Myths, Fantasies, and Rumors about Water System Design

September 1, 2006
Pharmaceutical Technology, Pharmaceutical Technology-09-01-2006, Volume 2006 Supplement, Issue 4

Water systems require a commonsense approach and good engineering rather than unquestioned acceptance.

The design of pharmaceutical water systems has always been part science and part alchemy, and unfortunately, it is not likely to change in the near future. Daily rhetoric is rife with statements from informed and intelligent people alleging design requirements that do not actually exist.

The problem is that if you do not have documentation from a reliable source that supports or contradicts these statements, they are nearly impossible to refute. As a result, they are often incorporated into system design.

Take, for example, the use of bacteria-retentive filtration in a water system. For years, variants of the following alleged statement have circulated verbally: "FDA doesn't allow filters because it considers them to be band-aids on a badly designed system." As a result, many systems were designed without filters, rooted in a fear that the US Food and Drug Administration would disapprove. Meltzer et al., after researching this very subject, wrote that current FDA staffers indicated they would not disapprove a system simply because it contained these filters (1). Was this statement ever really made?

The answer is a qualified yes, as this author personally heard it numerous times from more than one FDA employee, albeit many years ago. None-theless, it was said in conversation and never written in an approved regulatory document or guide.

Written guidelines . . . or lack thereof

As far back as FDA's 1993 Guide to Inspections of High Purity Water Systems, one can find comments that, taken out of context, could be interpreted as support for either side of the argument.

For example, in that Guide, Section "X-Reverse Osmosis," paragraph 6 states, "The systems also contain 0.2 micron point of use filters which can mask the level of microbial contamination in the system" (2). It also goes on in the same paragraph to state, "If filters are used in a water system there should be a stated purpose for the filter, i.e., particulate removal or microbial reduction, and an SOP stating the frequency with which the filter is to be changed" (2).

Hence, in spite of the urban legend, it does not appear that FDA ever took a definitively negative stance against bacteria-retentive filtration, although some individual inspectors did so, possibly based on a specific system design or set of individualized circumstances.

What's missing.The normally accepted design criteria for US Pharmacopeia (USP) water systems (purified and water-for-injection) are neither published by FDA nor are they tabularized in a recognized volume for easy review and discussion, although much is written about them by industry pundits. Nonetheless, most or all of the writings include at least some of the items listed in the sidebar, "Commonly accepted design criteria for USP water systems." Some of the items listed in the sidebar are supported in reliable resources, but many are not, and few confirmations exist in regulatory print.

Regardless of whether these items are documented in the literature or not, they are very often accepted by owners, engineers, designers, users, validation experts, and operations personnel as rules or laws.

The good news is that none of these features, properly applied, have ever compromised a water system, although some absolutists have come very close. Consider the example of the engineer who insisted an 1100-ft long, 2-in. line be pitched ⅛ in./ft even though the roof elevation was only 12 ft (it took a contractor with a jackhammer to make the point).

Microbial control. What is paramount and possibly not obvious is that of the items listed in the sidebar, every one relates either directly or indirectly to microbial control or sanitization of the system.

The reason for so much focus on sanitization is that achieving the required chemical purity for USP waters (they're basically the same for purified and water-for-injection) is fairly simple with available technologies. Keeping the system under microbiologic control, however, takes finesse.

Microorganisms are living entities and, hence, they are unpredictable in the sense that we cannot know for certain where or how fast they will grow, or when they will show up in our sample containers to make for an unpleasant day. We do know that heat is an extremely effective sanitant and many in the industry have come to believe in the virtues of ozone. Most often, however, we are unable to maintain sanitization (heated or ozone) limitlessly.

The best advice any consultant, water expert, or designer can give is to use the KISS principle of design: keep it simple and sanitary.

Once you have determined how to treat the water chemically, focus on microbial control. Most likely, you will need to expend at least four or five times more effort than you did for the chemical design.

Watch the front end. Remembering that organisms can occupy the treatment as well as in distribution is critical because the front-end of a system often provides the inoculation that causes long-term problems.

Commonly accepted design criteria for US Pharmacopeia (USP) water systems

Recently, on a job site that had been experiencing significant microbial excursions, it was noted that one environmentally aware individual attempted to reduce waste of pure water by piping the effluent from all of the slip-stream monitors (i.e., total organic carbon) back into their break tank. As a result, there were dozens of feet of translucent, small-diameter tubing that had been neatly tucked away so as to be inconspicuous. Not surprisingly, the tubes' interior had substantial growth varying from shades of green to black. Once this problem and a few other similar issues were resolved, the system quickly came back under control and returned to reliable operation. (Forgetting microbial issues for a moment, where was their change control?)

For anyone routinely inspecting systems, this type of situation is a flashing neon sign; however, the personnel at that facility missed this problem entirely and the result was that tens of thousands of dollars were spent on inspection, testing, and cleaning when the original intent was simply and practically to save a few gallons of water each day.

An issue of unique concern is cyclic sanitization. Engineers often design using this concept because they believe it will bring the system back to its original condition after each sanitization. Unfortunately, this is far from the truth and results in the biofilm buildup that is difficult to remove. Cyclic sanitization might as well be compared with Russian roulette because the odds are that you will eventually shoot yourself.

Referencing standards

In our haste and to justify our design dictates, we often reference specifications such as 3A, CGMP, FDA, ISPE, and USP along with an endless list of industrial codes, the majority of which most designers are not intimately acquainted.

One favorite document is the CGMPs, especially 21 CFR, which provides general counsel regarding the requirements for the manufacture of quality pharmaceutical products. Far too often, designers waggle their fingers at unsuspecting vendors, contractors, and even owners, threatening that failure to install an inane feature they have specified will not comply with CGMP and will result in the wrath of the entire federal government raining down upon them.

Another often-misunderstood document is the 3A sanitary standards. Originally developed by the International Association of Milk, Food and Environmental Sanitarians under the Dairy Industry Committee of the US Health Service, this set of specifications was developed for milk, cheese, eggs, and other related food products. It was the first definitive specification for the manufacture, finishing, and assurance of cleanability of sanitary equipment. Vendors who were approved to imprint their products with the 3A symbol were accepted more readily by users in the food industry.

These standards can be extremely valuable in the pharmaceutical arena if properly applied, but for the most part, specifiers fail to understand the implication of including 3A in their specifications. In fact, 3A was the underlying technical resource upon which much of today's pharmaceutical cleanliness was built, including automated cleaning systems, known as clean-in-place (CIP).

Sanitation remains the common thread throughout the codes, standards, and regulations because adulterated products are potentially dangerous to the public.

Water-system design requires features that ensure quality and cleanliness and, as a result, will comply with the generalities defined in the regulations. Because, with few exceptions, regulations do not require specific features, this can become a circular exercise.

Commonly accepted design criteria

A review of the items in the sidebar, "Commonly accepted design criteria for USP water systems" and their associated implications follows.

316LSS materials of construction. This chromium–nickel–molybdenum steel alloy is one of the most corrosion-resistant materials in the original family of 300 series stainless steel; however, 310SS and 317SS are very similar in their capabilities. In addition, more corrosion-resistant, and in some cases less expensive, super austenitic and super duplex stainless steels have been developed. Furthermore, materials such as titanium, hastelloy, and inconel offer far superior corrosion resistance and mechanical strength, albeit at a significantly higher price.

Are these materials outlawed from use in pharmaceutical product contact? No. As a matter of fact, precedent has been set numerous times for the use of specialty alloys such as hastelloy and AL6XN when warranted by conditions. Conversely, 304SS has been dairies' material of choice for decades.

Has 304SS (similar to 316LSS but with higher carbon and no molybdenum) been outlawed? It appears not, although the pharmaceutical industry seems to have set 316LSS as the standard in part because USP water systems held at elevated temperatures are notoriously aggressive.

Sanitary clamp fittings or orbital welds. Sanitary clamp fittings and orbital welds are not a panacea. Although they are often recommended, their application is not required. Even published industry guides mention the use of other joining mechanisms. Notwithstanding, many systems have experienced problems caused by poor-quality manual welding as well as from the use of threaded fittings. Although unrelated to water, one threaded system used for processing heart medication is an example of what problems could occur. The piping exploded upon disassembly because nitroglycerine had migrated into the crevices of the threaded fittings. Microbial contamination can certainly do the same.

Sanitary materials are typically tube outer diameter size. Their tolerances are more strictly controlled, thus making them a more appropriate choice for polished sanitary environments, as confirmed by the food and dairy industries. These more-precise materials lend themselves to orbital welding, thus resulting in improved reliability not only because of an inherent quality advantage, but also because of the rigor that goes hand in hand with their application.

Sanitary polish. The mechanical finishing of sanitary components improves their appearance and ostensibly their cleanability. Mechanical finishing may or may not remove surface imperfections, but it invariably damages the passive film that gives stainless steel its inherent corrosion resistance. In addition to providing an attractive and cleanable surface, electropolishing also passivates. Nonetheless, either method of finishing will most likely require preoperational passivation to ensure suitable service.

The finishing of materials for water system use can be viewed from two perspectives. If the system operates hot, microbial proliferation will be minimal. Hence, polishing, especially to high levels, provides the least added benefit. Nonetheless, heated water is more chemically aggressive such that electropolished material may indeed offer some advantages. Alternatively, cool systems may have greater potential for microbial contamination. Finishing may serve to impede the formation of biofilm and retard organism growth.

Lines sloped to drain. The sloping of lines to drain is an advantage when draining the system for maintenance or rework. For most systems, maintenance occurs 1–2 times per year with outages lasting only hours or a few days. At all other times, except for batch systems, water is continually present, usually circulating. Hence, drainability may not be as great a concern for all systems, especially small systems that could easily be blown dry with clean, filtered air. The requirement for a pitch of ⅛ in./ft is relatively arbitrary and can be relaxed, provided the pitch is measurable and has no obstructions. Such obstructions could include improperly torqued gaskets, incorrectly installed reductions, or mismounted diaphragm valves, all of which could easily cause a problem even if pitched ⅛ in./ft. The requirement for drainability of condensate from steamed systems, usually water-for-injection, is appropriate. The drainability of other systems makes good engineering sense but is not a requirement.

Recirculated loop. The recirculation of pharmaceutical water systems is probably the single most common design feature, and is alleged to inhibit microbial proliferation by agitation and constant scouring of the pipe wall while maintaining a flooded condition. Every published document recommends recirculation, although they do not typically consider it a requirement. Stagnation in dead-end systems often results in high bacterial counts. The only suitable alternatives to recirculation are flushing or repeated sanitization.

Unfortunately with flushing, the waste of expensive, treated water may not be cost-effective or ideal, depending on the system size and the facility requirements. Repeated sanitization also may be problematic because of the expense and time consumed. Additional advantages of circulation can include more-uniform sanitant distribution and more-uniform samples. The velocity of recirculation had been accepted as 3–5 ft/s for many years. Recently, however, several firms have modified specifications to require only turbulent flow. In most instances, this is a substantial departure from past practice, and technical support for this change is incomplete and counterintuitive.

Sanitary pumps. Sanitary pumps for water-system application are typically specified with a 45° discharge to avoid any air entrapment in the volute. This specification is reasonable because there is no downside and virtually all sanitary pump housings can be rotated to any position. Casing drains are specified to allow for drainage during maintenance or downtime. As noted previously, the time required for maintenance and service usually accounts for a very small portion of time and hence, may be addressed with alternative measures that cost less than drain fittings with valves, which routinely need service. Nonetheless, if the system is designed with an installed spare pump, drain fittings may be appropriate to ease switchover.

Drain fittings make housings less likely to be available from stock in an emergency and are substantially more expensive. Neither feature is required by any code, regulation, or guide, however. Many more substantial issues relating to pumps are commonly ignored such as proper seal and seal material selection, ferrite content, and microcavitation.

Filters on tanks and high point vents. Sanitary vent filters are common accessories for water-system storage tanks that filter the air entering and leaving the tank during emptying and filling. Filters allow vessels to operate at atmospheric pressure while eliminating bacteria that might enter during air inrush. Filter maintenance is of concern especially for tanks that are not rated for pressure or vacuum because a plugged filter might cause tank collapse or rupture, especially if other protective devices are not in place. Vent filters on system high points are extremely rare and should be avoided whenever possible because installation and service issues are complicated and difficult.

Vent filter ratings have ranged from 0.45 μm to the more common 0.20 μm and down to 0.10 μm or lower. Integrity testing may be required and system steaming can pose additional design constraints based on steam collapse during cool down. Issues relating to vent filters associated with ozone applications pose a unique problem because most vent filter elements have an extremely short life in ozone service. Vent filters typically are considered an absolute necessity in spite of the lack of definitive written documentation.

Use of ball, plug, butterfly, gate, globe, or disk valves. Sanitary valves typically come in only two designs, although this is not always the case. The valves most commonly specified and used for water systems are diaphragm-type valves, although most vendors of sanitary components manufacture valves (i.e., stem valves), which also are qualified as sanitary. Unfortunately, no single standard has been accepted for sanitary water system application. Ball valves, plug valves, and other valves that can trap product in their center are usually not selected because the potential for contamination is high. This, however, is not the case for clean steam systems where ball valves are the valves of choice. In addition, these valves have seats and seals that can be contaminated or wetted as the valve is actuated, and it is alleged that stems can provide for bacterial entry. These criticisms also apply to globe, gate, butterfly, and disk valves in varying degrees.

Under most circumstances, gate and globe valves obviously will not be suitable for use because of their materials of construction, finish, and connection. Nonetheless, many manufacturers of sanitary components also produce ball, plug, butterfly, or disk valves that might appear to be sanitary because they are constructed in stainless steel and with sanitary clamp ends. These valves are not banned by regulation, but are more difficult to sanitize, require more extensive maintenance on a regular basis, and usually are less robust than diaphragm valves. Typically, only 3A qualifies and differentiates valves.

Use-point fittings and valves. Use-point fittings became the norm early in the evolution of water-system design and remain an important component for bringing water as close to the point of actual use as possible. Subsequently, an alternative design using valves that are integrated into the loop has gained in popularity because it minimizes the outlet length or dead-leg. Neither of these options are mandated, and individual preference is normally the deciding factor. Use-point fittings are difficult to fabricate and polish and can be weak near the pull-out, especially for fittings with large reductions. Alternatively, GMP valves are considerably more expensive on an installed basis and are less likely to be available from stock if an emergency occurs.

Air-breaks in drains. The use of drains with air-breaks is overlooked frequently. Equipment in water systems, including use-points that flush to drain, tank drains, and even instrument drains can be easily compromised when the drain piping is extended too far into the drain or hard piped such that a back-up can be siphoned into the system. This problem occurs more often than most people realize, although it is seldom discussed because of the associated embarrassment. The requirement for this commonsense issue is not only rooted in local plumbing codes (although they may not be applicable to process systems), but also implied within the CGMP regulations.

Dead-legs. The discussion of dead-legs began many decades ago and was explored by the sanitarians developing CIP technology for dairy plants. As a result, the 6D rule was created, thereby limiting dead-legs or stagnant distances to six times the pipe diameter. Interestingly, years ago, the pharmaceutical industry adopted an almost identical rule, with little or no documentation.

Over the years, the rule has seen dozens of iterations and interpretations, however, such that the pharmaceutical industry has virtually no generally accepted limit today. Some organizations require 6D, whereas others demand 4D, or even 2D. Depending on how the dead-leg is measured, these specifications often cannot be met, thus causing frustration for vendors and especially for contractors.

Dead-legs should be minimized for various reasons, not the least of which is sanitization. Setting unachievable rules should be avoided, however, and specifications should be worded to allow for all reasonable situations.

Biofilm is unavoidable.That biofilm is ubiquitous in all natural environments is well recognized; however, the word from the biologists is that engineers (myself included) must "get with the program" and understand that biofilm also is ubiquitous in pharmaceutical water systems. In most situations, especially where cyclic sanitization occurs, this is reasonably understandable. Nonetheless, when new systems are placed into service and maintained continuously hot (85 °C), it is difficult to imagine bacterial counts substantial enough for biofilms to form. Without access to additional and specific research, we are obliged to believe this is indeed the case and plan accordingly. Because believing constitutes a worst-case scenario, there is little down-side except for cost.

No added substances. Without a doubt, the USP specifies in several places that added substances are not acceptable in many formulations, excipients (water), and other specified applications. What is not clear is what constitutes an added substance.

Based on discussions with technical staffers at USP and industry experts who participate on their committees, an added substance is an unacceptable material that remains in the final product, not something added and then removed during processing. On more than one occasion, ozone has been discounted for use in a water system because it is regarded as an added substance. This is not the case if the ozone is removed from the water before its use. On one occasion, the chlorine injection system at the inlet to a pretreatment system was removed because it was believed to constitute an added substance. Certainly, it must be verified that anything added to the water is removed. Overzealous interpretation of language, however, can be extremely debilitating.


Water systems require a commonsense approach and good engineering rather than unquestioned acceptance. Challenging convention is not irresponsible, but it is prudent not to assume those who have gone before you were fools. The issues that must be addressed during design are not unique, and many varied and appropriate solutions are at our disposal.

Joe Manfredi is the president and CEO of GMP Systems, Inc., 14 Madison Road, Unit F, Fairfield, NJ 07004, tel. 973.575.4990, fax 973.808.9201,


1. T.H. Meltzer, M.W. Jornitz, and R. Tetzlaff, "Pharmaceuticals FDA on Filters in Water Systems at Points-of-Use, Storage, and Distribution," Ultrapure Water, Jan. 2004.

2. US Food and Drug Administration, Guide to Inspections of High Purity Water Systems (Rockville, MD, 2003).

Related Content:

Information Technology