Last month’s article discussed the state of building automation system (BAS) alarming in many buildings, and it’s not good. Building operators are far too often saturated with alarms, and the intended purpose of alarming is lost. Many of these alarms are nuisance alarms, meaning they’re going off unnecessarily. One of the primary reasons for the high quantity of nuisance alarms is the lack of understanding of the concept of latching.
Explanation
Alarms can be configured to generate when some unexpected condition is observed by a BAS. A latching alarm requires both the alarm condition to have gone away and acknowledgment from the building operator for it to be reset (i.e., go inactive). A non-latching alarm does not require such acknowledgement; if the alarm condition goes away, the alarm will automatically reset.
Some alarms require an automated response by the BAS, often to protect the system’s components from damage. Such alarm responses can be either latching or unlatching as well.
Examples
Table 1 provides several examples of alarms and, where applicable, alarm responses. Latching requirements for the alarm and alarm responses are identified as well.
In Table 1, Example 1 shows an alarm condition that does not warrant any alarm response. The normal automation of the system should be working to restore SAT back to set point. Even if the alarm condition persists, there is minimal risk to damaging the system.
Example 2 is an alarm often called a software freezestat. It has an alarm response to protect against the freezing of coils within the AHU. Such an alarm and its response are latched out of an abundance of caution to ensure a building operator is made aware their investigation is warranted. As can be seen with this example, if an alarm response is to be latching, its associated alarm needs to be latching as well. The acknowledgment of the alarm will release both the alarm and the alarm response and return the system to normal operation.
Conversely, an alarm can be latching and its associated alarm response be unlatching. Example 3 in Table 1 depicts such a scenario from a recent project. The fan coil unit (FCU) serves a critical space in a hospital. If the float switch in the condensate pan triggers, indicating insufficient drainage, the cooling valve will be closed to prevent the condensate pan from overfilling and an alarm will be generated. If the pan then drains enough for the switch to stop triggering, then the cooling valve is allowed to operate again as to mitigate any overheating in the critical space. Thus, the alarm response is non-latching. However, the alarm itself remains latched as to alert the building operator that he or she better go take a look as the situation is likely to repeat itself. Since the condensate float switch triggered even once, the condensate drain is probably partially obstructed, or there is an unexpected source of humidity in the space.
To further complicate matters, several types of alarms can be either software or mechanically latched. For example, high static pressure safety switches (Example 4) can come with a mechanical latching option. If static pressure in the supply duct exceeds its adjustable setting (e.g., 4 in. w.g.), then the switch will trigger, and the supply fan is disabled. When the fan disables, pressure in the supply duct dissipates, but the switch will remain triggered (i.e., latched) until the push button located on the front of the switch is pressed (see Figure 1).
Such high static pressure safety switches can also come with an auto resetting option. When such a switch triggers due to excessive static pressure, the BAS will monitor that switch, generate an alarm when the switch triggers, disable the AHU (to include the supply fan), and latch both the alarm and alarm response. When the supply fan disables, the pressure dissipates and the switch will reset itself, but the alarm will remain latched due to logic programmed on the BAS. In this scenario, the building operator can reset the alarm remotely via the BAS. This approach of not forcing a building operator to physically inspect the unit carries risk, but it is desired/required by certain facility operation staffs.
The application of latching can get creative. Example 5 depicts an AHU Low SAT alarm, just like Example 2. However, on a recent project, the facility operations staff insisted on this alarm being non-latching as to avoid having to deal with acknowledging “nuisance trips.” The project team objected, stating the SAT dropping below 40°F is an indication of a real issue, and no such nuisance trips should ever be associated with this alarm. The project team advocated strongly for a latching alarm. The compromise between the parties was for the BAS to be programmed to latch the alarm and the associated alarm response only if the alarm triggered three times in a 24-hour period.
Whether an alarm and/or its alarm response is to be latched or not depends on a lot of factors. Those include but are not limited to:
- Nature of the alarm;
- Risk of damage to the system;
- Owner preference; and
- Likelihood of alarm repeating itself if the alarm response is unlatched.
Inappropriate Lack of Latching Leads to Alarm Saturation
There are certain situations that require a latching alarm and alarm response; otherwise, instability in the system can occur. Such instability may damage equipment and contribute to alarm saturation. This is best explained with an example. Figure 2 shows a condenser water system serving a water-cooled chiller. This system configuration allows for all suspended water in the system to be collected into the remote sump, located in a conditioned mechanical room, when the condenser water system is disabled. This avoids the need to seasonally drain the condenser water system in the winter, and the system can remain available for use year-round, should the need for cooling present itself.
The remote sump has a water level indicator that reports to the BAS. If the water level gets below a low limit threshold, this would indicate an issue with the makeup water valve or a system leak. An alarm will generate, and the condenser water pumps will be disabled to protect them from damage. However, this alarm condition will quickly go away as all the suspended water in the system will gravity drain back to the remote sump with the pumps off. If this alarm and its response were non-latching, the pumps would re-enable a few seconds after disabling only to have the water level drop and the alarm condition present itself again. Such instability would lead to thousands of pump cycles and alarm notifications over the course of a day. This scenario actually happened and is the poster child for alarm saturation. The remedy? Latching was added to the alarm and alarm response and investigation was performed into the root cause of the observed low water level.
Excessive Latching Leads to Alarm Saturation
Unnecessary latching of alarm and alarm responses will lead to alarm saturation as well. As described above, there are some alarms and alarm responses by their very nature that need to be latched. Such alarms make up a small portion of the alarms configured for a building’s HVAC systems. Imagine the chilled water system serving a building went down for several hours during a hot day. If all high-temperature alarms on the AHUs and in the served spaces were latching, the building operator may have to go and manually acknowledge a large quantity of alarms after chilled water service was eventually restored. This is a waste of time and a distraction from more pressing matters.
I would error on the side of keeping as many alarms unlatched as possible. That is not to say the alarms cannot be logged. They should still be logged in the history but should not remain active, as it is the active list of alarms the building operator is generally monitoring.
Conclusion
Taking the time during the design and construction phase of a project to evaluate latching requirements of both alarms and alarm responses can pay dividends when it comes to future building operations. The commissioning provider should seek to facilitate such discussions with project team members when the specified sequence is lacking or open for interpretation. Having a solid understanding of these concepts can help more efficiently clean up an existing BAS that is currently saturated with alarms as well.