This study aims to understand the dynamic behaviour of sprinkler piping systems under seismic floor excitation to provide grounds for a better formulation of the procedure used for estimation of seismic demands and to emphasize the need for establishment of seismic design criteria consistent with the target seismic performance in the New Zealand Standard (NZS) 4541: Automatic Fire Sprinkler Systems. Hospitals, malls and apartment buildings in New Zealand were surveyed to identify typical features of sprinkler piping systems and these observations were complemented by input from experienced industry members to understand the state-of-practice. The collected information was utilized in designing and configuring specimens of sprinkler systems that were used in experimental testing and numerical investigations to address different aspects of seismic performance.

Different variants of brace assemblies typically used for restraining piping systems were tested under monotonic tensile and compression loading to investigate the potential for ductility capacity. Both brittle and ductile failure modes were observed, and the need for proper evaluation and documentation of ductility capacities has been discussed for use in design practice. The tests also provided evidence on the strength hierarchy within brace assemblies that has implications for the performance of anchors. Utilizing the test results and data provided in the catalogs by the manufacturers, it has been demonstrated that anchors can become vulnerable to failure if proper capacity design is not carried out.

The procedure for estimation of seismic demands in NZS 4541 does not properly account for the interaction between the supporting structural system and the piping system, which could lead to considerable underestimation or overestimation of demand. Sprinkler piping specimens were subjected to recorded floor motions using a shake table. The testing showed that seismic demands can be considerably higher if a piping system is in resonance with the supporting structure. Further, it was observed that displacements were three times higher for cases in which resonance occurred with the fundamental mode of the supporting structure in comparison to displacements resulting from resonance with higher modes. Six specimens of sprinkler piping systems were tested on a shake table to investigate the relevance of pipe diameter to brace spacing. Despite different diameters of pipes, the differences in periods of vibration among the specimens were not significant, and consequently, no specific trend could be identified between pipe diameter and seismic demand. It has further been demonstrated that the diameter of a pipe is important from the perspective of achieving a certain target performance, such as the avoidance of leakage, but should not be considered as the only variable in the selection of brace spacing.

To investigate the effects of plenum depth of the distribution pipe on the dynamic behaviour of a piping system, three specimens were tested with a plenum depth of the distribution pipe equal to 1000mm, and were then re-tested by reducing the plenum depth of the distribution pipe to 400mm. It was found that larger plenum depth of the distribution pipe can influence the period of vibration in systems with flexible pipes and no significant influence on period was observed in specimens with bigger diameters of pipes (e.g., 40mm vs. 100mm). NZS 4541 permits braces to be omitted if pipes are individually supported within 150mm of the structure with some ambiguous conditions. Shake table tests were conducted on two piping specimens with plenum depth of the distribution pipe in the specimens equal to 150mm and the entire specimens were supported by 10mm threaded rods. No damage was observed despite the recorded pipe accelerations being four times the limit set in NZS 4541 on the allowable design coefficient for such systems. The test results have been utilized to argue that the Standard should specify proper design limit states for the performance of hanger rods, avoidance of leakage and clearance rather than irrational design provisions that result in systems with an unknown performance.

The guidelines in NZS 4541 for the calculation of tributary seismic mass to estimate forces in the braces have not been validated. Braces installed on piping specimens in shake table testing were instrumented with load cells to measure the axial force demand during shaking. It was found that the addition of braces to a system could influence the force demand in other braces in the system. However, contrary to the understanding led by the tributary mass approach, comparison of the recorded forces showed that the addition of braces did not always reduce the force demand in existing braces. This was because the addition of braces modified the period of the system, which either increased or decreased the force demand in existing braces, depending on the change in the acceleration demand on the system due to the change in period. The recorded data was utilized to study the conservatism of the guidelines for tributary mass and it was found that the degree of conservatism varied with the location of the brace. Another finding from the analysis of the recorded forces was the contribution of hanger rods to seismic resistance; this finding is significant in that these rods are just provided for gravity support and are not designed for seismic resistance. Recommendations have been made for obtaining a conservative estimate of force demands in braces on different segments of a piping system.

Comparison between the periods of specimens obtained from numerical models and those calculated based on the recorded experimental data showed that the numerical models under predict the periods of vibration. Such differences in periods led to differences between the experimental recordings and the numerical predictions of different response parameters. In particular, the numerical models consistently under predicted the force demand in braces on the distribution pipes. The differences in forces have been attributed to the differences in periods and the inability of the numerical models to account for the high frequency effects due to the local vibrations of the pipe, and due to the segmental nature of the brace assemblies that caused impact forces. Further, reliability of the equivalent static force approach in estimating different demands has been studied, and recommendations have been made for a conservative estimate of different demands. Dynamic and static analyses of a hypothetical sprinkler piping system revealed the contribution of higher modes to seismic demands. It was further found that response spectrum analysis predicts various response parameters quite closely to the same obtained using response history analysis. Being more suitable for practical use, this finding encourages the use of response spectrum analysis for the estimation of seismic demands for such systems. Guidelines for the estimation of seismic demands for sprinkler piping systems using hand-based calculations have also been provided to facilitate practicing engineers in case the facility of numerical modeling is not available.