Sampling in the face of variability

The concentrations of IAQ contaminants almost always fluctuate over time.  Therefore is it best to log data over time, whenever possible.  Technology, and more importantly, budget constraints may prevent us from logging data over time.  That means we must show up on a given day and measure that point in time.  This method gives us a random reading from a normal day.

If we only have one shot at taking a measurement, some advocate aggressive sampling to simulate the absolute worst case maximum scenario.  Let me give you some examples of trying to simulate the worst case maximum scenario:
  • Measuring carbon monoxide: Turn every single combustion appliance on at once.  Everything that creates a negative pressure in the home is turned on including kitchen and bathroom exhaust fans the clothes drier and the centralized vacuuming system.
  • Measuring airborne mold spores: A leaf blower blows air over carpeting, upholstered furniture, and in the ductwork.
  • Measuring VOCs: Although the cleaning crew normally cleans after hours, you have the cleaning crew clean an office while it is occupied.  You have the occupant wear the highest amount of perfume possible and wear dry cleaned clothes.  Oh, and don’t forget adding in an air freshener and shutting down the HVAC system.

Most of these extreme situations create a worst case maximum scenario that is unlikely to ever happen.  If it did happen, it would likely be for a very short period of time.

Is there some in-between ground?  Yes. We can adjust some of the variables, within reason, to simulate what I call a “worst case normal” scenario.  A worst case normal scenario is different from a worst case maximum scenario in that you are only adjusting variables within the parameters of normal building operation.

For example, this might mean having the commercial building be at minimum ventilation before measuring VOCs.  This would be accomplished by having the VAV system at minimum supply and ensure the outdoor intake dampers are at the minimum opening.  Or it might mean having the home owner close the windows and turn on the HVAC system before taking air samples for mold.

There are compelling arguments for either taking a random sample, which may be anywhere along the normal scale, or taking a worst-case normal sample by manipulating a few variables.  Which do you think is the best?  Please leave a comment on my blog!

3 thoughts on “Sampling in the face of variability

  1. Ian,

    I read this post with interest as it is a topic we frequently talk about at our office, especially since most of our IAQ guys have recently taken the BPI course (where we learned worst-case sampling for combustion gases).

    We rely on lots of data points over a one to three-hour test period. I recently collected eighty-five 1-minute readings for TVOCs, CO, CO2, temp, RH and PM10 resulting in over 200 data points. The first hour was during still conditions with no air movement. The next hour was with the HVAC fans in operation. The last hour was with the HVAC turned to a normal setting (it did come on once) and scuffing up carpet, tapping on upholstered furniture and moving clothing around in the closets. It’s amazing to chart out and sort the data in an Excel sheet – patterns emerge and I am often surprised by results.

    Add variables such as pets, children, popcorn, old furniture, books from Grandma’s house, etc., and it’s often a wild ride!

    No answers here, Ian – as IEPs we just need to do our best to collect meaningful data. Thanks for making us think!

    1. Craig,
      Thanks for your insight! I agree that the worst case scenario makes sense for carbon monoxide because the consequences are so extreme.
      Keep up the good work!
      Ian

  2. Interesting post, Ian. It really stands to illustrate the challenges of coming up with the truth, esp. in the context of weighing all variables. Building on that: even if one was to assume worst-case normal, as described in your post, as IEPs we are standing in a point-in-time as guests to the structure and not full-time occupants. What normal is is relative to the structure’s actual experience as full-time body. Compare that to full-time occupant response and you get a broader time/space relationship-expression (not always in scientifically measurable terms and conditions). It really makes one think, doesn’t it? Good post!

Leave a Reply

Your email address will not be published. Required fields are marked *


*