Duty cycling refers to the practice of operating the Ultrasonic Personal Air Sampler (UPAS), or another air sampler, such that the pump turns on and off at regular intervals throughout the sample (e.g., on for 15 s and then off for 15 s) instead of remaining on continuously.
Users who choose to run the UPAS on a duty cycle typically do so because they want to accomplish one of the following:
You might ask, “Why not just reduce the duration of the sample instead?” Reducing the sample duration can also prevent the UPAS battery from depleting or the inlet and filter from becoming overloaded before the sample is complete; however, you might want to characterize pollution associated with events that occur over a certain time period (e.g., 24 or 48 hours), and you might not be willing to sacrifice a portion of that period.
You might also ask, “Why not just reduce the sample flow rate?” Reducing the sample flow rate can also prevent the UPAS battery from depleting or the inlet and filter from becoming overloaded before the sample is complete; however, the flow rate of a PM sample cannot be set arbitrarily. A PM sample is typically collected using a size-selective inlet (i.e., a cyclone or impactor) that select particles in a specific size range (e.g., particles smaller than 2.5 micrometers, also called PM2.5) based on the inertia of those particles. Such inlets are designed to operate at a very specific volumetric flow rate (e.g., 1 or 2 L min−1) and must be operated at that flow rate to sample particles in the expected size range.
The filter sampling pump in the UPAS can operate on duty cycles ranging from 13% to 100%. The duty cycle operates on a 30-s interval. For example, if a duty cycle of 50% is selected, the pump will turn on for 15 s and then off for 15 s, repeatedly, until the sample ends. If a duty cycle of 20% is selected, the pump will turn on for 6 s and then off for 24 s, repeatedly, until the sample ends. If a duty cycle of 100% is selected (the default setting), the pump will run continuously until the sample ends.
Yes, the piezoelectric pumping elements in the UPAS reach the target flow rate within milliseconds of turning on. Example volumetric flow data from a UPAS v2.1 PLUS sampling at 1 L min-1 on a 50% duty cycle and logging data on a 1-s interval are shown in Figure 1. The data in Figure 1 reflect actual air flow rates measured during each 1-s log interval by the mass flow sensor in the UPAS. The measured mass flow rates have been converted to volumetric flow rates using the air temperature, pressure, and relative humidity values measured by the UPAS. You can see that the transition between zero flow and the target 1 L min-1 sample flow rate that occurs every 15 s is abrupt. When the pump turns on and the flow rate transitions from zero, a flow rate of 1 L min-1 is measured the very next second. Similarly, when the pump turns on and the flow rate transitions from 1 L min-1, a flow rate of zero is measured the next second.
Note: We do not recommend logging UPAS v2.1 PLUS data at 1-s intervals in the vast majority of applications. The 1-s log interval produces large files that are difficult to work with and reduces UPAS battery life substantially. We recommend logging data at 30-s intervals in most applications.
Yes! The agreement between PM2.5 concentrations derived from filter samples collected using UPAS v2.1 PLUS that sampled at 2 L min-1 with a 50% duty cycle, UPAS v2.1 PLUS that sampled continuously (100% duty cycle) at 2 L min-1, and Personal Modular Impactors (PMIs) connected to Casella Apex2 Pro personal sampling pumps that sampled continuously (100% duty cycle) at 3 L min-1 is illustrated in Figure 2.
The results shown in Figure 2 are from a series of eight tests that were completed by Mattie Morris—an environmental engineering student at Colorado State University who was one of our 2024 Energy Institute Cogen Summer Interns. During each test, sodium chloride aerosol from a nebulizer was sampled from inside a laboratory enclosure for between 4 and 15 hours by two UPAS v2.1 PLUS that sampled at 2 L min-1 with a 50% duty cycle, two UPAS v2.1 PLUS that sampled continuously (100% duty cycle) at 2 L min-1, and two PMIs connected to Casella Apex2 Pro pumps that sampled continuously (100% duty cycle) at 3 L min-1. The time-averaged concentration in the enclosure was varied between tests by varying the frequency at which pump supplying air to the nebulizer turned on and the duration for which that pump remained on each time. The air inside the enclosure was mixed by four small computer fans mounted in a rectangular pattern on the inside of the top surface. All samples were collected onto 37-mm-diameter polytetrafluoroethylene (PTFE) filters (MTL PT37P-PF03). Each x- and y- data point shown in Figure 2 represents the average of the duplicate samples collected with that sampler type.
The mean difference, averaged across all eight tests, between the mean time-averaged PM2.5 concentrations measured using the two UPAS that sampled with a 50% duty cycle versus the two UPAS that sampled continuously (100% duty cycle) was 5%. The mean difference, averaged across all eight tests, between the mean time-averaged PM2.5 concentrations measured using the two UPAS that sampled with a 50% duty cycle versus the two PMIs connected to Casella Apex2 Pro pumps was -4%.
The data shown in Figure 2 were collected using UPAS v2.1 PLUS only, but the pumping electronics and hardware, as well as the algorithms used to control pump flow and duty cycling, are the same across the UPAS v2.0, UPAS v2.1, and UPAS v2.1 PLUS. Therefore, we consider these results to be representative of duty cycling performance for all UPAS models.
A duty cycle has been employed in numerous studies in which particulate matter was sampled with the UPAS, the largest of which was the PURE-AIR study (Shupler et al., 2020). In the PURE-AIR study, one in-kitchen PM2.5 sample was collected in each of 2541 homes. Additionally, one personal PM2.5 sample was collected for each of 998 participants recruited from 696 of those homes. Each of these 3539 UPAS PM2.5 samples was collected for 48 h at 1 L min-1 with a 50% duty cycle.
Karaca et al. (2024) also operated the UPAS at 1 L min-1 with a 50% duty cycle when collecting 48-h samples of PM2.5 pollution in the kitchens of Kazakh homes.
Koehler et al. (2023) sometimes operated the UPAS with a 50% duty cycle when collecting week-long samples of PM2.5 pollution in Baltimore-area homes at 1 L min-1. Koehler et al. employed the 50% duty cycle “to reduce overloading of filters in homes where high PM2.5 concentrations were anticipated due to known smoking in the home.”
Alli et al. (2021) operated the UPAS at 1 L min-1 with a 50% duty cycle when collecting week-long samples of outdoor PM2.5 pollution at stationary sites in Accra, Ghana as part of the Pathways to Equitable Healthy Cities study. Alli et al. employed the 50% duty cycle “to avoid overloading filters and to also conserve battery power.”
The duty cycle can be selected in the mobile application. In the CSU UPAS app (available for Apple and Android devices) that controls the UPAS v2.0, this setting appears in the Main Menu under the name “Duty Cycle.” In the AST UPAS app (also available for Apple and Android devices) that controls the UPAS v2.1 and UPAS v2.1 PLUS, this setting appears in the Main Menu under the name “Filter Flow Duty Cycle.” The default setting is 100% (i.e., continuous pump operation), but duty cycles between 13% and 100% can be selected in 3% or 4% increments. For example, duty cycles of 46%, 50%, and 53% are available.
The time-averaged particulate matter concentration measured during a sample, PM (μg m-3), is calculated as PM = m/V, where m is the mass of particulate matter accumulated on the sample filter (μg) and V is the volume of air sampled through the filter (m3).
The total volume of air sampled through the filter can be found in the header portion of the UPAS log file, in the SAMPLE SUMMARY section, next to the label “SampledVolume.” The SampledVolume is reported in liters in the UPAS log file. Divide the SampledVolume in liters by 1000 to convert to cubic meters.
Example data from a UPAS v2.0 running firmware version 138 are shown below. This UPAS sampled air through a PM2.5 1 L min-1 inlet, using a 50% duty cycle, for 10 minutes. The following values are from the SAMPLE SUMMARY section of the log file header.
| SampledVolume | SampledRuntime | LoggedRuntime | AverageVolumetricFlowRate |
| 5 | 0.084 | 0.164 | 0.997 |
The SampledVolume is 5 L, because the UPAS ran at 1 L min-1 for 10 minutes with a 50% duty cycle. Because of the 50% duty cycle, the UPAS only sampled for half of each 30-s period. As a result, the SampledVolume is 50% of the air volume that would have been sampled if the pump ran continuously at 1 L min-1 for 10 minutes. In other words, the SampledVolume in the log file header accounts for how the duty cycle affects the volume of air sampled.
The SampledRuntime is 0.084 hours, or 5 minutes. The UPAS ran for 10 minutes with a 50% duty cycle, so the UPAS only sampled for half of the 10-minute period. The LoggedRuntime is 0.164 hours, or 10 minutes.
The AverageVolumetricFlowRate is 0.997 L min-1. This value reflects the average volumetric flow rate while the pumps were on. In other words, this value does not account for the 50% duty cycle.
The data shown in Figure 3 are from the SAMPLE LOG in the same file. The mass and volumetric air flow rates recorded at each 30-s log interval (“MassFlow” and “VolumetricFlowRate,” respectively) reflect the values time-averaged over the portion of that interval when the pumps were on. In other words, these values do not account for the fact that the pumps were off for a portion of each 30-s interval. In contrast, the cumulative volume of air sampled (labeled “SampledVolume” in the SAMPLE LOG) does account for the fact that the pumps were off for a portion of each 30-s interval. In the example shown in Figure 3, 5 L of air is sampled while operating at 1 L min-1 for 10 minutes with a 50% duty cycle.
Example data from a UPAS v2.1 PLUS running firmware version 157 are shown below. This UPAS sampled air through a PM2.5 1 L min-1 inlet, using a 50% duty cycle, for 10 minutes. The following values are from the SAMPLE SUMMARY section of the log file header.
| OverallDuration | PumpingDuration | OverallFlowRateAverage | PumpingFlowRateAverage | SampledVolume |
| 0.173 | 0.086 | 0.484 | 0.969 | 5.02 |
The OverallDuration is 0.173 hours, which reflects the total time for which the UPAS ran. The PumpingDuration reflects the total time the pumps were on during the sample; that’s why, for this sample collected with a 50% duty cycle, the PumpingDuration is 50% of 0.173 hours = 0.086 hours.
The OverallFlowRateAverage is 0.484 L min-1. The OverallFlowRateAverage value is the effective sample flow rate that results from using the selected duty cycle; that’s why, for this sample collected with a 50% duty cycle, the OverallFlowRateAverage is 50% of the 1 L min-1 flow rate required by the inlet. When the pumps were on, the UPAS sampled at an average flow rate of 0.969 L min-1 (the PumpingFlowRateAverage value), but the pumps were only on 50% of the time, so the OverallFlowRateAverage was 0.484 L min-1.
The SampledVolume is 5.02 L, because the UPAS ran at 1 L min-1 for 10 minutes with a 50% duty cycle. Because of the 50% duty cycle, the UPAS only sampled for half of each 30-s period. As a result, the SampledVolume is 50% of the air volume that would have been sampled if the pump ran continuously at 1 L min-1 for 10 minutes. The SampledVolume in the log file header accounts for how the duty cycle affects the volume of air sampled.
The data shown in Figure 4 are from the SAMPLE LOG in the same file. The mass air flow rate recorded at each 30-s log interval (“MassFlow”) reflects the value time-averaged over the portion of the interval when the pumps were on. Similarly, the pumping volumetric flow rate recorded at each 30-s log interval (“PumpingFlowRate”) reflects the value time-averaged over the portion of the interval when the pumps were on. In other words, these two values do not account for the fact that the pumps were off for a portion of each 30-s log interval. In contrast, the overall volumetric flow rate recorded at each 30-s log interval (“OverallFlowRate”) reflects the value averaged over the full 30-s interval. In other words, this value does account for the fact that the pumps were off for a portion of each 30-s log interval. The cumulative volume of air sampled (labeled “SampledVolume” in the SAMPLE LOG) also accounts for the fact that the pumps were off for a portion of each 30-s interval. In the example shown in Figure 4, 5 L of air is sampled while operating at 1 L min-1 for 10 minutes with a 50% duty cycle.
The data included in this post were analyzed in R. UPAS data were read into R using our astr package, which you can install from GitHub!
Read more about how to set UPAS up to sample on a duty cycle and how to interpret the data from UPAS log files in the User Guide for the UPAS v2.0 or the UPAS v2.1 and v2.1 PLUS.