The installation of non-metallic piping systems to handle process fluids, utilities and waste streams is an every day occurrence in the chemical processing industry. Many of these installations replace existing metallic piping systems that have failed due to internal or external corrosion. The design of these non-metallic systems generally parallels the process used in metallic pipe design but with an increased factor of ignorance (safety factor). And, unfortunately, most of the design criteria specified for such systems is based on a metal mind set, especially on retrofit or replacement type piping systems. There are various software packages that address stress analysis of piping systems based on physical properties and geometry input by the user. These packages do not calculate the local stresses at supports other than the general affects of longitudinal beam bending. This paper will review effective length design methods for the determination of local pipe stresses caused by externally applied support loads. Results will be compared with those predicted by one of the available finite element software packages.
Determination of local bending stresses, caused by support reaction loads on fiberglass piping systems, has been generally neglected because of the difficulty in calculating such stresses. Some factor of comfort has been realized by installing such systems according to manufacturers? published data (span charts, installation details, etc.) and generally recognized support methods (clevis pipe hangers, pipe shoes with encirclement saddles, etc.). However, this factor of comfort cannot be backed up with any data relative to the particular system as installed. In order to get a handle on the theoretical stress levels caused by externally applied support loads, several ring bending cases were superimposed to create effective length models that could easily be used to calculate theoretical maximum hoop bending stresses in pipe walls. These stresses plus those resulting from the internal pressure environment could then be used as a check to determine a numerical comfort factor. These results could also be compared to those generated by finite element analyses to determine model validity.
