Learning from Surprises III – Foam Fractionation

In previous blogs I discussed being surprised on at least two occasions.   One dealt with trying to determine the ventilation rate in a newly constructed experimental room of bare wallboard using C02 as a tracer gas.   The second covered the development of an indoor air model of a wood preservative.   These are available in previous blogs in this series.   This week I want to talk about what was perhaps the biggest surprise I have received to date.   It deals with the measurement of a very high level of airborne contaminant from a source that was not expected to present anything near the level actually measured.

The product of interest was lipophilic and slightly water soluble molecule (300 ppm w/w in water) which at 500 µg/m³ in air was considered to be irritating to the upper respiratory tract.   The product was being tested for its ability to treat thousands of gallons of water in a plant.   It was mixed into this large quantity of treated water systems at 5 ppm. 

Part of the testing was to have IH monitoring of the operators and various places within the plant where airborne exposure to the product might occur. 

Before the monitoring, I modeled the potential exposure and even considering a reasonably large activity coefficient,  I did not anticipate that we would have any airborne product above about 50 µg/m³.

As you might imagine I was very surprised when one of the results (and area sample) came back at approximately 600 µg/m³.   Indeed, the head-space concentration of the concentrated product was only about 900 µg/m³.    So we had a measured airborne concentration that was greater than 50% of headspace for concentrated product over an aqueous solution of the compound at only 5 ppm or  0.005%!

We double checked everything and found no mistakes.  I had to understand this result.  I was not present during the sampling but I trusted the IH that did the work.  I felt like I had no choice but to get on a 2 hour plane ride to the plant and reproduce the work done by the IH.    The single high concentration sample was taken in a man-way located above a sump which contained a majority of the treated water.

As soon as I saw the water in the sump and breathed the air above it, I started to understand what was happening.   There was a 3 inch layer of foam on top of the water.  If you put your head near the surface you could hear the bubbles popping much in the same way as you can here soap bubble pop in a bubble bath. 

The data and the above observation of foam provided a hypothesis as to what might be happening.  The foam was obviously a lipophilic or at least a surface active compound that was present in the water along with the compound of interest.   The compound, which has a relatively high octanol-water partition coefficient (see previous blog on Kow) was partitioning into this layer.   The foam bubbles were breaking or “popping” into the air above the foam.  As they did so they were releasing aerosol particles that were rich in the compound.    These resulted in a very high concentration of airborne compound both as an aerosol and as a vapor as the aerosol particles evaporated.

To test this hypothesis we sampled the foam.   It had concentrated the compound by over 3 orders of magnitude above the concentration in the water so that there were thousands of ppm of the compound dissolved within the foam.  

After doing some research on this phenomenon, I found it has a name:  foam fractionation.    I presented it here just to show how Mother Nature can through us some curves on occasion and when she does we get a chance to learn.

Questions for the LinkedIn Group:

Have you ever come across foam fractionation in your work?

What have been your biggest surprises in the realm of chemical exposure assessment?

Using IH MOD to Visualize Bolus Exposures

Two-weeks ago this blog  discussed the potentially very significant difference in toxic effect that could occur when what might be considered a relatively steady time-integrated exposure actually occurred as a bolus.     In the context of this discussion, we define time-integrated exposure as an exposure that happens at a relatively constant rate over a time period of hours.   A bolus exposure is defined in this discussion as a very high level exposure that happens over a period of seconds to minutes.   The point of the previous blog on this subject was that on those occasions that a bolus exposure occurred, but was measured as a time-integrated sample, that the average concentration might appear very tame compared to an 8 hour OEL while the actual exposure experienced by the worker could have been very high.    

My colleague Matt Le asked if we might be able to portray this difference in time graphically and if so what that portrayal might teach us.   I immediately thought of IH MOD as a tool that can readily help to visualize the time elements of inhalation exposure.  As such I set about to essentially reproduce last week’s Scenario 1 and Scenario 2.   In the first situation (Scenario 1), the breathing zone concentration of this compound is relatively constant throughout the 8 hour shift and results in a measured value 7 ppm after repeated daily sampling as an 8 hour time-weighted average (TWA).    In the second scenario (Scenario 2) the worker receives, for whatever reason, essentially no exposure for most of the sampling period but a bolus exposure to the compound that lasts only 2 minutes and also results in a measured integrated 8 hour TWA exposure of 7 ppm.

After firing up IH MOD it a pretty simple matter to simulate these two scenarios.

Scenario 1 is show below:

 Assuming it was indeed a constant source, 7 ppm was used as the starting concentration Co in the above model.   As you can see it is pretty much a flat line for both point-in-time and time-weighted average concentration out to the limit of the simulations which was 60 minutes.   The source as constant for the entire 60 minutes.

Scenario 2 used the same inputs (for room volume (V) and ventilation rate (Q)) as Scenario 1 but without a starting concentration.  The time of generation was set to 2 minutes and the generation rate was boosted to render a peak concentration slightly higher than 1600 ppm.  Indeed, the results would have been about the same if I had included 7 ppm as Co, it simply would have been washed out by the bolus.

Scenario 2 is below and shows the dramatic exposure potential for the 2 minutes in which the source was turned on.  Of course, the source was shut off after 2 minutes and the simulation allowed to run out for 60 minutes.  The time weighted average after 60 minutes was 50.1 ppm.   Running the model out for 480 minutes renders a calculated time-weighted average exposure of 6.8 ppm.

Other than providing a clear graphic picture of the differences in these two scenarios, that both represented 7 ppm as an 8 hour time-weighted average exposure, what did IH MOD tell us?   Well, the bolus dose (Scenario 2) required an intense short term (2 min) rate of contaminant input to the model that was over 230 fold higher that of Scenario 1.  The total amount of contaminant emitted during the bolus was about 8 fold higher than the total for Scenario 1 for one hour.  These ratios will vary with the modeling inputs selected but will always been a much larger value for the bolus exposure. 

Clearly bolus exposures represent unique and potentially critical events relative to worker exposure.

Questions for LinkedIn Group Discussion:

Do you have the potetial for bolus exposure in your workplace?  

If so how to you address them?

What do you think about having C or ceiling OELs for these possibilities especially for respiratory irritants or do you think the excursion limit of 5xOEL would be sufficient?

Exposure Modeling Data Base Needs

Quite a few years ago, Neil Hawkins and I wrote a few papers about the potential value of exposure modeling to the Industrial Hygiene community.   For those of you who know Neil, he is a very bright, creative and well-organized individual with an abundance of managerial acumen.  In short, he is great collaborator.  More important, we shared a belief in the utility of modeling.  In Neil’s case he had a vision of what needed to happen for the science to move forward.   I was in the weeds working on the details of the models while he was in the clouds looking at the big picture.  He started talking with me about the awareness and changes that needed to happen in the IH community in how we do our jobs in order for modeling to really take hold.

Indeed, if you want or need to evaluate an exposure you have essentially two ways to approach its assessment.  The first is very direct and the manner in which many of us were taught in school.  You measure the exposure directly and then compare that exposure to some toxicological benchmark.  To do this you needed a methodology and sampling protocol.

The other method, the method we were advocating in those days, was to conduct a model estimation of the exposure.   To do that you needed a model which had the following general form:

Predicted Exposure = f (the factors that caused the exposure)

Thus, your model always needed to be “fed” with the factors or values of the variables that were driving the exposure. 

Surprisingly, at least to me at the time, was the fact that most of the factors that were needed to feed our models were not commonly available.   The very entities that were driving the exposures to occur were not being captured and studied.  

This is not to say that there was not any exposure assessment data available.  Indeed, there were literally hundreds of thousands of measured exposures, typically breathing zone concentrations measurements in workers' breathing zone, in various databases.  Unfortunately, they were not particularly useful to the development of models which we believe represents the most basic element of the science of exposure assessment.    That is, they provided the left-hand side of the above equation but not the determinants of exposure on the right hand side.

As such, we set about to outline the data needs that would ultimately feed the modeling and exposure assessment process.  This resulted in Neil and I putting out the following paper which is now almost 20 years old:

Exposure Database Improvements for Indoor Air Model Validation

         APPL. OCCUP.ENVIRON.HYG. 10(4), APRIL 1995

I will be happy to send you a copy of this paper if you request it from me: mjayjock@gmail.com

Some of the bullets from this work are presented below.

We identified some “big picture” exposure determinants in need of capture

·         Source Characterization

·         Time Course

·         Sinks

·         Ventilation and Dispersion

To bring it down to earth in an operational scene, we proposed constructing a check-list for capturing the following minimal information in exposure databases to go along with the measured exposure:

·         source characteristics

·         areas and types of sinks

·         general type of building

·         room dimensions

·         qualitative ranking of smoke dissipation

·         assessment of equilibrium

I am sad to say that this idea never really caught on and, for the most part, we are still not capturing these determinants of exposure concurrently with our breathing zone monitoring data. 

Questions for discussion in the LinkedIn Groups:

Am I wrong?  Is there anything like this going on in your organization?

Why do you think this never got any traction?

Would you like to see such data captured and paired to monitoring data?

8 hour OELs versus Acute Bolus Exposures

Relying on full shift integrated sampling for comparison to an 8 hour occupational exposure limit could be problematic under some circumstances. 

Consider a hypothetical chemical with an 8 hour ACGIH TLV of 100 ppm (no STEL and no Ceiling TLV). Let’s also conditionally accept that we are okay with the AIHA exposure strategies criteria that a measured exposure of less than 10 percent of this OEL is considered an exposure that is not unacceptable.

Now let’s consider two very different exposure scenarios.   In the first situation (Scenario 1), the breathing zone concentration of this compound is relatively constant throughout the 8 hour shift and results in a measured value 7 ppmv after repeated sampling as an 8 hour time-weighted average (TWA).    In the second scenario (Scenario 2) the worker receives, for whatever reason, essentially no exposure for the vast majority of the period but a bolus exposure to the compound that lasts only 2 minutes and also results in a measured integrated 8 hour TWA exposure of 7 ppmv. 

Obviously, these represent two very different exposures that are measured as the same average over a one work-day time interval.   Depending on the chemical they could have resulted in very different effects on the worker being exposed.   For the sake of this example, let’s assume that the maximum instantaneous exposure was 10 ppmv in Scenario 1 with a minimum around 3 ppmv.  In other words, it occurred at a fairly constant rate of exposure with an average of 7 and a maximum instantaneous breathing zone concentration of 10 ppmv.   In Scenario 2 the average concentration over the 2 minute period of bolus exposure would be calculated as follows:

7 ppmv * (480/2)  = 1680 ppmv

Another way of checking this is to use the standard time-weighted averaging equation:

8 hour TWA  =  (C1* t1  + C2*t2  + … Cn*tn)/(480 minutes)

If the above initial equation is correct, then C1  In the second scenario is equal to 1680 ppmv, t1 = 2 min,  t2 = tn = 0 and total time = 480 minutes.    It checks, the 8 hour TWA comes out to be 7 ppmv.

The graph below illustrates this dramatic difference in maximum average breathing zone concentration for the different time periods for the two scenarios:

The right-hand bar would be 840 ppmv if it occurred over 4 minutes, 420 ppmv over 8 minutes and so on.

Supposed that the threshold mechanisms that protect the body by detoxifying or otherwise handling the chemical can be overcome with short-term, high concentration exposure; clearly, exposure Scenario 2 could result in a much stronger toxicological effect than Scenario 1.

As such, acute bolus exposure is of concern and the ACGIH addresses such situations with the following approach:

“Excursions in worker exposure levels may exceed 3 time the TLV-TWA for no more than a total of 30 minutes during a work day, and under no circumstances should they exceed 5 times the TLV-TWA, provided that the TLV-TWA is not exceeded”  [emphasis added]

Indeed, if we could anticipate the occurrence of any bolus exposure in a specific workplace location and time then it could be identified and addressed separately as its own scenario.    We may not always have this luxury however and may sometime be “stuck” with an integrated sample that indicates an acceptable exposure but an employee who has experienced a relatively high exposure and an untoward health effect.  One potential answer is to use real time monitoring of the employee’s breathing zone with a concentration-set alarm when the possibility of a bolus exposure exists. 

Questions for Discussion within the LinkedIn Groups: 

How would you handle a situation in which you had high bolus exposures occurring infrequently and somewhat unpredictably?  

Easy answer perfect world answer:  You monitor everyone everyday with real time monitors with alarms.   Real world question:    What might you do with hundreds of potentially exposed workers and bolus occurrences which are, at least at this point, unpredictable and are relatively infrequent?