of a sufficient number of different angles for an analysis to be made. In this instance the data did not vary greatly over the range Re = 1 x 105 to Re = 4 x 105 so taking mean values could be justified and the comparison can be regarded as reliable, see Figure 4. (The Idelchik relationship lay well above all three curves of Figure 4.)

These curves enable values of z to be obtained for any angle once the value for a 90° bend is known. The curves of Figure 4 may of course be different for bends of diameters other than 250mm and different again for bends of gentler curvature than r/d = 1.0, but we may hope the differences will not be large. Restricted as much of the data must be, it is still a major improvement on what went before.

When one component is closely followed by another, there is an interaction between the two; the total pressure drop will not be the same as the sum of the two in isolation. Surprisingly the pressure drop of the two in close proximity is frequently cited as less than twice the value of the single component. The European report found the opposite, namely that the effect of close coupling two bends was more than the sum of the two. Unfortunately this project only tested separations of up to five diameters, leaving much further work to be done. But the results were consistent. In the case of close-coupled bends of 75°, the results enabled the value for a single bend of 75° to be reliably estimated though in fact the single bend had not been tested. These extra points for the 75° bends in Figure 4 were essential.

Within the data of the very comprehensive European report there is much reassurance where different European centres had tested the same component. But sadly the results did not always agree. This must have been embarrassing for the coordinator, but presented me with considerable thought and analysis as to which to prefer and what data should be discarded. If I had chosen to reject all data having conflicting values from different research centres, I would have been obliged also to reject all previous data for which no attempts at corroboration had been made! The author has detailed many of these contentious judgements and decisions in the paper “The influence of Reynolds number and size effects on pressure loss factors of ductwork components”, BSER&T Vol 27 Issue 3, (2006) pp261-283.

In view of the amount of research carried out on circular components over the past 18 years, the amount of work needed to provide reliable data for rectangular ducts of various configurations is enormously more. The amount of earlier “original source” data for rectangular ducts is very sparse. Other data quoted in various guides is suspected to be “guidance” rather than “researched data”, some I am sure being based on early data for circular components and fudged using a “hydraulic diameter equivalent”. In the absence of reliable modern original source material we have had no option but to continue with the data of Idelchik. In short we are unable to place great confidence in figures for rectangular components, and have made this clear in the guide.

Some members may find disturbing the changes to habits and find the older inadequate guides easier to use. They should be reassured that every effort has been made to render the guide as simple as possible. The reduction of the 800-page European report ought to be persuasive. If in doubt, professional engineers should need no reminding that they have an obligation to make use of the most recent guidance. There may be a problem however in knowing precisely what data is concealed in the various software programs for pressure drop calculations. It may be some time before another major research programme is initiated, but when it is, I hope there will be a more positive plan to disseminate the results to those like CIBSE who would place it in the public domain. But Chapter 4 on fluid flow will always need revision as more research data becomes available.

Peter Koch, author of the newly revised Guide C Chapter 4, was formerly senior lecturer at Coventry University and Professor at IUT1, Grenoble University