Dead Legs, Slopes, Materials and Hygienic Design in Pharmaceutical Water Systems

In almost every WFI loop design review the first technical question is the same: “what is your L/D ratio?”. On its own it means nothing. It means nothing until you state where the branch length is measured from, which diameter goes in the denominator and under which operating condition that geometry has to work. Two suppliers can deliver the same tee, quote two different figures, and both be consistent with their own internal convention.

The same pattern repeats with slope, velocity and surface finish: these are the four parameters most often quoted as if they were GMP requirements, and none of the four is one in the numerical form in which it is quoted. Annex 1 asks for an outcome — prevent contamination, avoid dead legs, slope the pipework for drainage, keep the flow turbulent — and sets no numbers. The numbers come from industry standards, from the design specification and from the contract, and they become binding to the extent that the site has adopted them and can demonstrate them. The reading key throughout this article is the same: first the physical and microbiological phenomenon, then who sets the criterion, then where the value has to be written down and how it is verified.

It is worth fixing the physical picture first. From a microbiological point of view a distribution system is a large wetted surface held under conditions that ought to be unfavourable to colonisation. Biofilm does not form where the water flows well: it forms where turnover is slow, where the sanitisation temperature does not reach, where the chemical agent does not reach a useful concentration, where a liquid film remains after draining. L/D ratio, slope, velocity and roughness are design proxies: measurable quantities that correlate with that behaviour and that can be put in a contract. They are not the phenomenon.

What Annex 1 requires and what industry standards quantify

This is the most misunderstood part of the whole subject. Annex 1 of EudraLex Volume 4 (C(2022) 5938 final, in operation since 25 August 2023) addresses water in clauses 6.1-6.15, and for hygienic design of the distribution there are two direct references. Clause 6.7: the water treatment and distribution system must be designed, constructed, installed, commissioned, qualified, monitored and maintained so as to prevent microbiological contamination, minimising particulates, microbial proliferation and endotoxin/pyrogen, with pipework sloped for drainage and dead legs to be avoided. Clause 6.9: flow must remain turbulent to limit microbial adhesion and biofilm formation, with flow rate established during qualification and routinely monitored. Neither contains a number.

From this follows a practical rule: when a supplier presents a number as a “GMP requirement”, the correct question is which document states it, in which edition, with what scope of applicability. For ASME BPE the numerical criteria are copyright material and must be consulted in the applicable edition, which has to be fixed contractually: BPE is revised periodically, and “ASME BPE” with no edition is not a specification.

Dead legs and branches: the phenomenon before the number

In a closed branch, flow in the main pipe generates an eddy that penetrates the branch only to a certain depth. Beyond that depth mass exchange stops being convective and becomes essentially diffusive: turnover slows by orders of magnitude. The consequences are cumulative: planktonic cells entering the branch do not leave it and find a low shear stress surface to attach to; during thermal sanitisation the far end is heated by conduction through the wall and by limited exchange with the fluid, and may reach neither the temperature nor the contact time intended; and if the branch is not drainable, stagnant water remains for the whole duration of a shutdown.

Why the L/D ratio is an ambiguous proxy

The ratio of branch length to diameter is the classic way of making the phenomenon contractable. It is useful, but it is ambiguous for three reasons that must be resolved in the specification, not in the downstream technical argument.

• Origin of the measurement. From the inner wall of the main pipe, from the projected inner surface, or from the centreline? The three conventions give different numbers for the same geometry.
• End point. To the flange, to the sealing face, to the valve diaphragm in the closed position, or to the furthest wetted surface? On a diaphragm valve the difference is not negligible.
• Diameter in the denominator. That of the branch or that of the main? On a reduction the choice changes the result.

There is also a conceptual limit that no convention resolves: the ratio is a static, geometric quantity, while the phenomenon is dynamic and thermal. It does not describe the flow regime in the main, the orientation of the branch, the operating temperature or the sanitisation method; the same geometry behaves very differently in a hot loop under continuous circulation and in a cold loop sanitised periodically. The ratio should therefore be used as a design and documentary acceptance criterion, backed by a functional verification: thermal mapping at the far end of the branch during the sanitisation cycle, drainability verification, dedicated point of use sampling. The geometric criterion says the design complies with the specification; the functional verification says the system does what it is meant to do.

Design options at the branch

There are few families of solution: point of use valves with an integral forged body and the diaphragm presented directly to the loop flow; diaphragm valves on a short branch, oriented to be drainable; points of use placed on the return leg rather than the supply; dedicated sub-loops for remote users, replacing a long branch with a circulating circuit. Each family moves the problem: the integral valve reduces the dead leg but constrains cost and maintenance; the sub-loop eliminates the branch but introduces a pump, instruments and new sampling points. It is a design decision, to be justified on the use risk of that specific point. The overall layout logic is covered in designing the distribution loop.

Slope and drainability: a functional requirement, not a number

Slope serves one purpose: allowing the system to empty by gravity towards defined drain points. It is needed before qualification (emptying after flushing and passivation), in operation (shutdowns, maintenance, filter changes) and during steam sanitisation, where condensate that does not drain creates zones that never reach the intended conditions.

The value is quantified by industry standards — ASME BPE, Part SD, covers multi-use system design including drainability — and must be carried into the design and installation specification in the contractually applicable edition. The value alone, however, is not enough:

• The design slope is the as-built cold slope. Thermal expansion, settling of supports and loads can create local counter-slopes on long runs: pipe supporting is part of the requirement, not an installation detail.
• It must be continuous and in the right direction. A run that complies on average but has a local sag does not drain: the criterion should be expressed as absence of counter-slope, not as an average value.
• The acceptance criterion is functional: the system drains or it does not. What you take into an inspection is the documented drainability test (residual volume, inspection of low points, verification after final installation of supports), not the installer's declaration.

Turbulent flow and velocity: what Annex 1 actually asks for

Clause 6.9 requires flow to remain turbulent and the flow rate to be established in qualification and routinely monitored. The requirement is on the flow regime, not on a velocity, and the difference is not a formality: the regime depends on the Reynolds number, hence on diameter, velocity, density and viscosity. The viscosity of water changes significantly with temperature, so the same velocity produces a very different Reynolds number in an ambient loop and in a hot loop, and different diameters on the same loop give different regimes at the same flow rate.

“Minimum velocity of X m/s” is therefore not a universal GMP requirement and is incomplete as a sole criterion. It is a useful design parameter — industry guidance, in particular ISPE Baseline Guide Vol. 4 “Water and Steam Systems”, discusses typical design velocities — but the verifiable criterion is to demonstrate by calculation, in the worst case (maximum simultaneous draw-off, largest diameter, least favourable temperature), that the regime remains turbulent, and then to fix the operating flow rate in qualification and monitor it. Turbulence does not prevent biofilm: it reduces initial adhesion and improves the transport of heat and sanitising agent to the extremities. It is a necessary condition, not a sufficient one.

Metallic materials: “316L” is not a specification

Austenitic stainless steel 316L is the de facto standard for PW and WFI distribution, for corrosion resistance, weldability and availability. Quoting the grade in an enquiry does not, however, define the suitability of a component: within the same designation composition varies between heats, with direct practical effects. Sulphur content influences penetration and weld pool shape in automatic orbital welding, to the point that welding components from very different heats with the same parameters can produce an asymmetric or incomplete bead; delta ferrite content, as-supplied surface condition and dimensional tolerance also matter.

With reference to the relevant ASME BPE parts — MM for metallic materials, DT for dimensions and tolerances — the following should be specified: material designation and required composition limits; material certificates with heat traceability for tubing, fittings, valve bodies and components; dimensional and ovality tolerances compatible with automatic welding; as-supplied condition and surface finish. Where practicable, requiring heat consistency across components to be joined is good engineering that reduces weld defects, not a regulatory requirement. The same discipline applies to the components that get overlooked — thermowells, in-line instrument bodies, spray balls, sampling valves — for which the industry reference is ASME BPE Part PI.

Polymers and elastomers: the components that age

Valve diaphragms, gaskets, O-rings, flexible hoses, filter housings and some instrument components are the part of the system that degrades with time and with cycles; ASME BPE Part PM covers polymeric materials. The variables to control in the specification are: compatibility with operating temperature and pressure and with sanitisation conditions, which are more severe; compatibility with the chosen sanitising agent, where ozone is the critical case because it is aggressive towards many common elastomers; extractables and leachables, with supporting documentation from the supplier; and compression set and creep behaviour, which determine sealing over time.

The operational point is that these components are not “chosen once”: they are managed across the lifecycle. You need a replacement plan based on cycles and operating hours, traceability of the installed batch, and change control when the supplier changes formulation under the same commercial part number. A diaphragm replaced with a material that looks similar but is not qualified for the thermal cycle is a recurring cause of leaks after maintenance. The consequences of the sanitising agent choice are discussed in hot, cold and ozonated systems.

Welds and joints: where most defects originate

Automatic orbital welding under inert atmosphere is the reference technique for permanent joints; ASME BPE Part MJ covers multi-use joints. The defects that matter microbiologically are not those of mechanical strength but those of the internal surface: lack of penetration, excessive concavity, misalignment creating a step, and oxidation from insufficient backing gas, which degrades the passive layer in the heat affected zone.

What to require in the specification and to insist on as a deliverable: qualification of procedures, operators and machines; test coupons at the start and end of a shift and at every change of parameters or material lot; backing gas control with residual oxygen monitoring; a weld log with a unique identifier carried onto the isometrics; documented borescopic inspection. The percentage of welds to be internally inspected is a contractual and design decision — based on accessibility, criticality of the run and coupon results — not a number fixed by GMP legislation, and it has to be justified as such. Every dismountable joint is then a potential crevice: an undersized gasket recedes into the bore creating a cavity, an oversized one extrudes into the flow. The criterion is twofold: minimise their number to what maintenance genuinely requires, and map those that remain, with gasket size and material defined and verified on reassembly.

Surface finish: why an Ra value alone is not enough

The roughness of the product contact surface influences cleanability and initial cell adhesion: a smoother surface offers fewer anchoring sites and rinses better. The relationship is not linear, however, and it is not the only variable: surface chemistry — the chromium oxide layer, free iron, inclusions — and the sanitisation regime matter at least as much. Below a certain threshold, reducing roughness further gives a marginal benefit at significant cost.

There is also a metrological limit: Ra is an average parameter and does not describe localised defects. A surface with a low Ra but a deep scratch, a weld undercut or a residual crevice is microbiologically worse than a surface that is on average rougher but free of defects. The specification must therefore contain more than a number: it needs the finish designation, the measurement method and location, the treatment (mechanical finishing or electropolishing, which beyond roughness changes surface chromium enrichment), the acceptance criterion and the required evidence. Designations and measurement methods are defined in ASME BPE Part SF; the applicable value is chosen by the site, written into the specification and verified during inspection. How these criteria enter the founding document is covered in the URS for pharmaceutical water systems.

Passivation: a process, not a declaration

Passivation is neither cleaning nor a coating: it is the chemical treatment that removes free iron and surface contaminants and promotes formation of a chromium-rich oxide layer, which is what gives the steel its corrosion resistance. The technical references are ASTM A967/A967M-25 (chemical passivation of stainless steel, approved 24 January 2025) and ASTM A380/A380M-25 (cleaning, descaling, pickling and passivation, approved 15 January 2025), which describe treatments and acceptance tests.

What must be defined and documented: the procedure applied, referenced to standard and revision; chemical agent, concentration, temperature and contact time; degreasing sequence and, where necessary, pickling; final rinse and the quality of the water used, a design choice to be justified against the intended use of the system; acceptance tests performed and results; a map of the zones treated and of those not reachable. Two points are frequently weak. Passivation must be repeated after modifications, tie-ins and welding on an existing system, and the repetition managed under change control. And initial passivation is not a permanent guarantee: rouging is a surface oxidation phenomenon that develops over time, particularly in hot systems, and must be monitored and managed as such — see biofilm and rouging: investigation and recovery.

Inspectability: what is not documented cannot be verified

An inspector does not measure a slope or count eddies in a branch: they read documents and check their consistency with the field. The package that makes hygienic design choices defensible comprises as-built isometrics with unique weld identifiers; a branch register with the declared geometry and the measurement convention used; a weld log with inspection outcomes; material certificates with traceability; surface finish and passivation documentation; drainability test records; worst-case flow regime verification; thermal mapping during sanitisation. Building this evidence through FAT, SAT and qualification is covered in FAT, SAT, IQ, OQ and PQ of water systems. Inspectability is also a physical requirement: access, dismountable sections and borescopic inspection points have to be designed in, because a technically perfect run that cannot be inspected becomes, after a few years, an area of permanent uncertainty.

Good Engineering Practice or GMP requirement?

Two conclusions. Meeting the number in an industry standard does not in itself demonstrate GMP compliance if the outcome — absence of non-sanitisable zones, drainability, turbulent regime — is not demonstrated. Symmetrically, departing from that number is not automatically a non-compliance: it is a choice to be justified with data and a risk assessment, and documented. The methodological framework is that of quality risk management applied to water systems.

Worked example: Site Delta

Illustrative example, fictitious site. Site Delta adds four points of use to an existing hot WFI loop. In design review it emerges that the supplier declares the branches compliant with the contractual geometric criterion by measuring from the centreline of the main pipe to the flange, while the site specification — written years earlier and never updated — intended measurement from the inner wall to the diaphragm sealing surface. On the same geometry the two figures differ materially: two branches turn out to be non-compliant under the site's convention.

The resolution was not a negotiation over the number. Site Delta formally declared the measurement convention in the revised specification, re-measured all branches and updated the register, modified the two non-compliant branches with integral-body point of use valves, added a functional verification to qualification (thermal mapping at the far end of the new branches during the sanitisation cycle) and verified drainability after final installation of the supports, finding a local counter-slope caused by a settled bracket. The biggest cost was not the two valves: it was the time lost before the design review, because the convention was written down nowhere.

Decision matrix: connecting a point of use

Weights should be assigned according to your own context — criticality of the user point, sanitisation method, accessibility, budget — and the column is deliberately left blank.

Common mistakes and red flags

• The number without the convention. A geometric ratio with no origin, end point and reference diameter guarantees a dispute at acceptance.
• “ASME BPE” with no edition. An unspecified reference: the applicable edition has to be fixed contractually.
• The geometric criterion as the only evidence. No functional verification that sanitisation conditions are reached at the far end of the branches.
• Slope declared and never verified. No drainability test after final installation of the supports, no management of counter-slopes.
• Velocity presented as a GMP requirement. No worst-case calculation, no flow rate confirmed in qualification and monitored as clause 6.9 requires.
• Ra as the only surface criterion. No designation, no measurement method, no management of localised defects and weld zones.
• Passivation forgotten after a tie-in, elastomers replaced without batch traceability, as-built isometrics not updated after field modifications: the branch and weld registers lose their evidential value.

If these topics are part of your daily work, the newsletter The Pragmatic GMP offers practical analysis on the design, qualification and operation of GMP systems.

Key takeaways

• Annex 1 asks for outcomes (avoid dead legs, slope for drainage, turbulent flow) and sets no numbers: the numbers come from industry standards, the design specification and the contract.
• The L/D ratio is a static proxy for a dynamic, thermal phenomenon: without a declared measurement convention it is not a criterion, and without functional verification it is not evidence.
• Slope is a functional requirement: the acceptance criterion is demonstrated as-built drainability, not the value on the drawing.
• The requirement is the turbulent regime, not a velocity: it must be demonstrated by worst-case calculation and confirmed by the flow rate established in qualification and monitored.
• “316L” is not a specification: you need composition, certificates with heat traceability, tolerances and surface condition; polymers and elastomers are finite-life components.
• Surface finish requires a designation, a measurement method and management of localised defects, not just an average Ra; passivation is a documented process, must be repeated after modifications, and does not protect against rouging over time.
• The applicable ASME BPE edition must be fixed contractually and its numerical criteria consulted in the purchased edition. The rest of the cluster is in the Pharmaceutical Water & WFI Systems hub.

Regulatory and technical references

• EudraLex Volume 4, Annex 1 (C(2022) 5938 final), in operation since 25 August 2023 — clauses 6.7 and 6.9: health.ec.europa.eu
• EMA, Guideline on the quality of water for pharmaceutical use (EMA/CHMP/CVMP/QWP/496873/2018), effective 1 February 2021: ema.europa.eu
• USP <1231> Water for Pharmaceutical Purposes, official since 1 December 2021 — informational chapter; covers system design, materials and sanitisation.
• ASME BPE — Bioprocessing Equipment. Relevant parts: SD (multi-use system design, including drainability), MM (metallic materials), PM (polymeric materials), SF (process contact surface finishes), MJ (multi-use joints), DT (dimensions and tolerances), PI (process instrumentation). Applicable edition to be fixed contractually.
• ASTM A967/A967M-25, chemical passivation of stainless steel (24 January 2025); ASTM A380/A380M-25, cleaning, descaling, pickling and passivation (15 January 2025).
• ISPE Baseline Guide Vol. 4, Water and Steam Systems, 3rd edition, September 2019 — industry guidance, not regulation; PIC/S PI 009-4, Aide-Memoire Inspection of Utilities, effective 1 January 2021: picscheme.org