Published on: 

Rainer and I don’t often disagree on turbomachinery related technology issues.  But we could not reach a consensus on the topic of surge control valves.  Therefore, this one is a “Rainer says – Klaus says” type column.

Rainer is of the opinion that two-surge control valves, one primary surge control valve and one hot gas bypass valve (also often called a fast stop valve) are a more robust solution in many instances.

Klaus believes that the latest advances in transient analyses and improved design of surge control systems can erase the headache of having a separate hot gas bypass in most cases.  So let’s have at it.

Rainer’s argument

Centrifugal compressors are protected from surge by an anti-surge control system. The control system operates one or multiple valves that can bypass or recycle gas. The science behind designing these systems is well understood. Reliable design tools are available and in use.

What is frequently not discussed is that an anti-surge control system has at least three functions, and these functions have different requirements regarding valve characteristics, valve size, and noise attenuation.

The three functions are Start, Process Control and Shut Down. Starting the compressor usually requires the valve to be at a fixed position until the compressor comes online. A cooler in the loop is often advantageous.

Process control requires a relatively small, equal percentage valve, allowing precise positioning. If the valve is too big, the control function cannot be met. It is often advantageous to have a cooler in this recycle loop to avoid overheating.

Noise attenuation might be required.  Shutdown or particular emergency shutdowns require a very fast (and relatively large) valve, and piping with a small volume. Neither a cooler, nor noise attenuation are necessary or desirable.

We can conclude that it is usually easier to combine the start and process control functions into a single valve, but the emergency shutdown function will be more difficult to combine with either function. In reality, there are many compressor stations where all three functions can be covered with a single valve, but many of them do not have coolers. In all cases, a finely tuned compromise between the requirements for process control and emergency shutdown has to be found.

Since valves require long lead items in many projects, the decision about number, size and type of the valve has to be made early. Tools are available to make this choice, although experiences from past projects often guides the decision.

Since the piping layout of the station is typically not known early in the project, the tools have to be simple, and accuracy should not be overestimated. The valve size is essentially determined by the requirement to control the compression system, and thus depends significantly on the aerodynamic behavior of the compressor.

The capability to avoid surge in an emergency shutdown can be determined by simplified estimating tools. Two fundamental ways are used: Non-dimensional numbers, taking into account piping volumes, inertia and compressor power; and a more consistent way, using a “low fidelity” dynamic simulation, which combines the physics of the system, usually in a lumped representation with experimental data. The shutdown behavior of the driver or the opening characteristic of the valve are often known, and have to be considered. Such an analysis can identify whether a single valve solution is even feasible.


Simplified lumped parameter tools give a reasonable amount of information about system behavior to decide how many valves are necessary early in the project. Non-dimensional number models are not sufficiently accurate for this analysis. More importantly, having separate valves for fast shutdown and for control will in many instances provide a system that can be controlled more accurately.

The Klaus view

Compressor manufacturers require that surge control systems keep a compressor out of surge, even short transient surge during a fast emergency shutdown (ESD)-type shutdown. With increasing compression ratios and more demanding operating conditions, the use of hot gas bypass valves to avoid compressor surge during ESD events has become common.

But most operators dislike a second hot gas bypass valve in their surge control system because of added complexity and maintenance headaches. These valves can be expensive and extra piping is problematic when space is at a premium. In most cases, they are installed due to caution and design conservatism. They could be eliminated if the rest of the surge control system piping, check valve placement, valve selection, and controller design is done properly.

Hot gas bypass valve avoidance is effectively the same as transient ESD surge avoidance since the only time the hot gas bypass valve is used is to avoid transient surge during a fast shutdown. A system design that avoids a hot gas bypass valves requires a surge control system that avoids transient surge during fast shutdowns.

This can be achieved with a carefully designed primary surge control system.  But it cannot be achieved with a basic analysis using simplified rules of thumb, non-dimensional number models, or basic lumped volume models.

Typical surge system designs are usually based on a minimum design surge margin and a 10% surge control line using compressor flow and head. If a dynamic analysis is not accurate within a fraction of that percentage, it is not useful in determining if the compressor will go into surge during an ESD or if a hot gas bypass valve is required.

Neither non-dimensional number or low fidelity lumped volume approaches are accurate enough since their inherent uncertainty is higher. The accuracy of these approaches is too low to determine whether a single or dual valve system is needed. They are also of no use in determining if a system is susceptible to dynamic surge.

But a properly implemented full time-domain finite element 1-D dynamic compressor analysis, calibrated with real test data can provide modeling uncertainties as low as 3 to 5%. These tools have improved over the last 10 years and have been validated with test data.

Even at those uncertainties, analysis results have to be conservatively interpreted to make reasonable design decisions. But the potential exists for proper surge system engineering decisions. If an operator desires to implement a surge control system that eliminates the need of a hot gas bypass valve, a full 1-D transient finite element surge dynamic analysis needs to be performed. Otherwise the approach outlined by Rainer is the safe path.

Both Rainer and Klaus agree that a lumped parameter model is a good first screening tool but often a more detailed analysis is required.