Impact of Graveyard Shift
Anecdotally, we all think turnout times are increased during the graveyard hours. Crews are sleeping, need to dress, and have to travel from quarters to the responding apparatus, which might require traversing several flights of stairs or a fire pole. Several past studies [1, 2, 3] have confirmed our theory, some concluding that time of day has the greatest impact on response times, even more so than station design. In one study, across all station designs, the graveyard shift, defined as midnight to 6 AM, had increased turnout times between 50.8% and 58.9%.
Lets explore two methodologies on how to quantify the impact of graveyard runs within our own department.
Filtering For Graveyard Hours
A naive, but useful visualization is to simply calculate 90th percentile turnout for the two blocks of time
- Graveyard hours, midnight to 6 am
- Non-graveyard hours, 6 am to midnight
- Create a new Metric Visualization, using the incidents index
2. First, visualize the overall 90th percentile turnout time
a) Increase time range to last 1 year
b) On Metrics panel, change count to Percentile
c) For field, type durations.turnout.seconds
d) Remove all other percents, and keep one at 90
e) Add a custom label, 90th Percentile Turnout
f) Press play
3. Now, lets break the metrics into two buckets, one for graveyard hours and the other for the non-graveyard hours
c) On Buckets panel, click split group, and choose a range Aggregation
d) Change field to description.hour_of_day
c) Create the two ranges
- 0–5 (any incident opened between 0000 and 0559)
- 6–24 (any incident opened between 0600 and 2359)
d) Add a custom label, Hour
g) Press play
The final visualization, now shows two distinct metrics. In this case, we show that 90% of the time, the first unit turnout is 114 seconds when the incident is opened between midnight and 6 AM. Alternatively, when the incident occurs during non-graveyard hours, the first unit turnout time is 85 seconds, 90% of the time.
Turnout Time by Hour of of Day
A more advanced, but another common methodology, often found in Standards of Cover is to illustrate turnout times and call volume, by hour of day in a line chart.
- Create a new Line Visualization, using the incidents index
2. First, visualize the count of incidents by hour of day
a) Increase time range to last 1 year
b) On Metrics panel, change the label to Incidents
c) On Buckets panel, add X axis, select Terms aggregation, and add description.hour_of_day
d) Change Order By to Alphabetical
c) Change Order to Ascending
e) Increase size to 24, to account for 24 hours in a day
f) Add a custom label, Hour
g) Press play
3. Now, lets add 90th percentile turnout as a second Y Axis
a) On Metrics panel, click add Metrics, select Y Axis -> Percentile
b) For field, type durations.turnout.seconds
c) Remove all other percentiles, and keep one at 90
d) Add a custom label, 90th Percentile Turnout
e) On Metrics & Axes tab, expand the 90th Percentile Turnout section, and change Value Axis to New Axis
f) Press play
In either methodology, our suspicions are confirmed, from the hours of midnight to 6 AM, the 90th percentile turnout times increase from approximately 80 seconds to greater than 110 seconds, despite the decrease in call volume. In the second visualization, we gain additional insights — such as seeing the trend start earlier than midnight, closer to 10 PM.
How does your department turnout trend in the graveyard hours?
References
- Reglen, Dennis, and Daniel S. Scheller. “Improving Fire Department Turnout Times: Training v. Sanctions in a High Public Service Motivation Environment.”
- Reglen, Dennis, and Daniel S. Scheller. “Fire department turnout times: A contextual analysis.” Journal of Homeland Security and Emergency Management 13.1 (2016): 167–189.
- Upson, Robert, and Kathy A. Notarianni. Quantitative evaluation of fire and EMS mobilization times. Springer Science & Business Media, 2012.
