Pinellas County, FL Pool Study 1992 (1994)

This thread presents my analysis of a study entitled, "A Comprehensive Study on The Microbicdal Properties of Stabilized and Unstabilized Chlorine and The Relationships of Other Chemical and Physical Variables in Public Swimming Pools; A Report on A Study Carried Out in Pinellas County, Florida, Summer/Fall, 1992". This link goes to a page at Oxy-Chem that contains the detailed study data. [EDIT] The links on the page were previously broken, but are now fixed. [END-EDIT]

The first thing that I did was to compile the source study data into a spreadsheet. This link is that spreadsheet put into HTML format so you can readily look at it. (If anyone wants the actual spreadsheets themselves, let me know and I'll post them or E-mail them to you). There were 486 pools in the study (the "N" column goes up to 494 because pools 14, 425, 434, 465, 466, 471, 473, 492 ended up not being measured or used). Originally they intended to monitor 500 pools, but 486 is what was ultimately used.

I then extracted out a subset of the columns of data and sorted by Free Chlorine and highlighted in red the bacteria and algae counts that were considered to be a problem (according to study criteria) and this may be found at this link. Finally, I calculated theoretical disinfecting chlorine (HOCl) concentrations and sorted by those and this may be found at this link.

The official position from a manufacturer I contacted was that they don't necessarily agree with my conclusions on the effects within pool water. They then talk about going through rigorous tests and studies before they are submitted to the EPA for federal registration. [EDIT] I am waiting for additional communication from them. [END-EDIT] So I'd like you to take a look at the same data and see if you can tell me what conclusions make sense.

I'm not concerned with bacteriological safety with high CYA in pools since it takes very, very little disinfecting chlorine to kill "easy-to-kill" bacteria. Hot tubs are another matter due to the bacteria that causes hot tub itch, Pseudomonas aeruginosa, and its much higher CT value (so it takes much higher chlorine concentrations to kill it). I think that the main problem with high CYA in pools is in terms of not preventing algae since higher chlorine levels seem to be needed to keep algae away, but this study makes absolutely no sense with regard to green algae (as you will see later).

(CONTINUED ON NEXT POST...)

(...CONTINUED FROM PREVIOUS POST)

The following is essentially what I sent to the manufacturer and talks about my analysis of the data. I would have preferred to have discussed this in private, but don't see much choice but to publish this [EDIT] since I haven't heard from them in a while [END-EDIT] and I'd like to continue to get feedback, more data, and do studies to confirm or refute my hypothesis. This was not my first E-mail to them, but it was the most recent where I used actual calculated HOCl values instead of a proxy estimate (FC/(CYA+1)), plus I thought more about what might be going on.

-----------------------------------------------------------------------

PINELLAS COUNTY STUDY DATA WITH HOCl
Attached is the spreadsheet of the Pinellas County 1992 commercial pools study where I have calculated the disinfecting chlorine (hypochlorous acid, HOCl) concentration in ppm Cl2 units (as all chlorine is measured). The spreadsheet has the 486 pools in the study sorted from highest to lowest HOCl concentration. Though the conclusions in the study are largely based on overall correlation analysis, I don’t think that this sort of analysis is particularly useful for this kind of data because the correlations aren’t strong enough to be used with reliability and safety. The mechanisms for chlorine inactivation of pathogens and inhibition of algae growth are more like step functions so if the data doesn’t show that kind of behavior, then the model is wrong and there are missing parameters that are not getting measured. For the Pinellas County data, this seems to be true both with the study assertion that FC is the sole parameter to be used and with my assertion that the calculated HOCl from pH, FC and CYA should be used. Neither approach is valid for this study data, so something is missing.

LOOKING AT GROUPS, NOT CORRELATION
To see this more clearly, let’s take a look at a percentage grouping analysis which would normally make any sort of “step function” much more readily apparent. The groupings were done using geometric (logarithmic) sizes since it is well known that chlorine inactivates mostly in a linear fashion (i.e. CT) so that doubling the chlorine should double the rate of inactivation. Something approaching a step function is to be expected because the amount of chlorine that inactivates roughly half of the pathogens (or algae) in the time it takes for the pathogen to double in quantity (i.e. the generation time) is a “critical” amount of chlorine. Below that amount, the number of pathogens is expected to increase while above that amount it is expected to decrease. So an “S” shaped curve is expected (actually, an upside-down “S”, so more like a “Z”, if chlorine increases on the X-axis and pathogen counts increase on the Y-axis).

The following table shows the percentage of failures (HPC>500, TCOLI,FCOLI>0, NCOLI>200, BK,YL,GN,PK<>”N”) in each grouping that has “Count” data points.

Count .. FC ...... HPC ....... TCOLI .... FCOLI .... NCOLI ... PSEUD .. TSTAPH . FSTREP .... BK ......... YL ........ GN ....... PK
49 ... < 0.1 .... 48.98% ... 36.73% . 20.41% .. 57.14% .. 6.12% ... 6.12% .. 2.04% .. 48.98% . 14.29% .. 0.00% .. 0.00%
. 4 .. 0.1-0.19 . 25.00% .... 0.00% ... 0.00% .. 75.00% .. 0.00% ... 0.00% .. 0.00% .. 75.00% ... 0.00% .. 0.00% .. 0.00%
14 .. 0.2-0.39 . 14.29% .... 7.14% ... 7.14% .. 14.29% .. 7.14% .. 28.57% .. 0.00% .. 42.86% . 14.29% .. 7.14% .. 0.00%
35 .. 0.4-0.79 .. 8.57% ... 11.43% ... 5.71% .. 20.00% .. 2.86% .. 11.43% .. 0.00% .. 34.29% ... 5.71% .. 2.86% .. 2.86%
44 .. 0.8-1.59 .. 9.09% .... 6.82% ... 4.55% .... 9.09% .. 2.27% .... 2.27% .. 0.00% .. 43.18% ... 6.82% .. 4.55% .. 0.00%
111 . 1.6-3.19 .. 4.50% ... 6.31% ... 4.50% .... 9.01% ... 0.00% .... 7.21% .. 0.00% .. 41.44% ... 7.21% .. 0.90% .. 0.90%
189 . 3.2-6.39 .. 4.76% ... 5.82% ... 3.70% ... 12.17% .. 0.00% .... 3.17% .. 0.53% .. 34.39% ... 5.29% .. 0.00% .. 0.00%
29 .. 6.4-12.79 . 0.00% ... 6.90% ... 3.45% .... 0.00% ... 0.00% .... 3.45% .. 0.00% .. 20.69% ... 3.45% .. 0.00% .. 0.00%
11 ... > 12.8 .... 0.00% ... 9.09% ... 0.00% .... 0.00% ... 0.00% .... 0.00% .. 0.00% .. 18.18% ... 0.00% .. 0.00% .. 0.00%

486 . Overall .... 9.88% ... 9.67% ... 5.76% ... 15.84% ... 1.23% .... 5.56% .. 0.41% .. 37.65% ... 6.79% .. 1.03% .. 0.41%

Now let’s look at the same table using the sorted table based on HOCl.

Count ... FC ............ HPC ...... TCOLI .... FCOLI ..... NCOLI .... PSEUD .. TSTAPH .. FSTREP .... BK ......... YL ......... GN ......... PK
58 .... < 0.001 ...... 43.10% .. 31.03% .. 17.24% .. 51.72% ... 5.17% ... 5.17% ... 1.72% .. 50.00% .. 12.07% ... 1.72% ... 0.00%
10 . 0.001-0.0019 . 30.00% .. 20.00% .. 20.00% .. 50.00% ... 0.00% .. 10.00% ... 0.00% .. 70.00% .. 10.00% . 10.00% ... 0.00%
22 . 0.002-0.0039 ... 9.09% .... 4.55% ... 4.55% .... 9.09% ... 0.00% .. 13.64% ... 0.00% .. 59.09% .... 9.09% ... 9.09% ... 4.55%
44 . 0.004-0.0079 ... 2.27% .. 13.64% ... 6.82% ... 13.64% .. 4.55% .... 9.09% ... 0.00% .. 36.36% .... 4.55% ... 0.00% ... 0.00%
92 . 0.008-0.0159 ... 7.61% .... 6.52% ... 3.26% .... 9.78% ... 0.00% .... 2.17% ... 0.00% .. 39.13% ... 8.70% ... 1.09% ... 1.09%
94 . 0.016-0.0319 ... 4.26% .... 5.32% ... 2.13% ... 12.77% .. 0.00% .... 6.38% ... 1.06% .. 24.47% ... 8.51% ... 0.00% ... 0.00%
55 . 0.032-0.0639 ... 3.64% .... 3.64% ... 3.64% ..... 3.64% .. 0.00% .... 0.00% ... 0.00% .. 32.73% ... 1.82% ... 0.00% ... 0.00%
28 . 0.064-0.1279 ... 0.00% .... 0.00% ... 0.00% ..... 7.14% .. 0.00% .... 0.00% ... 0.00% .. 25.00% ... 3.57% ... 0.00% ... 0.00%
83 .... >0.128 ......... 4.82% .... 8.43% ... 6.02% ... 10.84% .. 1.20% .... 9.64% ... 0.00% .. 40.96% ... 3.61% ... 0.00% ... 0.00%

486 ... Overall ......... 9.88% .... 9.67% ... 5.76% .. 15.84% ... 1.23% .... 5.56% ... 0.41% .. 37.65% ... 6.79% ... 1.03% ... 0.41%


174 ... < 0.011 ...... 20.69% .. 18.39% .. 10.34% .. 27.59% .. 2.87% .... 6.90% ... 0.57%
312 .. >= 0.011 ...... 3.85% .... 4.81% ... 3.21% .... 9.29% ... 0.32% .... 4.81% ... 0.32%

312 ... < 0.03 .................................................. .................................................. .......... 39.10% .. 8.97% .. 1.60% .. 0.64%
174 .. >= 0.03 .................................................. .................................................. ......... 35.06% .. 2.87% .. 0.00% .. 0.00%

You can readily see something that looks like a bit of a step function in the first one or two categories, but this step function does NOT go to 0 above a certain chlorine level (except FC>5.0). The earlier Pinellas County research study of about 1500 pools over 8 years (1973-81) stated the following:

They showed that as long as a minimum of 1 ppm of free available chlorine was present, algal growth was controlled and coliform bacteria were absent, even for pools containing up to 800 ppm of cyanuric acid. In addition, they showed that pools maintained the desired minimum of 1 ppm of free available chlorine more readily when cyanuric acid was present.

The 1992 study said the following:

The results of this study verified the results of the 1973-81 Pinellas County study.

Of course, that is simply not true since the first study claimed that 1 ppm FC was sufficient when clearly it is not. In fact, the 1992 study also concludes:

The free chlorine standard of 1.0 - 5.0 ppm of the Florida swimming pool code increases the probability that a swimming pool will be bacteriologically satisfactory for swimming. Hence, this standard is a more effective than the 1.0 - 3.0 ppm free chlorine standard employed by some of the other state regulatory agencies.

The main problem, of course, is that a free chlorine standard of 1.0 – 5.0 ppm says nothing about what FC level to actually use. Should it be 1.0? Should it be 5.0? Something in between?

Also, pool #318 had an HPC of 31000 and an NCOLI of 320 (and 2 TSTAPH) and yet had 5 ppm of Free Chlorine with a pH of 7.2 and no Cyanuric Acid. Just looking at HPC alone, there were 5 pools with an HPC>500 and FC >= 5.0; and a total of 9 pools (i.e. 4 more pools) with an HPC>500 and FC >= 4.0. How can one possibly have a standard or claim that FC alone makes a pool safe when so many pools are bacteriologically unsatisfactory?

I also find it very odd that pools with no measurable FC have lots of bacterial counts, but no green algae. Could the bacteria be consuming nutrients needed for algae? The study showed that some pools with no FC and lots of bacteria still had nitrates — but we don’t know about phosphates. Anyway, I just thought it particularly strange. Could it be that there is some use of algaecides in these pools (and maybe other pools as well)?

(CONTINUED ON NEXT POST...)

(...CONTINUED FROM PREVIOUS POST)

WHAT COULD POSSIBLY BE GOING ON?
Consider the following pieces of information:

• Many studies have shown that hypochlorous acid (HOCl) is the disinfecting form of chlorine for bacteria, protozoa (not that good for cysts, however), and viruses and that it is the form that inhibits algae though monochloramine is also effective for some forms of algae (yellow/mustard). These studies are mostly done in distilled water though some are done in simulated pool water, but simulated only in terms of a carbonate buffer and calcium, not in terms of organics from bather load.
• Clifford White’s “Handbook of Chlorination” and other sources refer to “base potential” or “poise” of different waters such that different amounts of chlorine are needed to reach a given ORP value. Or put another way, an amount of chlorine added to different waters such that they each measure with the same FC can produce different ORP values and therefore have different HOCl concentrations.
• Several studies have found that ORP is a far better predictor of water quality than FC alone. Other studies found that ion-specific membrane HOCl sensors were also effective and that ORP seemed to track HOCl concentration at a given pH and temperature (ORP varies with pH and temperature in ways beyond corresponding to the HOCl concentration).
• Cyanuric Acid (CYA) combines with hypochlorous acid (HOCl) to form several chlorinated isocyanurate compounds. Several studies have shown that these compounds are not effective sanitizers. Experience of users on pool forums shows that they also do not inhibit algae (notwithstanding one of the two studies I sent earlier that disputes this, but whose methodologies were flawed). The HOCl is released from the Cl-CYA compound at half-life rates on the order of seconds (0.25 seconds for one species; 4 seconds for another) such that the Free Chlorine tests (OTO, DPD, FAS-DPD) register the HOCl, OCl-, AND the Cl-CYA species by taking up (with a dye) the chlorine from all of these compounds.
• The Total Chlorine test which allows one to determine Combined Chlorine (CC), releases the bound chlorine from chlorinated organics and chloramines.
• Data from pool forums with residential pool users strongly validate the chlorine/CYA relationship with respect to green algae and appear to do so for mustard/yellow algae as well though there are too few incidents with that algae to be as certain. Residential pools do not appear to exhibit the sort of behavior of commercial pools as found in the Pinellas study. Instead, residential pools appear to exhibit the theoretical S-curve behavior.

If you haven’t figured out where I am going yet, consider the following link to a study that shows that monochloramine will register in the FC test if given enough time :

http://pubs.acs.org/cgi-bin/abstract.cgi/esthag/1984/18/i05/f-pdf/f_es00123a011.pdf?sessid=6006l3

The paper in the above link states the following:

This study confirmed that NH2Cl oxidizes DPD to the colored intermediate at a rate of 5.6 and 6.0% of the NH2Cl concentration, in the first minute, at 25ºC and 5x10^-5 M (3.5 mg as Cl2/L) and 1x10^-4 M (7.0 mg as Cl2/L) NH2Cl, respectively.

Have you guessed what I am about to hypothesize yet? Think about it. Monochloramine is normally considered to be a “combined chlorine” that should not get measured in the Free Chlorine test. And yet it does slowly give up its chlorine to be measured in the Free Chlorine test. We already know that Cl-CYA compounds will give up their chlorine relatively quickly (in seconds) to be measured in the FC test. So why can’t there be other organic compounds that behave like CYA in that they combine with HOCl to form new compounds, but are not bound as tightly as combined chlorines measured in the TC/CC test, but rather are more loosely bound more like the Cl-CYA compounds? These would be part of the “base potential” or “poise” of different waters referred to above.

In fact, if there were such chemicals, then they would precisely explain the results of the Pinellas study. The bulk of the pools exhibited the expected S-curve behavior requiring only a very low amount of around 0.002 ppm HOCl or perhaps even 0.001 ppm HOCl (measured in units of ppm Cl2, of course). This is, in fact, the expected amount of HOCl based on the hetrotrophic 2-log CT(99%) value off 0.08 since this corresponds to a CT(50%) value of 0.08*0.301/2 = 0.012 and using 15 minutes as the low end of the 15 minute to 1 hour typical generation time for bacteria this implies 0.012/15 = 0.0008 ppm HOCl.

If there were a CYA-like substance, then pools with higher calculated HOCl (using pH, FC, CYA) could have far lower actual HOCl. But because we are only measuring CYA and not these other substances, we don’t know about them nor account for them to find proper correlations! So let’s take a look at the worst-case for this, namely pool #318. What CYA level would it take to make that pool have a ppm HOCl of 0.001 and therefore be much more theoretically likely to have a high HPC? It would take about 2000 ppm of CYA. Now that sounds like an extraordinary amount, but consider that this mystery chemical (or chemicals, since there could very well be more than one) could bind to chlorine tighter than CYA (i.e. the Cl-Mystery has a smaller equilibrium constant of hydrolysis to release HOCl) and this normally implies that the rate of release of HOCl would also be slower. We know that it can’t be too slow or else it wouldn’t measure in the FC test, but it could be, say, 5 times slower and bind 5 times as strongly so that the amount of this mystery substance could be 2000/5 = 400 ppm in that worst case pool. In fact, it could be 10 times slower and still be measured as FC since 10*0.25 = 2.5 seconds for half-life.

As for why residential pools do not seem to experience this mystery chemical in sufficient quantities to disrupt the S-curve behavior, I can only speculate. My best guess is that residential pools tend to be smaller than commercial pools and have lower bather load. Many residential pools have DE or sand filters that are regularly backwashed (or cleaned, in the case of DE) so that pool water gets replaced over time. So between a low bather load and higher water replacement, there is less buildup of organics, including the “mystery” chemical. In fact, some residential pools with smaller volumes and that have regular backwashing can use Trichlor as their sole source of chlorine and not build up CYA to unacceptable levels — they can keep the CYA below 50 ppm. I looked at the TDS of pools in the Pinellas study to see if there was any sort of trend, but I couldn’t find any — most pools had rather high TDS and a few of the pools with high calculated HOCl but high bacterial counts had low TDS, so regular changing of the water isn’t necessarily the only answer. However, if TDS was measured using a conductivity test, then the “mystery” chemical could very well be neutral, though polar to dissolve in water, and not show up in a TDS test — CYA itself does not show up as TDS, though the CYA- ion does.

HIGH CYA HAS OTHER PROBLEMS
One of the recommendations from the study was the following:

Review with state regulatory agencies and the EPA the feasibility of increasing the cyanuric acid limit in public pools.

Raising the CYA limit above 100 ppm is not a wise thing to do because there are other problems with high CYA other than just needing higher FC (which I believe, but you may still not). The following links show how high CYA levels cause pitting in plaster surfaces:

http://www.tricitypoolservice.com/tc-plaster_study.html
http://www.findarticles.com/p/articl...4/ai_n15932555
http://www.aquamagazine.com/data/archive/AQ-406-61.pdf

Note that the above study was done by Arch Chemicals who manufacture CYA so this is an example of good corporate responsibility so that products will be used responsibly. [EDIT] This study on CYA and plaster has since come into question and may not be valid. [END-EDIT]

(CONTINUED ON NEXT POST...)

(...CONTINUED FROM PREVIOUS POST)

FURTHER EXPERIMENTS AND FIELD STUDIES
So my hypothesis that there are other chemicals that behave like CYA but that are not being measured is intriguing, but how do we go about proving if this is true or not? If we can find a pool that is like #318 or any other pool without CYA but with a reasonable FC, such as 1, and perhaps not in too much sun (so we know that the FC is being maintained), then we can measure the ORP in that pool as well as the specific HOCl concentration (using any of several techniques) and see if they are indeed measuring very low. I suspect that they will since that was the entire basis for the ORP industry and their studies.

If it is confirmed that pools measuring normal to high FC are measuring very low HOCl without any CYA to explain it, then samples of such water should be analyzed for their contents. I’m not sure what technique would be best, but a mass spectrometer might give a clue as to the chemical weights of the mystery chemical and we already know that in a pool such as #318 there will be a reasonable amount of it for measurement (in the hundreds of ppm). Another approach might be to look for nitrogen sites since they seem to be the ones where hydrogen is substituted for chlorine in both CYA and ammonia as well as other chlorinated organics (combined chlorines). Anyway, this now goes well beyond my abilities since I’ve been out of college for over 23 years and haven’t kept up with the latest in chemical analysis (I’m not a chemist; I do mostly software engineering and imaging and color algorithms).

I’d be very interested to know what you think about all of this and if you would be willing to pursue this investigation further.

-----------------------------------------------------------------------

In addition to sending the above as an E-mail (with attachments) to a manufacturer/distributor of pool chemicals, I also sent it to a manufacturer of Trichlor/Dichlor products, [EDIT] and I have heard back from the latter and they have been very busy so perhaps that is why I have not heard back from the former [END-EDIT]. Ideally, I would like to put together a presentation that sums up what we've learned here on this forum, not only about chlorine/CYA, but also about lowering TA to reduce rising pH (especially in SWG pools), how to lower TA properly (through low pH and aeration), chemical equivalents among product brands (and store products -- e.g. baking soda, bleach), etc. I then would like that presented at the next Pool & Spa Expo or similar trade show so that pool dealers and distributors can get some better education and resolve customer pool problems more quickly. Many people use a pool service because they simply do not want to bother with brushing their pool or adding chlorine regularly and people still need to buy equipment and occasionally some specialty chemicals so it's not as if the industry will go away with better education. Some people will simply not maintain proper chlorine levels regularly and using an algaecide will help them in that case -- it's not cheap, but that's a choice a consumer can make. The goal is to at least get rid of the false claims or misinformation that seems fairly rampant in this industry and to shift the focus to a higher level of integrity based on truth and real consumer needs. Any suggestions for how these goals can be best accomplished are most welcome.

[EDIT]
This link to the U.S. Patent Office shows a patent that describes how glycoluril binds to chlorine more strongly than CYA does, thereby increasing protection from sunlight to go from a chlorine half-life of 6-7 hours to around 25 hours. It also appears to have lower disinfecting chlorine for the same total chlorine level. It also shows lower rates of disinfection by-product production and outgassing of monochloramine. This is all consistent with the "mystery" chemical that I hypothesized might exist. What I do not yet know is whether chlorine bound to glycoluril will get released fast enough to show up in the Free Chlorine test. If it does, then this is exactly the kind of chemical that could explain the Pinnelas study results, though this is just an existence proof of the concept, not an actual identification that this is the actual chemical in the Pinnelas pools (nor a proof that such a chemical is truly what is going on). Nevertheless, this is very encouraging and may lead to better tests to predict bacteriological safety using standard chemical tests (as opposed to ORP or HOCl sensors).

So far, what this means to residential pool users, is that though the chlorine/CYA relationship is still very likely to be true (the laws of equilibrium chemistry have not changed), there may be other chemicals that act like CYA and get bound to chlorine reducing its effectiveness. We haven't seen data from residential pool users for this since there hasn't been a single reported case of green algae when maintaining chlorine levels vs. CYA levels according to Ben's table (roughly the 0.03 ppm HOCl column in my table for Ben's Min equivalent). At least I could not find any exception -- if anyone can, please let me know. I do know that the few cases of mustard/yellow algae required higher levels of chlorine to keep away, but we sorted that out to be roughly the Max level in Ben's table (roughly the 0.07 ppm HOCl column in my table). Black algae seems to be even more rare in residential pools and brushing plus chlorine seems to get rid of it most of the time.

I also found this patent which combines Trichlor with glycoluril to have it be more slow-dissolving, but it would also add glycoluril to the water that would have the characteristics described above. It does not appear that this combination is in any current products. Also, some of the pools with high FC and high HPC counts in the Pinnelas study had no CYA so direct organic introduction of a CYA-like substance seems more likely. Glycoluril is formed from a combination of urea (obviously present in some pool water, though usually reacts with chlorine to form chloro urea intermediates and to then break down to nitrogen and carbon dioxide gasses) and glyoxal (not clear where that comes from -- *maybe* an oxidized breakdown product from some organics??? more information may be found here). The concentration of urea in urine is about 400 millimole/liter or 24,000 mg/l (ppm) while in sweat the concentration of urea is around 22 millimole/liter or 1300 mg/l (ppm). The concentration of glyoxal in human urine is 132 micromoles/liter or about 7.7 mg/l (ppm). However, the production of glycoluril from glyoxal and urea normally requires heat and a very acidic environment so I doubt the glycoluril is coming from people. Again, I do not know if glycoluril is the mystery chemical explaining the anomalous data or if something else is going on.
[END-EDIT]

(END OF MULTIPLE SEQUENTIAL POSTS)

Ben, I cannot edit my post (if possible, please give me that ability even for old posts). I need to correct the link to the Pinellas County study since it has been dead for a while. I asked Oxy Chemical to rehost the study on their site since NSPF took it down and they have now down so. So the first "This link" in the first post should go to this link where there are links to the study and its appendices under the "Technical Information" section. Thanks.

It's fixed. -ben