Published on February 8, 2021.

In Compliance | NFPA 72

Determining the extent of fire alarm acceptance testing


Some of the most common questions received by NFPA staff have to do with additions and modifications to fire protection systems. In today’s world of shell space buildings, increasing tenant turnover, and frequent building modifications both large and small, it’s no wonder that the systems providing fire and life safety must be constantly evaluated and tweaked to maintain compliance.

NFPA’s technical question service (TQS) recently received a question about the extent of testing that would be required when adding three initiating devices to a fire alarm control unit (FACU). In this case, the FACU was connected to multiple networked FACUs.

The correct place to start here is Chapter 14 of NFPA 72®National Fire Alarm and Signaling Code®, which includes criteria for initial acceptance testing, routine inspection, testing, and maintenance, as well as a section on reacceptance testing for fire alarm systems. The first and easiest answer to the question is that the three new initiating devices need to be functionally tested. This is stated clearly and concisely in the code, and you’d be hard pressed to find anyone who might argue with that.

For a conventional system where no other changes are made, including to the site-specific software, that might be all the testing that’s necessary. The way the question was asked, however, indicated that this situation involved an addressable fire alarm system and therefore required changes be made to the site-specific software. When this is the case, NFPA 72 requires more extensive testing. This will first require that all the functions known to be affected by the change be fully tested. In this case, any function activated by the three new initiating devices must be tested and could include notification appliances, elevator recall, pressurization, other relay functions, and annunciator labelling, to name a few.

In addition to functionally testing the three devices themselves, as well as all of the associated outputs, an additional 10 percent of initiating devices, up to a maximum of 50 devices, that are not affected by the change must also be tested in order to verify correct operation. This is because, while modern software-driven fire alarm systems are easily modified to accept new devices or to change the address or response of existing devices, these steps are often handled by changes to the site-specific software in the fire alarm control unit. That means that what may seem like a minor change may have unintended consequences to inputs or outputs not meant to be changed.

This is where the question submitted through TQS got interesting. As mentioned, the three new initiating devices were being added to a single FACU that was connected to several other FACUs. The submitter wanted to know if the additional 10 percent of initiating devices (up to 50 total) requiring testing was for the single FACU or for initiating devices connected to the networked FACUs. The impact of this determination was that where the three initiating devices were being added was part of a control unit that had a total of 50 initiating devices, whereas the total amount of initiating devices for the networked system was over 1,000. That meant the difference between testing an additional five initiating devices (10 percent of 50) versus an additional 50 initiating devices—a significant difference in time and effort.

To answer this, we needed to know the functions of the specific system in more detail than we were provided. If activation of devices that report to the FACU in question are limited to that control unit, then the 10 percent could be limited to the single control unit. If the activation of devices that report to the FACU in question transmit and cause functions to occur at other FACUs via network or other communication means, then those would need to be included as part of the 10 percent, or up to 50 devices.

Whichever situation is appropriate, the 10 percent sample should be selected randomly and include at least one device per initiating device circuit or signaling line circuit (SLC). If all the devices are installed on one SLC, multiple devices (a 10 percent sample) should be tested at different sections of the circuit. Use of comparison algorithms, or programs that compare two sets of data to determine the differences between them, can also help determine where program changes may have occurred, but are not an alternative to the testing.

With buildings constantly changing and the ease with which modern fire alarm systems can be modified, the need for this reacceptance testing is common. While routine inspection, testing, and maintenance should catch issues that arise and need to still be performed in accordance with the rest of Chapter 14 of NFPA 72, the correct operation needs to be confirmed at the time the changes are made, and that is what the short but important requirements on reacceptance testing are meant to accomplish. 

Jonathan R. Hart, P.E. is technical lead, fire protection engineering at NFPA. NFPA members and AHJs can use the Technical Questions tab to post queries on NFPA 72 at