Fluid-structure interaction (FSI) occurs in all fluid-carrying pipe systems. More often than not, its consequences are minor and may be neglected for practical design purposes. In some cases, however, FSI can be responsible for unacceptable operational behaviour - e.g. sustained or sudden noise - or the failure of the pipe or its support system. The challenge is to know whether "your" system is (or will be) one of the benign majority or one of the troublesome minority. This website might help you to decide.

The underlying mechanisms of FSI in pipes are well understood and, in principle, the phenomenon can be analysed with good accuracy. Unfortunately, there is an important proviso, namely that the actual accuracy is strongly dependent upon the correctness of the input data defining geometrical and physical properties of pipes, valves, pumps and supports, etc. In reality, it is rare to be able to prescribe these with high accuracy in a practical application. In fact, it is even rare to be able to do so in controlled, laboratory experiments.

The overall aim of this website is to help you to move closer to deciding whether your pipe system might be susceptible to FSI. If you conclude that it is at serious risk, you will need expert help (see "Sources of Advice"). In the majority of cases, however, we hope that the site will enable you to conclude that a visit to the doctor is not necessary.

The site contains various levels of self-help. At one extreme, it provides simple guidance that should be sufficient to enable most readers to reach a clear conclusion. This guidance is based on descriptions of features that are likely to be found in many pipe systems. At the other extreme, it contains software that you may download free of charge. You may use the software to begin to quantify the ideas set out in the qualitative descriptions.

Flow-Induced Vibrations

The term "Fluid-Structure Interaction (FSI)" is sometimes used interchangeably with "Flow-Induced Vibrations (FIV)". On this website, the latter term is avoided and so are many types of vibration to which it relates. In particular, no account is taken of interactions between pipes and external flows. Attention is focussed exclusively on interactions between pipes and fluid flows within them.

Can FSI be a "good" thing?

FSI is nearly always regarded as a "bad" thing - i.e. as a source of trouble - and this website is no exception. For the record, however, it can also be a very good thing. One example is the hydraulic ram pump. This can give decades of service in water supply with little or no maintenance - not bad for a device that doesn't need an external power supply!

0.1 FSI and FIV

Some interactions between fluids and pipes occur in steady flows whereas other exist only in unsteady flows. Likewise, some occur with flows inside pipes whereas others are associated with flows outside pipes. A typical example of the latter is the vibration of pipes (or pipe groups) in a cross-flow.

Most of the FSI phenomena considered in this website are interactions between pipe movement and the unsteady flow of a fluid within the pipe. The initial cause of the interactions may be a disturbance to the flow (e.g. valve or pump activity) or a disturbance to the pipe (e.g. vibrating machinery). In both cases, the resulting dynamics will involve an extensive collection of pressure waves in the fluid and stress waves in the pipe wall. Whenever one of these reflects at a support or a bend, etc, it induces new waves of both types. That is the interactions are inherently two-way.

When the term FIV is used in this website, it usually involves an interaction that is dominantly one-way. These occur in special cases of steady flow, but they are rare in unsteady flows.

These uses of the terms FSI and FIV are consistent with much of the general literature, but certainly not all. Take special care with all literature using these terms. It is easy to be misled. One reason for the potential confusion is that there are many different types of interaction between fluids and pipes. The following figure (given by Blevins, 1990) illustrates the breadth of possibilities - and also categorises them in an intuitively elegant manner.

0.2 Time-domain and frequency-domain

There is a tendency for experts in FSI to classify themselves as specialists in either "time-domain" or "frequency-domain" representations of the various phenomena. Only a small minority are truly fluent in both of these. So what are they?

Broadly speaking, a typical outcome of a time-domain analysis may be a series of graphs showing how parameters vary in time (see figure 1). This is an intuitively obvious way to model real phenomena and it is often an appropriate way too. However, it is not always the best.

A typical outcome of a frequency-domain analysis might be a series of graphs highlighting the dominant frequencies in the response of various parameters (see figure 2). This is a valuable representation because most responses are composed primarily of a small number of dominant frequencies. Such analyses can tell us very easily whether our pipe is likely to respond in a lively manner to expected stimuli.

Mathematically, time-domain and frequency-domain analyses contain the same information. It is possible, for example, to obtain frequency-domain results from a Fourier analysis of the output from a time-domain analysis. Likewise, it can be possible to reproduce time-domain behaviour from a frequency-domain analysis (although this is less common because most frequency-domain analyses intentionally discard information about initial conditions). Wavelet analyses combine time and frequency domains.

Figure 1: A typical time-domain output

Figure 2: a  typical frequency-domain output