Archive for the ‘Uncategorized’ Category


The Petroleum Equipment Institute (PEI) held its 2018 convention at the Las Vegas Convention Center from 07 to 10 October 2018. As usual, the PEI convention was held in conjunction with the much larger National Association of Convenience Store (NACS) convention. Today, I’ll focus on a few items that are particularly relevant to fuel and fuel system microbiology. I’m not going to attempt to provide anything approaching an overview of the entire convention. Instead I’ll report and discuss a few statements I heard from speakers during PEI’s Tuesday 09 October education sessions.

Regulatory issues

EPA Regulatory Update – Carolyn Hoskinson, Director of EPA’s Office of Underground Storage Tanks (OUST) and several members of her staff spoke to the current state of affairs regarding UST regulations. Tony Raia reported that with the 13 October 2018 compliance deadline looming, 32 states had updated their UST regulations to harmonize them with the 2015 updated US EPA regulations. Tony identified five state categories:
1. State Program Approval (SPA) States that have completed their updates and which are now in full compliance
2. Non-SPA States that have completed their updates to comply with the 2015 regulations
3. SPA States that have delayed revising their state regulations
4. SPA States that have updates in progress
5. Non-SPA States that have not yet updated their regulations per the US EPA 2015 regulations.
Bottom line is that we are entering a period during which there will be some confusion over compliance.

U.S. EPA UST Enforcement – Mark Barolo – the US EPA OUST official responsible for enforcement – noted that in nearly all cases, individual States were responsible for enforcement. Recognizing the confusion, Mark opined that inspectors were going to address violations on a case-by-case basis. Generally speaking, retailers who had been incompliance, had the required documentation, and demonstrated that they were making good-faith efforts to ensure that they remained in compliance, would experience less enforcement grief than those who have not. Mark’s colleague, Cho Yi Risher noted that the regulations do not prescribe the time permitted for site owners to repair or replace non-compliant equipment. Moreover, the inspections required by the 2015 regulations identify non-compliant equipment. There is no incentive for owners to institute predictive maintenance programs (see my 16 January 2017 post) that would detect failure trends before equipment became non-compliant.

Failure to detect uncontrolled microbial contamination and biodeterioration before they cause valves to seize or tanks and lines to leak is a false economy. A few pennies saved during regulation-mandated inspections can lead to remediation expenses in excess of $0.5 million.
Although for some of us, it’s hard to believe that the UST installed in 1987 – in compliance with the original UST regulations – are now beyond their 30-year warranty life. Discussion during the session’s question and answer period indicated that all stakeholders shared a common interest in ensuring that sites with tanks that were more than 30-years old would be able to continue to operate. I anticipate seeing articles to this issue in PEI Journal in the coming months.

Fuel Quality and Corrosion
Scott Boorse – PEI’s Technical Program Manager; recently retired from a major fuel retailer – made several observations that validated much of what I’ve been discussing in Fuel Microbiology What’s New posts. He suggested that in his experience, 100 % of all retail site fuel systems had some corrosion. He attributed much of this corrosion to the bacterial genus Acetobacter converting ethanol to acetic acid. I am convinced that most of the headspace and spill containment well acid production comes form chemical oxidation of ethanol to acetic acid. Microbes are involved in an estimated 50 % of all system corrosion issues, but – as I’ve written previously – microbes produce a variety of organic acids. These acids can react with chloride, sulfate, and nitrate in fuel-associated water to form organic bases (or salts) and strong, highly corrosive, inorganic acids – hydrochloric, sulfuric, and nitric acids, respectively. Still, Scott was on spot suggesting that UST system corrosion was much more wide-spread than most stakeholders realize.

Rebbeca Moore – GM and chair of the automotive industry’s Top Tier Detergent Gasoline and Diesel Fuel consortia – discussed the importance of fuel quality on engine performance. Top Tier is an auto industry sponsored compliance program intended to go beyond ASTM product specifications typically cited in state regulations. I’ll steer clear of the perennial debates between engine manufactures and petroleum producers that enliven our semi-annual ASTM D02 (Petroleum Products) subcommittee A (Gasoline and Oxygenated Fuels) and E (Burner, Diesel, Non-Aviation Gas Turbine, and Marine Fuels), but Rebbeca made an important point. ASTM specifications are often misused. They are meant to indicate whether a product (i.e., fuel) is fit for use at a single point and place in time (i.e., when and where the sample was collected). The petroleum industry’s infrastructure is vast and complex. Moreover, product ages (if it didn’t it wouldn’t combust so well in engines). Specification tests provide little information about how the product will age during storage.
Rebbeca illustrated the dilemma by listing the typical components of 7,500 gal of in-specification ULSD delivered to UST:
• 1 cup of dirt
• 1 to 2 gallons of water
• Up to 325 gallons of FAME (B5 ULSD is now included in ASTM Specification D975 Diesel Fuel Oils)
• 1 gallon of glycerin
• 5 to 40 gallons of additives
At sites that receive frequent deliveries, these trace amounts of dirt and water add up! Not surprisingly, along with measures that are outside the retail site or fleet owner’s control, Rebecca recommended more aggressive water removal and better dispenser filtration (there is an ongoing debate among stakeholders with some recommending that all dispensers have 5.0 µm, water absorbing filters and others arguing for 100 µm particulate filters). I share Rebecca’s view that all dispensers should have 5.0 µm, water absorbing filters. Marketers who are focused only on low rates and not product quality have argued for eliminating the filtration requirement completely.

Ryan Haerer – of US EPA’s OUST ( – wrapped up the session, sharing a few notable points. First, Ryan reminded attendees that UST regulations apply only to system components that are in contact with the soil. The US EPA does not regulate the condition of internal components that are not in direct contact with the soil (for example, submerged turbine pumps – STP – and their associated hardware). He also explained that under the UST regulations, there is a requirement that system components be compatible with the substance stored. This is likely to become interesting as new products (for example, E15 gasoline or substitution of ethanol with isobutyl alcohol) are introduced into the commercial fuel infrastructure – here I’m using interesting – in the same way it is used in of the phrase: “May you live in interesting times.” (Austen Chamberlain – British Foreign Secretary, 1924 to 1929 wrote that he had been told that this was an ancient Chinese curse, but his claim has never been verified).

Bringing it home

At this point we are enjoying an interesting paradox. Regulators, insurers, and an increasing number of retailers recognize that waiting until fuel systems fail is a problem. However, the system largely provides incentives for site owners to wait until failures have occurred. After failure, insurance covers component replacement costs. In many states, superfund monies cover remediation costs. When site owners invest in predictive maintenance, they only see the costs. Although there are benefits – not the least of which is customer satisfaction and a positive corporate image – they are intangible. How do we break the paradox?

Please share your thoughts on this issue with me at I’ll compile comments and post them anonymously as a future What’s New column.


Biocide treatment releases biomass – now what?

Disinfection using microbicides is only one element of the fuel system decontamination process. This month’s post covers what needs to be done after a fuel system has been treated with a microbicide.

When a moderately to heavily contaminated fuel system is treated with an effective biocide, masses of biofilm material – flocs – get suspended into the fuel. As illustrated in figure 1, some of this biomass quickly settles to the tank bottom and the rest is carried with the fuel to the submerged turbine pump. Dispenser filters are designed to capture fuel particulates and biomass flocs. When the fuel is clean, a 10 m fuel filter, mounted in a retail dispenser, can process 500,000 gal (1,900 m3) to a million gal (3,800 m3) of fuel before it needs to be replaced (Note: there are no consensus criteria for filter life. Some retailers replace dispenser filters when the flow is less than 8 gpm – 30 L min-1. Others wait until flow is less than 2 gpm – 7.6 L min-1. In my November 2016 post, I detailed the economics of dispenser flow rates. The opportunity costs caused by slow flow can be startlingly expensive! After a fuel system has been treated, the next step is to get the flocs out of the fuel.

Fig 1. Effect of biocide treatment on biofilm masses – a) photo of the bottom of an underground storage tank – submerged turbine pump is to left, just outside of view so that the left side of the photo is in line with tank’s bottom centerline; b) schematic representation of (a), showing biomass accumulation on either side of bottom-dead center; c) same as (B) but after biocide treatment – biomass flocs are not dispersed in the fuel.

Floc removal – fuel polishing

There are three options for floc removal. Option 1 is to stop using the dispensers that draw product from the treated tank, give the flocs a day or two to settle to the bottom, and then to vacuum out the bottom sludge, sediment, and water. I can’t think of any retailers or fleet operators who would choose to put a tank out of commission for a couple of days.

Option 2 is to let the dispenser filters do all the work. This can translate into multiple filter changes per day for several days (fig 2). When the fuel reaching the dispenser is loaded with biomass flocs, filters can plug after 2,000 gal (7.6 m-3) to 10,000 gal (38 m-3) have flowed through them. This option might be feasible at rural sites that sell fewer than 10,000 gal (38 m-3) per week but is not particularly practical at high volume facilities. Long fuel turnover periods (i.e., more than two days) give biomass flocs time to settle to the tank’s bottom. As is the case for option 1, settled sludge and sediment needs to be removed from the tank as soon as possible after biocide treatment. Optimally, this is done one or two days after the treatment.

Fig 2. Removing biomass flocs the hard way – flocs are pulled into the submerged turbine pump and carried to the dispenser filter. As shown in the inset, the dispenser filter quickly becomes covered with slime; preventing fuel from flowing to the dispenser’s nozzle.

Option 3 is to use a fuel polishing rig (fig 3). Filtration rigs come in numerous configurations. Filtration rigs come in nearly as many configurations as there are companies who offer fuel filtration services. The number of filter housings on a rig typically ranges from one to three – although there are rigs with more than three housings. Diverse filter media are available – each with advantages and disadvantages relative to the others. Filter housing designs differ by the types and number of filter elements they contain, and by fluid flow patterns they use to optimize filtration efficiency. I’ll leave it to the mechanics and engineers to provide details on filtration technology and rig design. In this post, I’ll describe a generic rig.

Rigs with multiple housings use filters in series. Fuel first passes through a coarse filter (for example, designed to remove particles and masses that are >100 m), and then through a polishing filter (nominal pore size between 1m and 5m). Figure 4 shows a 16-element filter housing that’s mounted on a skid. Some filtration rigs include a fuel-water separator. Others rely on coalescer filters to strip water out of the recirculating fuel.

Filtration rigs have one or more pumps to drive fuel recirculation. Most commonly, a pump pushes fuel into the tank to create turbulent flow. This turbulence helps to keep particles and biomass flocs in suspension. The pump’s discharge creates pressure and its intake creates a vacuum. The vacuum draws product through the return riser and line to the rig. In fig 3, the pump drives fuel into the tank via a stinger that is inserted into the fill tube fitting and draws fuel through a stinger that is dropped through the turbine riser fitting. The fuel discharge stinger can be rigid or flexible. Some stinger designs include high pressure nozzle meant to source biofilm residual material from tank walls.

Depending on how heavily contaminated the fuel is, three to seven fuel-volumes might need to recirculate through the filtration rig to complete the polishing process. Depending on rig design, filtration flow rates range from 75 gpm (0.28 m3 min-1) to 350 gpm (0.9 m3 min-1). For a tank containing 5,000 gal (19 m3) of fuel, this means that post-treatment fuel polishing can take from less than 30 min [(5,000 gal 350 gal min-1) x 3 cycles  23 min] to 8h [(5,000 gal  75 gal min-1) x 7 cycles  470 min  8h]. As I mentioned above, greater flow rate also helps to keep particles in suspension.

Fig 3. Removing biomass flocs the easy way – high-capacity filtration rig processes recirculating fuel at 75 gpm to 350 gpm. In this illustration pump pulls fuel through a 3-stage unit (fuel-water separator, coarse filter, and polishing filter) and discharges the filtered fuel – via stinger (in) back into the tank. The pump’s suction also creates sufficient vacuum to draw fuel, water, and particulates from the tank to the rig’s inlet. Using the filtration rig keeps dispenser filters (inset) in pristine condition.

Fig 4. High-capacity, two-stage filtration rig – inset shows 26 filter cartridges inside 1st-stage filter housing.

My fuel is now clean – is my system also clean?

The answer depends what you mean by clean. Periodic microbicide treatment and subsequent fuel polishing can be enough to prevent tank deterioration problems. However, if biodeterioration damage began before the tank was treated and fuel polished, more thorough cleaning might be needed.

Even the best fuel polishing equipment can only direct pressure at surfaces that are in direct contact with the fuel. High pressure systems can be used to remotely clean the surfaces of empty tanks. It might be necessary for workers to enter the tank (personnel performing this work must be properly trained and certified to operate in confined spaces) and clean its surfaces manually (fig 5).

Fig 1. Confined space entry to clean tank walls. Source:

The only way to know for certain that a tank is clean is by visual inspection. Available remote camera technology can only be used to see exposed surfaces (for example, see: They cannot be used to see surfaces that are below the fuel level. Consequently, for visual inspection by remote camera, tanks must first be emptied (product can be drawn into either a tank truck or frac tank). Confined space entry and direct inspection remains the most reliable means of evaluating tanks for cleanliness, coating condition, and corrosion.

Why do routine condition monitoring if visual inspection is the gold standard?

If direct observation is truly the only way to know how heavily contaminated a tank is, why bother with the various types of tests I’ve described in previous Fuel Microbiology blog posts? The answer lies in return on investment (ROI). Data from routine sample testing (see my December 2016 Blog post provides important infromation about the fuel system’s condition. Most frequently, easy tests that require little or no equipment, act as the canaries in the mine. If results indicate that something is changing for the worse, more advanced tests help to determine what is going on. One can run routine tests and complete preventive maintenance actions for years for the cost of a single visual inspection. Consequently, internal inspections are reserved for when data (or regulatory ordinances) indicate they are needed.

For more information about detecting and controlling microbial contamination in fuel systems, please contact me at


Take two gallons and call me in the morning – not!

In Part 21, I reviewed the three primary types of fuel treatment microbicides – classifying them by their respective solubilities in fuel and water. You’ll recall, that I recommend using products that are soluble in both fuel and water – what I call: universally soluble. If you don’t remember why I prefer universally soluble, fuel treatment biocide, please re-read Part 21.
In today’s post, I’ll discuss how to get the most effective results when you treat a fuel system with a microbicide. Spoiler alert, I do not recommend simply dumping the minimum dose (gal microbicide per gal fuel) into your system and hoping for the best.

Think strategically
Before treating a fuel system, ask yourself why you are adding microbicide. To many readers the answer will be obvious, but not necessarily correct. Yes, the objective is to kill microbes. However, have you first considered where the microbes are living, or how much biofilm has accumulated on fuel system surfaces? When you treat a heavily contaminated system, and biofilm sloughs off tank and pipe walls, where is it going to go? Is it reasonable to assume that a single dose will disinfect my system? How will you know if your treatment was effective? If you don’t think about these issues before you treat, you probably will afterwards.

Where are the microbes living?
Tank bottoms
Most often we base fuel treatment decisions on test results from bottoms samples. Growth at the fuel-water interface can be seen either as an invert (water in oil) emulsion (rag-layer), membrane-like layer (pellicle), or both (fig 1a). Profiles of bioburdens in fuel, interface, bottoms-water layers generally show that the greatest bioburden is in the interface layer (fig 1b).

Fig 1. Fuel-water interface. 1a: fuel over water, separated by rag-layer; 1b) schematic profile showing maximum bioburden within rag-layer, minimum in fuel, and intermediate bioburden in bottoms-water.

In Part 15 (November 2017), I wrote about biofilms. Biofilms are complex, slimy residues that can form at the fuel-water interface (as in fig 1a) and on fuel system surfaces, including bottom-sludge and sediment. Among their numerous fascinating properties, biofilms can act like a slime fortress – preventing microbicides from reaching biofilm microbes.

Treatment objective(s)
Based on the previous section, it should now be clear that most commonly, the objective is to disinfect fuel system surfaces. Treated surfaces can include any combination tank walls, pipe surfaces, valves, meters, and pumps. Treated surfaces will not include the tank ullage zone. If fuel is not in direct contact with a surface, neither will the microbicide. I’ll discuss disinfecting ullage surfaces in my next blog post (Part 23). If only tank surfaces need to be disinfected, then static soaking can be sufficient. However, if the objective includes disinfection of other fuel system component surfaces, the treated fuel will have to be recirculated to ensure that they are exposed to the microbicide.

Choosing the correct dose
Dosage is the volume of microbicide added per gallon of fuel (recall from Part 21 that I recommend against doing before removing bottoms-water). All microbicides list minimum and maximum dosages on their container labels. The minimum dose is based on laboratory tests that can provide optimistic results. I always recommend using the maximum permissible dose. Maximum dosage is based on the regulatory agency’s toxicological risk assessment of the microbicide’s active ingredient(s).
I recommend using the maximum permissible dose because the concentration of microbicide available to kill microbes begins to decrease once the product has been added to the fuel. Collectively, the factors contributing to the disappearance of microbicide active ingredient are called demand. There are chemical, physical, and microbiological demands. Figure 2 illustrates how microbicide concentration can decrease over time. The time axis in fig. 2 can range from hours to months.

Fig 2. Microbicide demand curve.

The most common physical demand is dilution. Each time untreated fuel is added to a tank containing treated fuel, it dilutes the microbicide concentration. Figure 3 illustrates what can happen once the microbicide concentration decreases to below its critical concentration – the minimum concentration at which the active ingredient is effective. Note how active ingredients that are quite effective when used as directed can actually stimulate growth once their concentration is less than the critical concentration. In fig. 2, microbial populations exposed to sub-critical microbicide concentrations are more than three orders greater than those in untreated systems. Conversely, at concentrations ≤50 % of the maximum permissible dose, the microbicide is fully effective.

Fig 3. Microbial population response to different microbicide doses (hormesis).

Dosing plan
Although a single dose is often sufficient when treating a lightly contaminated system, it can be insufficient for disinfecting moderately to heavily contaminated systems. The reason is microbicide demand. An effective treatment exposes microbes to adequate concentration of active ingredient for a sufficient period of time (the soak period). The optimal soak period is between 24h and 48h. Except for long-term storage tanks, operators rarely have the luxury of allowing their fuel tanks to stand idle for this long. Depending on the fuel turnover rate, it might be necessary to add microbicide in order to maintain the active ingredient’s effective concentration for at least 24h (this is most commonly an issue at sites that receive more then one fuel delivery per day).

Additionally, active ingredients are used up as they react with microbes. The more heavily contaminated a tank is, the more quickly the available microbicide concentration will decrease. Figure 4 illustrates a point I made above, regarding biofilms. The initial treatment is unlikely to remove the entire biofilm or to kill microbes deep within the biofilm matrix. One or more follow-up treatments might be needed to fully eradicate the biofilm community (fig. 4c, d, and e). Each treatment will cause masses (flocs) of biofilm material to slough away from the surface to which it was originally attached. Some of these flocs will settle to the tank’s bottom. Those that don’t will remain suspended in the fuel and be transported to filters. Rapid filter plugging is a common result of effective biofilm destruction.

Fig 4. Biocide interacting with biofilm – a) biofilm accumulation on a surface; b) first biocide dose penetrates into the biofilm partially, causing some biofilm material to slough off; c) second biocide dose treats most of the remaining biofilm; d) third does disinfects surface; e) after effective treatment, surface is biofilm-free.

When tanks or sufficiently contaminated to cause filter plugging after biocide treatment, polishing, fuel tank cleaning or both should be part of the remedial effort. In part 23, I’ll write about fuel polishing and tank cleaning. In the meantime, if you have questions or comments about today’s post, please contact me at

Microbes are ubiquitous. There are extraordinarily few habitats on earth where thriving, microbial communities have not been detected. In practical terms, this means that it is unlikely that operators will ever have a completely sterile fuel system or that they will reduce their fuel system biodeterioration risk to zero. Biodeterioration can still occur in the best maintained fuel systems. However, the risk of it occurring in an inadequately maintained system is much more likely


Fig 1. From Rime of the Ancient Mariner, Samuel Taylor Coleridge, 1798

Water, water everywhere…

Samuel Coleridge’s infamous mariner paid dearly for having killed an albatross (figure 1). Do fuel quality managers and personnel responsible for fuel system integrity pay dearly for underestimating the ability of small (<1 oz; 30 mL) pools of fuel-associated-water left behind after water has been nominally purged from a fuel tank? A water bit is any small volume of water that remains in a fuel tank after dewatering (my personal, technical definition).
In Part 20, I wrote: “No water means no bugs. Is it as easy as all that?” I also explained why the short answer to the question was: “No.” For emphasis, I’ll again share figure 3 from Part 20 (figure 2, here):

Fig 2. Scale: how 2 mm of water appears to a bacterial cell.

A – a 6’6” tall man standing at the base of Mt. Kilimanjaro; B- a bacterial cell “standing” in a pool of water that is >2 mm deep; the ratios between the height of Mt. Kilimanjaro and the man in A, and between the depth of the pool of water and the bacterial cell in B are the approximately the same.

What can we do about these traces of water?
I confess that I am not a big fan of dispersants. When water dispersants are used routinely as fuel additives, the dispersed water can act as a corrosive agent; damaging engine components. However, when used to complete the job started by draining or vacuuming most of the free-water out of a tank, dispersants can be quite effective.
Figure 3 illustrates how dispersants work. Most dispersants are organic molecules that have a polar (charged; water-soluble) head and a non-polar (non-charged; fuel-soluble) tail. When added to fuel over water (figure 3b), they move towards anywhere where fuel contacts water (figure 3c) and trap tiny (typically <1 µm; 0.0004 in dia) fuel droplets. The fuel “sees” only the dispersant’s non-polar tails, so the droplets disperse uniformly throughout the fuel (figure 3d). The dispersed droplets (micelles) get transported with the fuel and evaporate during combustion in the engine cylinder.

Fig 3. Dispersant action: a) fuel over bottoms-water; b) dispersant added to fuel – inset shows dispersant molecule with polar head and non-polar tail; c) dispersant heads and tails align in water an fuel phases, respectively; d) dispersants form micelles with water droplet trapped in center; typical droplet size is < 1 μm dia.
The use of dispersants is controversial. Dispersant manufacturers and marketers focus on dispersant effectiveness in keeping free-water from accumulating in fuel systems. Moreover, under most circumstances, microbes are unlikely to make use of water trapped within dispersant micelles. Conversely, engine manufacturers focus on the potential for dispersed water to corrode and erode injector nozzles; particularly on modern, high-pressure, common-rail diesel engines. Interestingly, there is at least one additive manufacturer that has tried to promote water-emulsion diesel fuels – diesel with a dispersant that enables the fuel to hold as much as 25 % water. As a microbiologist, I was looking forward to investigating microbial contamination problems in systems that handled 25-75 water-in-diesel blends. But that’s another discussion.
In my next blog, I’ll focus on fuel treatment biocides. In the meantime, if you have questions or comments about today’s post, please contact me at

Microbes are ubiquitous. There are extraordinarily few habitats on earth where thriving, microbial communities have not been detected. In practical terms, this means that it is unlikely that operators will ever have a completely sterile fuel system or that they will reduce their fuel system biodeterioration risk to zero. Biodeterioration can still occur in the best maintained fuel systems. However, the risk of it occurring in an inadequately maintained system is much more likely.



I’ll open this post with a disclaimer. Microbes are ubiquitous. There are extraordinarily few habitats on earth where thriving, microbial communities have not been detected. In practical terms, this means that it is unlikely that operators will ever have a completely sterile fuel system or that they will reduce their fuel system biodeterioration risk to zero. Biodeterioration can still occur in the best maintained fuel systems. However, the risk of it occurring in an inadequately maintained system is much more likely.


I’ll also take this opportunity to remind readers that biodeterioration (damage caused by organisms) and bioremediation (using microbes or other organisms to degrade or remove toxic or noxious chemicals) are the flip-sides of the biodegradation coin (figure 1).

Figure 1. Just as the two sides of a coin are its obverse (front) and reverse (back) sides, the two sides of biodegradation are bioremediation and biodeterioration.

Microbes degrade fuel quality and fuel system components. In high-turnover retail systems, product deterioration is unlikely. I consider any tank that is refilled at least weekly to be a high-turnover (or high-throughput) system. The time that product spends in the storage tank is too short for degradation to occur. Studies that investigate the rate at which microbes change fuel chemistry, typically show substantial changes after a month or longer. Consequently, fuel in tanks used for emergency generators, or seasonally operated equipment is at greater biodeterioration risk than fuel in retail underground storage tanks (UST), or frequently operated vehicles. That’s all I want to say about fuel biodeterioration for now. I’ll return to the topic in a future blog post.

For now, I’ll focus on fuel system biodeterioration. Most of the damage caused to fuel system components is caused by biofilm communities (see post #16 Microbes cause damage either directly or indirectly (more on this in a future post). The most obvious indications of biodeterioration are filter plugging and corrosion. Although it’s typically the first indication of a biodeterioration problem, filter plugging is a late symptom. I often compare it to a heart attack; a late – but often first recognized – symptom of coronary disease.

So how do we substantially reduce fuel system biodeterioration risk? Step one is cost-effective condition monitoring (CM). Step two is cost-effective predictive maintenance (PdM; see The former drives the latter. Why do I emphasize cost-effectiveness? As I see it, there is as little justification for investing $10,000 per year to detect problems that might cause $1,000 per year of damage, as there is in refraining from spending $10,000 per year to detect problems that could cost $100,000 per year. I’m not suggesting that any fuel system CM program should cost $10,000 per year. An effective program can cost less than $2,000 per year. My point here is that before setting up a CM/PdM program, stakeholders should invest a bit of time and effort to determine their actual annual biodeterioration-related costs.

Opportunity Cost:

Nearly two decades ago, I first argued that a 10% flow-rate loss at high-traffic, retail sites, can easily translate to more than $100,000 per dispenser per year opportunity cost (Passman, F.J., 1999. “Microbes and Fuel Retailing: The Hidden Costs of Quality.” Nat. Petrol. News 91 [7]: pp: 20-23). My model did not include lost C-store revenues related to customer discontent with their fueling experience. Retailers who have tested my model have invariably been shocked by the huge impact of seemingly minor flow-rate reductions on fuel sales volumes. I’m still trying to understand the psychology behind retailers’ general reluctance to even test my model (for model details, contact me at Bottom line: the return on investment (ROI) for well-designed and executed, CM can easily be >$1,000 return on each $1 invested.

Condition Monitoring:

In Parts 2 through 18 of this series, I’ve written about the details of condition monitoring. I won’t repeat that information here. Instead, I’ll offer a few basic guidelines:

1. Testing hierarchy – CM plans should include two or more tiers. Tests that are easiest and least expensive to perform should be done most frequently (checking UST for bottoms-water accumulation and dispenser flow-rate checks are great examples of Tier 1 tests). Tier 2 tests include bottom-sample visual inspection (optimally, samples should be collected from the fill, automatic tank gauge, and submerged turbine ports). A simple microbiological test (for example ASTM D7687) is indicated whenever the bottom-sample is turbid or when it includes water. When Tier 2 tests indicate that the biodeterioration risk is moderate to high, Tier 3 tests (generally performed by a qualified laboratory) are used to confirm the risk.

2. Test method selection – notwithstanding the examples I mentioned under Testing hierarchy each site owner should develop a CM plan that best meets their needs. You can read my test specific blog posts (or ASTM D6469) for discussions of the benefits and limitations of each test method.

3. Testing frequency – my rule of thumb, after you have determined how often a test parameter is likely to indicate an increased biodeterioration risk, divide that period in three. That gives you the optimal test interval. Testing more often typically translates into greater costs without any real ROI. Testing less frequently increases the risk of having to perform corrective – rather than preventive – maintenance actions.

4. Understanding trends – each of the three previous guidelines depends on a basic understanding of trends. For example, even fuel with an ISO 4406 cleanliness rating of 18/16/13 (see, will eventually plug dispenser filters. Consequently, dispenser flow-rates will invariably decrease. How many operators know what normal looks like? How many have a control limit? Typically, dispenser filters can process >250,000 gal of fuel before the flow rate will fall below 7 gpm. Replacing fuel filters when the flow-rate is <7 gpm strikes a balance between the opportunity maintenance costs. Knowing whether the flow rate has fallen to <7 gpm after 50,000 gal or 500,000 gal of fuel have been filtered serves to trigger additional sampling and testing. If the amount of fuel filtered before substantial flow reduction occurs is much less than expected, then additional testing is indicated.


In summary, microbial contamination control depends on a good CM program that is linked to a good PdM program. A reasonable investment in CM and PdM should be a fraction of the likely cost impact of not having those programs in place. A cost-effective CM program is driven by an understanding of system trends, definition of the methods that will provide the most useful, actionable information; selection of which tests to run; and determination of sampling and testing frequency. In my next blog, I’ll focus on preventive measures. In the meantime, if you have questions or comments about today’s post, please contact me at



Quick review:

In post #12 I provided an overview of microbiological testing.

Next (post #13) I launched my discussion of non-culture tests.

In posts #14 and #16 – post #15 captured my impressions of the fuel microbiology sessions at two conferences – I discussed how ATP testing could be used to measure microbiological contamination in both liquid and solid samples (surface swabs and sections of filter media).

Before moving on from microbiological test methods, I want to cover two more non-culture test methods. The first – ASTM D8070 (Method for Screening of Fuels and Fuel Associated Aqueous Specimens for Microbial Contamination by Lateral Flow Immunoassay; – is another test that can easily be run immediately after collecting a fuel, bottoms-water, or mixed sample. The second, quantitative polymerase chain reaction (qPCR) is a laboratory test that is still too complicated for use by field technicians. I’ll discuss qPCR in my next post.

ASTM D8070 is based on a test kit manufactured by Conidia Biosciences Ltd ( The lateral flow device (LFD) technology is very similar to that used for pregnancy tests. A few drops of test fluid are dripped into a well at one end of the LFD, the fluid wicks across a filter pad, and the fluid reacts with an antibody cocktail (a mixture of antibodies designed to detect the target microbes). The kit has six LFD assembled as three pairs on a panel (fig 1). The top pair detects bacteria. The middle pair detects fungi, and the third pair detects one particular type of fungus. Within each pair, one LDF detects high antigen concentrations and the other detects low antigen concentrations. For each microbe category the results are negligible, moderate or heavy (fig 2). Each LFD has a control line (always visible) and a test line (visible only if the antigen concentration is below the LFD’s detection limit). If target microbe molecules (antigens) are present in the sample, the test line remains invisible. Thus, figure 2, shows that the sample had heavy bacterial and moderate fungal contamination levels, and insignificant Hormoconis resinae levels. Historically, some fuel microbiologists were convinced that H. resinae was the predominant microbe that contaminated fuels. I suspect that the D8070 kit has retained the H. resinae LFDs for nostalgic, rather than technical reasons.



Fig 1. Unused ASTM D8070 panel showing red, control line in each of the six LFD.


Fig 2. Used ASTM D8070 panel. Top row: heavy bacterial contamination (no visible test line on either LFD); middle row: moderate fungal contamination (visible test line on left LFD, but not on right); bottom row: negligible Hormoconis resinae contamination (test lines visible on both LFD).

In 2015 an ASTM interlaboratory study was run on fuel and bottoms-water samples. The samples were tested by ASTM D7687 (Method for Measurement of Cellular Adenosine Triphosphate in Fuel and Fuel-associated Water with Sample Concentration by Filtration) and D8070. A comparison of the bacterial contamination results from the two methods showed 83 % agreement. Recall from post # 12, that each microbiological test method measures something different. Culture tests use microbes’ ability to reproduce and form visible masses (colonies) on nutrient media. The catalase test measures the concentration of an enzyme that’s present in many different types of oxygen-requiring microbes. ATP testing measures the concentration of a universal energy molecule. The LFD test detects antigens that react (actually: bind with) antibodies selected for their diagnostic usefulness. Consequently, strong agreement between two different test methods provides a means of validating both methods.

No method is without its limitations. D8070 provides attribute scores (three distinct categories) rather than quantitative results. For some users, this might be sufficient. Other users might prefer quantitative results (for example CFU/mL or pg ATP/mL). Inexperienced users might see test lines that are not there (for example, if you look very closely at figure 1, you can see a very faint test line on the unused, top right LFD), or miss lines that are. Still the method is useful for quickly distinguishing between bacterial and fungal contamination. In selected cases, I find it to be a great tier 2 test. If I detect sufficient ATP, I’ll run D8070 to determine whether the ATP is bacterial, fungal, or a combination of both.

If you’d like to learn more about fuel system microbiology, please contact me at


In September, I attended two conferences; each of which included a half-day, fuel microbiology session. Although most of the folks presenting fuel microbiology papers were onboard for both conferences, the information overlap was minor. My overall take home lesson is that when it comes to fuel microbiology, we are all still like the five blind men attempting to describe an elephant (if you are not familiar with this ancient, Indian parable, I invite you to look it up).

Although it has been more than 120 years since the first peer-reviewed paper about fuel biodeterioration was published, there is still much we do not understand. The papers presented at the International Biodeterioration and Biodegradation Society (IBBS17) conference during the week of 04 September and the International Conference on the Stability and Handling of Liquid Fuels (ICSHLF15) the following week shed new light on old questions. At the same time, they highlighted the need for more research.
In a nutshell, in my decades of investigating fuel system biodeterioration, I have often detected substantial microbial communities in tanks that showed no evidence of damage. Just as often, I’ve detected considerable damage in systems that seemed to have negligible microbiological contamination. We might just be getting to the point where we can reasonably investigate why some populations cause damage and others don’t. I’ll get to that in a bit.

For those of you who don’t have the patience or inclination to read this entire post, I’ll start with the highlights:
     1. Anaerobic fuel biodeterioration is an important, but often overlooked component of the overall fuel biodeterioration picture.
     2. Sulfur concentration has no impact of fuel biodegradability.
     3. New test methods, still under development, hold tremendous promise to improving our understanding of fuel and fuel system biodeterioration mechanisms.
     4. Fiber-reinforced-polymer biodeterioration is real.

I had the honor of being the keynote speaker, kicking off the IBBS17 session on fuel microbiology. My presentation focused on just how critical sampling is if microbiology data from fuel systems is going to be either relevant or meaningful. During the session, Prof. Joe Suflita (University of Oklahoma) presented the results of studies he and his team have done on microbiologically influenced corrosion (MIC) caused by anaerobic bacteria (anaerobes are microbes that grow only when there is no oxygen present). His two take-home lessons were:
     1. Anaerobes growing in seawater-ballasted diesel tanks cause MIC; and
     2. The fuel’s sulfur concentration (HSD to ULSD) does not affect, microbial growth, fuel biodeterioration, or MIC risk.

Next, Dr. Oscar Ruiz (Air Force Research Lab – AFRL, Dayton) summarized his recent work on genomic (techniques that profile microbial communities, based on the types of genetic material present and the relative abundance of each unique type of microbe – based on its unique genetic profile – in a sample) and metabolomic (techniques that determine which genes are turned on and which are turned off) testing of fuels and fuel-associated waters. Per my comment earlier in this post, it’s not unusual to detect heavy contamination, but not see evidence of biodeterioration. I suspect that as metabolomic testing becomes more practical to run on lots of samples, we will gain a critical understanding of the triggers that cause some microbial populations to cause damage and other to be benign. As an aside, I’ll note here that understanding these triggers has become a major focus of human and animal disease research. More often than previously understood, we get sick when microbes on which we normally depend turn rogue. The next great leap in microbiology will be to understand what genetic switches are turned on or off. After that, the key will be to learn what triggers these switching actions. I am very excited about the work that Dr. Ruiz is doing at AFRL.

Mr.Graham Hill (ECHA Microbiology, Ltd.) reviewed his Energy Institute sponsored work on the relationship between water and microbial contamination levels in biodiesel blends. Graham and his colleagues looked to the effect of fatty acid methyl esters (FAME) on dissolved, dispersed, and free water. Importantly, they found a critical relationship between dispersed and free water, and bioburdens. Microbial loads did not increase as dissolved water concentration increased. Only once fuel-associated water became biologically available, did bioburden increase. These results weren’t surprising, but it is always great to see hard data that support conventional wisdom.

Prof. Ji-Dong Gu (University of Hong Kong) shared some of the work he had done as a post-doctoral fellow at Harvard, in the early 1990’s. This U.S. Air Force sponsored research is still the most comprehensive study of fiber-reinforced polymer (FRP) biodeterioration that has been published (Prof. Gu has several peer-reviewed papers covering this work; several years ago, the Fiberglass Tank and Pipeline’s attorney demanded that I remove all reference to FRP biodeterioration from the BCA website Prof. Gu’s research demonstrated that a diverse range of polymers and fibers (including several that had biocides blended into the polymer) were susceptible to biodeterioration. He showed a number of very elegant electron microscope images that illustrated the attack mechanism. He also presented electrical impedance data that demonstrated that FRP lost structural strength as biodeterioration progressed.

Dr. George Dodos (Technical University of Athens) presented data demonstrating that FAME composition affected both the rate and specific nature of biodiesel (B5) biodeterioration. His work built on previous studies that showed similar results. Biodeterioration is more rapid when FAME molecules have more carbon-to-carbon (C=C) double bonds (this is called: degree of unsaturation). Dr. Dodos’ research focused on examining the chemical changes that occurred in different biodiesel blends. Publications of this sort of corroborative research is essential to scientific progress.

Prof. Egemen Aydin (Istanbul University) wrapped up the IBBS17 fuel microbiology session with his paper on the biodeterioration of water-soluble molecules in navy fuels. As many readers know, the refining processes used to produce LSD and ULSD adversely affect fuel lubricity, oxidative stability, and corrosivity. Although they are primarily fuel-soluble, additives used to restore these properties have some water solubility. Consequently, they are nutrients for microbes growing in fuel-associated water. Prof. Aydin’s presentation illustrated how water-soluble fuel molecules can stimulate bioburden development and biodeterioration.

ICSHLF15 followed immediately on the heels of IBBS17. As I noted above, many of the same actors attended and presented papers at both conferences. I’ll only mention the presentations that were unique to the ICSHLF15 fuel microbiology session.
Dr. Giovani Cafi (Conidia Bioscience, Ltd.) presented research being done at Conidia using genetic tools to detect and quantify anaerobic microbes in fuels and fuel associated waters. As I noted, apropos of Prof. Suflita’s IBBS17 presentation, except for sulfate reducing bacteria, historically, anaerobic microbes in fuel systems have been largely overlooked. Dr. Cafi reported that anaerobes are commonly part of the fuel microbiology community. Clearly, more research is needed to better understand what anaerobes are present and how they contribute to both product and system biodeterioration.

Mr. Gareth Williams (EHCA Microbiology, Ltd.) discussed EHCA’s recent investigations in which they compared the results of different fuel microbiology test methods. Not surprisingly, Mr. Williams reported that culture tests do not covary strongly with non-culture tests. As I discussed in my December 2015 blog post (, any single culture test is unlikely to detect >0.1 % of the different types of microbes present in a fuel or fuel-associated water sample. Put another way, the culture tests typically used for routine condition monitoring are 99 % likely to miss microbes that are present in samples, but won’t grow in the nutrient recipe used to manufacture the culture test kit. Unfortunately, as a manufacturer of culture test kits, ECHA presents methods comparisons as though culture test data represent a gold standard for microbiology testing. Conversely, in my own experience, I have routinely detected heavy microbiological contamination by non-culture methods in samples that appear to be microbiologically clean, based on culture test results. Interestingly, the ECHA data set indicated that ATP data obtained using a method other than ASTM D7687 appeared to have no relationship to other measures of microbial loads.

In addition to the oral presentations there were several noteworthy ICSHLF15 posters that addressed fuel microbiology issues.
Dr. Joan Kelly (Conidia Bioscience, Ltd.) presented the results of a survey that she and her collaborators performed on microbial contamination in U.S. retail site UST. The team collected samples from UST across two states. Not surprisingly (to me), Dr. Kelly’s team detected moderate to high levels of microbial contamination in most of the sampled UST.

Dr. Marlin Vangsness (University of Dayton Research Institute) presented a poster reporting bioburden in bulk storage tanks. Dr. Vangsness reported that most sampled tanks had moderate to heavy microbial contamination. Moreover, he reported that ATP data obtained using ASTM D7463 did not correlate with other microbiological parameters. Having spent 30 years working to separate interferences that had historically made ATP testing unusable for complex, organic chemical rich fluids like fuels and lubricants, I have argued that ASTM D7463 is an unreliable test method. D7463 does not separate water-soluble organic chemicals and salts from microbes in the test specimen. Consequently, ASTM D7463 results are subject to both positive (high values caused by chemical reactions) and negative (low values caused when chemicals in samples capture the light that is generated by the test reaction – see – before the light reaches the detector) interferences. Dr. Vangsness’ poster and Mr. Willams’ presentation both corroborate findings that National Research Defense Canada presented at a NATO conference, nearly a decade ago. In contrast to ASTM D7463, ASTM D7687 (see the previous hyperlink) effectively separates interfering chemicals from microbes before extracting ATP.

Speaking of ATP, Ms. Chrysovalanti Tsesmeli, a doctoral candidate at the Technical University of Athens, presented a poster reporting her use of ASTM D7687 and chemical analysis to explore the effects of FAME and hydrogenated vegetable oil (HVO) on marine diesel fuel biodeterioration. Her work showed that HVO-blended fuels were more biostable than FAME-blended fuels.
Last, but not least, Ms. Silvia Bozzi (Chimec S.p.A.) presented a poster that was similar to Dr. Kelley’s. Ms. Bozzi reported on a survey of retail UST in Italy. As in the U.S., the incidence of moderate to high microbial contamination levels in Italy’s retail site UST is considerably greater than generally recognized by site owners and operators.

One of the particularly gratifying aspects of both conferences was the number of young (i.e. under the age of 40) researchers who are investigating fuel microbiology. These young scientists are applying new techniques to ask new questions and to obtain answers that we cannot get using traditional microbiology methods. Moreover, often the young researchers come from non-microbiology disciplines. Because this reflects a multidisciplinary approach to fuel and fuel system biodeterioration it bodes well for the future of fuel microbiology.
Although I didn’t mention any posters or presentations made by Prof. Fatima Bento or her graduate students, I’ll close this blog with a special call out acknowledging the great research being done by this group at Instituto de Ciências Básicas da Saúde, Sao Paulo, Brazil. Few months pass when I don’t have an opportunity to review manuscripts submitted by members of Prof. Bento’s team. They have made many important contributions to our understanding of ULSD and biodiesel biodeterioration. I’d be sorely remiss if I didn’t mention their fine work.

As always, if you’d like to learn more about fuel and fuel system microbiology testing, please contact me at


Let’s pick up with: “If no method provides a perfect measurement of microbial contamination, which one should I use?”

Currently, the primary microbiological test that I use for testing fuels, fuel-associated water and fuel system components is ASTM D7687 Test Method for Measurement of Cellular Adenosine Triphosphate in Fuel and Fuel-associated Water With Sample Concentration by Filtration. The ASTM method is based on a test kit manufactured by LuminUltra Technologies, Ltd.; with whom I collaborated to develop and validate the method.

Adenosine triphosphate (ATP) is the primary energy transfer molecule in all metabolically active, living cells. Living cells can be metabolically active or dormant. When a cell is metabolically active, it carries out all of the activities we use to define life: respiration, ingestion, excretion, response to stimuli, etc. Many (if not all) microbes can shut down – become dormant – when environmental conditions are unfavorable (for example, while they are suspended in fuel). Recently, researchers have discovered viable microbes that have remained dormant for 10’s of thousands of years. The key here, is that ATP is not detectable in dormant cells. Consequently, ATP testing does not detect dormant cells (more on this in a bit). ATP testing is based on the detection of light given off by a specific reaction, unique to fireflies and glow worms. This is the reaction that produces the characteristic yellow-green firefly light (Fig 1)

Fig 1. Firefly with its tail aglow.

In 1954, microbiologists first reported a method for using firefly tail extract to measure microbial population densities in water samples. By the late 1960’s it was known that bacteria had – on average – 1 x 10-15 grams (1 femtogram; 1 fg) of ATP per cell. Even though the actual amount of ATP per cell was quite variable, the 1fg/cell ratio soon proved to be quite reliable. In the mid-1970’s I ran thousands of ATP tests on ocean water samples.

However, when I attempted to use ATP to test oilfield produced water samples, I learned about interferences cause by salts and organic chemicals. High salt concentrations (brines) and some organic chemicals depressed test results – indicate that the ATP-biomass was less than it actually was. Other organic molecules emitted light (i.e. were chemiluminescent) that gave false positive test results in samples that had few metabolically active microbes. The bottom-line was that I could not use ATP for testing microbial contamination in complex organic fluids such as fuels, lubricants, or metalworking fluids. Between 1980 and 2008, a number of ATP test kit manufacturers made claims about the stability of their kits for testing fuels, but the QGO-M method – developed between 2008 and 2009 – remains the only one that is not affected by either the salts or organic chemicals that make other ATP tests unusable for fuel or fuel-associated water testing.

In most samples, ATP can be found in three states (Fig 2). ATP within whole cells is called cellular ATP (cATP). ATP attached to cell fragments and dissolved ATP are nominally detected as dissolved ATP (dATP; it would probably be less confusing to call dATP extracellular ATP). Most commercially available test methods start by breaking open cells. Consequently, the results they produce are total ATP (tATP):

[tATP] = [cATP] + [dATP]

Where [tATP], [cATP], and [dATP] are the concentrations of total, cellular, and dissolved ATP, respectively.

Fig 2. ATP in fuel and fuel-associated water samples: a) cellular ATP – ATP within whole (intact) cells; b) dissolved ATP – individual ATP molecules in solution; c) cell fragment ATP – ATP that is bound to pieces of cell wall from cells that have broken apart. Because there is no easy way to differentiate between (b) and (c), the term dissolve ATP generally refers to the sum of (b) and (c).

The QGO-M method washes away dATP before breaking open cells to release ATP for detection. This means that it measures what’s important: the [cATP] in whole cells. In contrast, test methods that give only [tATP] results cannot determine whether the detected ATP came from whole cells, cell fragments, dissolved ATP or some combination of the three.

The QGO-M method doesn’t provide all the answers about fuel and fuel system microbial contamination, but it’s my go to, first test. First, I can run the test out of the back of my car. Second, I can complete a test in less than five minutes. Third, I get results as actual ATP concentrations ([ATP]). The other commercially available test kits give results in instrument specific relative light units (RLU; more on this in a bit). Fourth, as explained above, the QGO-M method detects [cATP]. The other test [tATP]. Consequently, ASTM D7687 is the only ATP test method that provides reliable test results for fuel and fuel-associated water samples.

Coming back to RLU: as I mentioned above, the ATP test measures the intensity of light produced by a biochemical reaction. The device used to measure the light intensity is called a luminometer. Basically, it counts the number of photons (light emissions). The unit of measurement is RLU. The RLU is different for every luminometer. Unless RLU are converted to [ATP] data from different luminometers cannot be compared. The QGO-M method includes determination of the RLU produced by an [ATP] = 1 x 10-9 pg mL-1. It also provides an equation for converting test sample RLU to pg mL-1. Thus, numerous technicians each collecting data using their own luminometer can share data. Moreover, the ATP data can be pooled into large databases so that the relationships between [ATP] and biodeterioration can be better understood. I consider this to be a major advantage over other ATP tests.

I use QGO-M data for two purposes:

  1. As a primary microbial contamination screening tool; and
  2. As a first-tier test to tell me whether I should run additional microbiology tests (see Part 8).

Table 1 shows the criteria I use for screening fuel and fuel associated water samples. For routine condition monitoring and predictive maintenance (PdM), [cATP] data alone are sufficient to guide PdM actions. When results are in the negligible zone no additional PdM is needed. When they are in the moderate zone, it is time for preventive action, and when they are in the heavy zone, only corrective action will suffice (I’ll discuss preventive and corrective actions in a future blog post).

Table 1. QGO-M test method PdM criteria.

Part 15, I’ll discuss how to use ATP to detect and quantify biofilms (microbiological growth on surfaces). In Part 16, I’ll discuss using these data to determine what additional microbiological tests are needed. In the meantime, if you’d like to learn more about fuel and fuel system microbiology testing, please contact me at


In Part 13, I discussed culture testing. One of the points I made was that any given culture test (of which there are >5,000) is unlikely to detect >1 % of all of the microbes present. Before moving on to discuss methods that detect more of the microbes present – in terms of percent detection of each type of microbe and the fraction of the different microbes present that are detectable – I will invoke one of Donald Rumsfeld’s most famous quotes:

“There are known knowns. These are things we know that we know. There are known unknowns. That is to say, there are things that we know we don’t know. But there are also unknown unknowns. There are things we don’t know we don’t know.”

Although, in February 2002, when Secretary of Defense Rumsfeld offered this statement, he was discussing the possibility that Iraq had weapons of mass destruction; he could just as well been talking about microbial contamination condition monitoring. In Part 12’s fig 1, I indicated that genomic testing (you’ll have to wait until Blog Post 15 or 16 for more on genomics) detected a greater proportion of the total microbiome (all of the microbes present in a particular environment) than any other method currently available. However, I also noted that I doubted if current genomic testing detected more than 80% of a given microbiome. This begs the question: “If no method provides a perfect measurement of microbial contamination, which one should I use?”

The perhaps ungratifying answer is: “It depends on your intention.” Let’s start with an illustration. Fig 1 illustrates three ways to take a measurement. You can use a ruler or tape measure to determine an object’s dimensions. If It is a liquid, you can use a measuring cup or graduated cylinder to determine its volume. You can also use a scale to determine its weight. Each of these is a valid measurement, but each provides different information.

Fig 1. Three different ways to measure.


It’s the same thing with testing from microbial contamination. Each method that I illustrated in Blog Post 12, figure 1, provides useful information about the microbial population, but each provides different information. If you need to have pure cultures of microbes, on which to do research, culture testing is the most appropriate tool. If, however, you want to quickly determine how heavily contaminated your system is, then one of the chemical microbiology test methods is a better choice.

A chemical microbiology test method is a method that detects specific molecules that are either part of or are produced by microbes. The three chemical microbiology methods illustrated in Fuel Microbiology Part 12 are: catalase activity, adenosine triphosphate concentration, microbial antigen detection.

Today, I’ll write about the catalase test. In the interest of full disclosure, in the early 1980’s, after a University of Houston graduate student developed the HMB catalase test method (, I did most of the method validation for a variety of industrial applications. I also developed ancillary HMB tests to verify that the test results were due to microbes. Starting in 1982, and for the next 27 years, the HMB was my primary field test for detecting and quantifying microbial contamination in industrial fluid systems.

The catalase test is based on the reaction between the enzyme catalase and hydrogen peroxide. Catalase is the enzyme that made life in an oxygen-rich atmosphere possible. Cells that grow in normal air (aerobes) produce hydrogen peroxide as part of their energy metabolism. Catalase converts that hydrogen peroxide into water and oxygen. What makes the HMB test quantitative are its two primary components: a patented, electronic pressure gauge (figure 2a) and a stoppered reaction tube (figure 2b).

Fig 2. HMB catalase test system. a) pressure measurement device; b) stoppered reaction tube


The HMB pressure gauge is unique because there’s very little volume between its probe and its sensor.

The stoppered reaction tube provides a fixed volume, so that headspace pressure increases as the concentration of oxygen gas increases within that space (the head space is the space between the top of the liquid and bottom of the stopper).

To run the test, add a standard sample volume (typically either 3 mL or 10 mL) to a reaction tube, and then add concentrated hydrogen peroxide (one drop – = 0.05 mL – per mL of sample). Quickly replace the tube’s stopper (it is a septum cap that re-seals itself after it has been pierced with a needle) and briefly vent the tube. This ensures that the headspace pressure is 0 psig when the reaction starts. If there are aerobic microbes in the sample, they will race to convert the hydrogen peroxide to water and oxygen gas, before the hydrogen peroxide kills them. In the meantime, as oxygen is produced, it accumulates in the reaction tube’s head space. The universal gas law teaches that if temperature and volume are constant, the pressure in an enclosed space is proportional to the concentration of gas in that space. Simply put: the more catalase enzyme in the sample, the more oxygen in the headspace; the more oxygen the greater the pressure increase (fig 3). The reaction runs its course in <15 min. At 15 min, stick the reaction tube with the needle that’s attached to the pressure gauge (fig 1a) and read the psig. The psig reading at 15 min is proportional to the microbial contamination load. Correlation between culture test data and HMB catalase test data is generally very strong.

Fig 3. Catalase reaction with hydrogen peroxide in reaction tube. a) negligible contamination = negligible oxygen accumulation = negligible pressure increase in headspace; b) heavy contamination = substantial oxygen accumulation = large pressure increase in headspace.


However, the HMB test has its limitations. First: it only detects organisms that have the catalase enzyme. This excludes all anaerobes (microbes that only grow in oxygen-free environments) and aerobes that don’t have a complete catalase enzyme. Second: dissolved iron reacts with hydrogen peroxide to release oxygen gas. Samples with dissolved oxygen will appear to have microbial contamination. Third: at ∼25 psig the pressure is sufficient to launch the reaction tube’s stopper. The noise can be disconcerting and flying stoppers can be eye hazards. Moreover, the foam pouring over the reaction tube’s wall creates a mess. When microbiological contamination is negligible, it generates <1.5 psig pressure. Heavily contaminated samples (many bottoms-water samples) will foam over before the reaction tube’s stopper can be put in place (have you ever seen the reaction when sulfuric acid is poured over a sugar cube; fig 4?). When this occurs, the sample must be diluted to get a quantitative test result. On the few occasions when curiosity has compelled me to get a quantitative answer, after observing a violent reaction in the original sample, I’ve found that the actual psig was 20,000 to 30,000 psig (yes, I had to dilute samples 10,000 to 50,000-fold in order to get a psig reading). Normally, either being unable to get the stopper onto the tube, or having the stopper launch before the end of the 15 min test period, provide the information I need to determine that the sample is heavily contaminated.

Fig 4. Column of sugar charcoal formed after adding sulfuric acid to sugar. The reaction is violent and exothermic (give off lots of heat).


Earlier, I mentioned that I had developed ancillary tests for the HMB catalase test. One is used to determine if dissolved iron is producing a false positive result. The other is used to inactivate any enzymes in the sample. When testing unknown samples (i.e.: I don’t know whether they sample is likely to have dissolved iron), I run four tests: hydrogen peroxide (H2O2) only, H2O2 + a chelating reagent (prevents the dissolve iron reaction), H2O2 + a poison (inhibits catalase activity), and H2O2 + chelating reagent + poison (serves as a background control). The H2O2 result tells me if there is a contamination issue. If the chelating reagent reduces the psig by >90 %, then the psig observed in the H2O2 only test is due to dissolve iron. Similarly, if the chelator has no effect but the poison reduces the psig by >90 %, then the psig observed in the H2O2 only test is due to microbes. If both the chelator and poison are needed to reduce the psig by >90 %, then the sample has substantial concentrations of dissolved iron and microbial contamination.

With all of these limitations, why use the HMB test? The truth is, for those 27 years during which I relied on it, the HMB test was the best test available for my specific objectives: to be able to obtain a sample and obtain reasonably reliable, quantitative microbiological data, quickly (15 min), near the point of sampling. These days, I compare the test method to early portable phones and so-called laptop computers (the former weighed in at > 10 lb., and the latter at > 20 lb.) At the time they were introduced, they did their respective jobs better than anything else available. I hope that you are now wondering: What test replaced the HMB test? That will be the topic of Part 14. Stay tuned…

In the meantime, if you’d like to learn more about fuel and fuel system microbiology testing, please contact me at


Since November, this series has progressed through fuel system sampling, sample handling and non-microbiological tests used to detect biodeterioration. This post, and the three to follow, will cover microbiological testing.

Let’s take another look at the figure (fig 1) that accompanied Part 3 (December 2016):

Fig 1. Ability of different microbiological test method to detect all microbes present in a microbiome.

The largest circle represents the total microbiome – all the microbes present in a particular environment. The parameters portrayed in the figure are not exhaustive. For example, the figure does not include direct counts: the use of a microscope to examination samples and count the number of microbial cells per microscope field (the area visible when looking through a microscope’s lenses). Nor does it include tests that measure the concentration of building block molecules such as proteins, carbohydrates, or fatty acid methyl esters (FAME). Direct counting is labor intensive. Moreover, it can be difficult to distinguish between microbes and microbe-size inanimate particles. Finally, there is considerable debate about whether direct counting includes both live and dead cells. Although, theoretically, direct count methods detect 100% of the total microbiome, direct counting is rarely used in practice.

Culture testing is currently the most commonly used tool for determining bioburdens in fuel and fuel associated water samples. Culture testing depends on microbes captured in a sample to be able to reproduce (proliferate) either in or on the growth media used to perform the test. The growth media can be either solid (ASTM Practice D7469), semisolid (ASTM D7978), or liquid (for example: LiquiCult Test Kits – LiquiCult is a trademark of MCE, Inc.; First developed in the late 19th century to detect disease causing microbes, culture testing is now often used without any real understanding of its real purpose or its limitations.

To produce a visible colony (mass of cells), a microbe must reproduce. A generation is the time required for a population to double: for one cell to become two; two to become four, etc. A visible colony has at least 1 billion cells. It takes 29 generations to get from one cell to a billion cells (fig 2).

Fig 2. Microbe proliferation from individual cell to visible colony.


To reproduce, a microbe must have the right nutrients and environmental conditions. There are more than 5,000 different recipes for microbiological growth media. Each one is optimized for the nutrient requirements of specific types of microbes. No individual type of microbe will grow on all media. Additionally, different microbes have unique preferences for growth conditions (atmosphere with oxygen present versus oxygen-free atmosphere; acidic, neutral, or alkaline environment; cold, temperate, or hot – >40 °C/104 °F; etc.). Consequently, the 1% recovery estimated in fig 1 doesn’t reflect the detection power of all test method combined. It reflects the sensitivity (actually: insensitivity) of any individual culture method. If an analyst ran thousands (perhaps millions) of different combinations of growth media and conditions, the combined results might detect 50% to 60% of the total microbiome population. There are still many microbes that we do not know how to culture.

In addition to the selective effects of any combination of growth medium and incubation conditions, time affects culture test sensitivity. Known microbe generation times range from 15 min for the fastest growing bacteria to 30 days for the slowest. The fastest growing microbes can proliferate from single cells to visible colonies in less than a day. A microbe with a 4h generation time needs nearly five days to from a visible colony, and one with a 30-day generation time needs nearly 2.4 years! Most commercial test kits recommend observing colonies daily, for up to three days. Any microbe with a generation time longer than 2h is unlikely to be detected. Analysts testing samples contaminated with microbes that have generation times of >2h will incorrectly conclude that the samples are uncontaminated.

Notwithstanding these limitations, culture testing has been used with reasonable success for more than a century. It remains the only tool available for obtaining pure cultures on which to do additional testing. Consequently, the take home message is not to dismiss culture testing. Rather it is to recognize that culture testing has specific uses. Obtaining an estimate of total levels of microbial contamination (i.e.: bioburdens) is not one of them. In the next several blog posts, we’ll look at tools better suited for that purpose.


  • Consulting Services
  • Condition Monitoring
  • Microbial Audits
  • Training and Education
  • Biocide Market Opportunity Analysis
  • Antimicrobial Pesticide Selection