Bolted Flange Connections Under Piping Loads: Equivalent Pressure to VIII-2 Part 4.16

A B16.5 or B16.47 flange rating covers internal pressure only — it says nothing about the bending moment and axial force a piping stress model puts on the joint from thermal expansion, weight, wind, seismic events or relief-valve thrust. Code basis: ASME B31.3-2024 and B31.1-2024 for the piping, ASME B16.5-2025 and B16.47-2025 for standard flanges, ASME BPVC Section VIII Division 1 and Division 2 (2025 edition) for flange design, and ASME PCC-1-2022 for assembly. Paragraph numbers below are from those editions — if your contract invokes a different edition, check the paragraph before you rely on it. This guide works through the methods in the order a stress engineer normally escalates: the Kellogg equivalent-pressure screen, ASME VIII-1 UG-44(b), a full VIII-2 Part 4.16 flange design, and the alternatives an owner specification may name.

First published 2026-09-23 · content last reviewed 2026-09-23 · page regenerated 2026-09-23 at build.

Written and reviewed by Matthew Norris, P.E. — active P.E. licensure in Arizona, California, Kansas, Missouri, North Carolina, Texas.

Why this check exists

Neither piping code gives a mandatory equation for it. B31.3 ¶301 — ¶301.2 through ¶301.9, including ¶301.5 Dynamic Effects (impact, wind, earthquake, vibration, discharge reactions), ¶301.6 Weight, and ¶301.7 Thermal Expansion and Contraction — lists the loads a designer must consider, and B31.3 ¶308 (flanges, blanks, flange facings and gaskets) leaves the joint evaluation itself to the designer. That gap is why several methods exist, why owner specifications pick one, and why stress reports get rejected for not stating which one they used.

A bending moment unloads the gasket on the tension side of the joint before it overstresses the flange ring, the hub or the bolts. When the gasket stress on that side drops below what it needs to seal, the joint leaks — often well within the flange's stress and bolt-load allowables. So the flange check in a stress report is really a leakage check, not a strength check.

Four methods cover the range from a five-minute screen to a full design calculation:

1. Kellogg equivalent pressure. A fast, conservative screen against the flange's pressure rating.
2. ASME VIII-1 UG-44(b). The Code's own leakage-based method for standard weld-neck flanges. Less conservative than Kellogg.
3. ASME VIII-2 Part 4.16. A full flange design calculation that carries the external loads through the bolt-load and flange-moment equations, plus the flange rigidity check.
4. Alternatives an owner may specify. NC-3658.3, EN 1591-1, and FEA.

The later sections cover the gasket and bolt data every method needs, and a worked example that carries one joint through the escalation.

Step 0 — Get the loads right before you check anything

A flange check is only as good as the loads fed into it. Before running any method:

• Take loads at the flange node, not at the nearest restraint. Use the forces and moments the stress program reports at the flange face. If the flange is modeled as a rigid element, use the end of that element facing the joint.
• Resolve into joint axes. The methods below need the axial force F (normal to the gasket face, tension positive) and the resultant bending moment M = √(M_y² + M_z²) about the two axes in the gasket plane. Torsion and shear are excluded from the equivalent-pressure methods — friction between the flange faces carries shear, and torsion is worth checking separately when it is large relative to the bending moment.
• Check every load case that matters for leakage: operating (sustained + thermal), sustained alone, each occasional case (wind, seismic, PSV thrust, hammer), and hydrotest if the joint is inside the test boundary.
• Keep the sign convention conservative. A compressive axial force helps sealing; the conservative convention is to take compression as zero rather than as a credit, unless the owner specification says otherwise.
• Record the pressure used with each case. For the design check that is design pressure; for the test case it is test pressure, checked against the flange's test capability, not its rating.

Method 1 — The Kellogg equivalent pressure screen

The equation

The M. W. Kellogg method, from Design of Piping Systems, converts the external moment and axial force into an extra "equivalent" internal pressure, added to the design pressure and compared with the flange rating:

P_eq = P + 16·M / (π·G³) + 4·F / (π·G²)

PASS if P_eq ≤ P_R

P is the design (or case) internal pressure, M the resultant bending moment at the joint, F the axial force (tension positive; compression taken as zero), G the gasket load reaction diameter, and P_R the flange pressure–temperature rating at the case temperature from B16.5-2025 or B16.47-2025 for the material group and class of the flange. The moment term 16M/(πG³) is the pressure that produces the same peak line load on the gasket circle as the moment does; the axial term 4F/(πG²) is the axial force spread over the area inside the gasket reaction diameter.

Choosing G

G is the diameter where the gasket reaction acts — it is not the gasket OD and it is not the bolt circle, and it changes the moment term with its cube. Compute it the way the flange design rules do:

• Basic gasket seating width: b₀ = N/2 for a flat ring gasket on a raised face, where N is the gasket contact width.
• Effective width: b = b₀ if b₀ ≤ 0.25 in. Otherwise b = 0.5·√b₀ (U.S. customary units).
• G: the gasket contact OD minus 2b when b₀ > 0.25 in; otherwise the mean diameter of the contact face.

The b₀ rules are in ASME VIII-1 Mandatory Appendix 2, Table 2-5.2 — retained in the 2025 edition even though the surrounding flange-stress and bolt-load clauses (¶2-5 through ¶2-14) were deleted in favor of BPVC VIII-2 Part 4.16, per Appendix 2 ¶2-1(a)(4)–(6). VIII-2 Part 4.16 carries the equivalent table. For a spiral-wound gasket, N is the width of the sealing element — the guide ring and any inner ring don't count. Take the sealing-element dimensions from B16.20-2023 for the gasket style you are actually buying.

Common error: using the bolt circle diameter for G. It shrinks the moment term by (G/C)³ — often a factor of 2.5–3.5 — and can turn a failing joint into a passing one.

Why this is conservative

The Kellogg screen treats the rating as a hard ceiling on total equivalent pressure. B16.5 ratings are set by shell and body stress, not by leakage, and a properly assembled joint has sealing margin above the rating pressure that Kellogg gives no credit for. Most flanges that fail Kellogg pass UG-44(b) or a full flange calculation. A Kellogg failure means "look harder." It does not mean "redesign."

When not to use it

• Non-standard flanges — they have no P_R to compare against. Go straight to Method 3.
• Flanges at rotating equipment where the vendor sets flange load limits. The vendor's limits govern.
• Joints where the owner specification names a different method — some owner specifications route vessel flanges directly to UG-44(b), for example.

Run it: the free Flange Equivalent Pressure Calculator implements this method and reports the utilization P_eq/P_R, with both load terms shown separately so you can see which one governs. Get P_R from the B16.5 Flange Rating Calculator.

Method 2 — ASME VIII-1 UG-44(b)

What it is

UG-44(b) is the Code's own leakage-based way to put external loads on standard weld-neck flanges without a full flange design calculation. It applies to welding-neck flanges chosen under UG-44(a)(2) (ASME B16.5), (a)(9) (ASME B16.47 — both Series A and Series B, Table UG-44-1 gives each its own moment factors), or (a)(10) (forged nozzle flanges meeting the B16.5/B16.47 dimensional-substitution rule). BPVC VIII-2-2025 carries the direct counterpart at §4.16.12, Eq.(4.16.19), with F_M from Table 4.16.12 and the same four-condition structure. UG-44(b) traces to an earlier ASME Code Case; once a Code Case's content is folded into mandatory Code text, ASME's usual practice is to annul the case, so treat Table UG-44-1 of the edition you are using as the current source for F_M — not a separately-numbered Code Case.

The acceptance criterion has the same structure as Kellogg, with one extra term:

16·M_E + 4·F_E·G ≤ π·G³ · [ (P_R − P_D) + F_M · P_R ]

M_E is the external bending moment, F_E the external axial force (tension positive), G the gasket load reaction diameter, and P_R the flange rating at design temperature — all as in Kellogg. P_D is not design pressure. The Code defines it as the vessel MAWP, corrected for the static pressure acting on the flange, at design temperature — a value that generally differs from the design pressure used in the flexibility analysis. F_M is the moment factor from Table UG-44-1, by flange standard, class and size.

The one-line insight: UG-44(b) is Kellogg plus a bonus — under one condition

Set F_M = 0 and divide through by πG³. The left side becomes exactly the Kellogg screen's moment and axial terms, and the inequality collapses to P_eq − P ≤ (P_R − P_D). That reduces to the shortcut below only when P (the pressure used in the Kellogg run) equals P_D — the common conservative case where design pressure is taken equal to MAWP with no static-head correction:

Required moment factor F_M,req = P_eq / P_R − 1

UG-44(b) passes when F_M (from the table) ≥ F_M,req

When the job's design pressure and P_D genuinely differ — a static-head correction applied, or MAWP carried above design pressure — reconcile the two pressures first: recompute F_M,req with P_D in place of P before comparing it with the table. Where they coincide, this is the natural second step after a failed Kellogg screen: compute F_M,req, look up F_M for your flange standard, class and NPS in your own copy of Table UG-44-1, and compare. Table UG-44-1's own General Note states that combinations of size range and flange class it gives no factor for are outside the table's scope — for those, go to Method 3.

Conditions you must satisfy

UG-44(b) is not a free pass; the Code text states four numbered conditions, all of which must hold:

• MAWP vs. rating. The vessel MAWP (corrected for static pressure acting on the flange) at design temperature must not exceed the flange's pressure–temperature rating.
• Assembly. The actual assembly bolt load complies with ASME PCC-1, Appendix O. A joint assembled with uncontrolled torque does not qualify — a single assembly note in the construction specification makes the difference.
• Bolting. The bolt material's allowable stress is equal to or greater than SA-193 B8 Cl. 2, at the specified bolt size and design temperature — not ambient.
• The equation. The combination of MAWP, external moment and external axial force satisfies the boxed inequality above.

Gasket type is not one of the four conditions. Spiral-wound to B16.20 is common practice on joints qualified this way, but it is not a listed UG-44(b) requirement. Table UG-44-1's General Note (b) adds one more piece of judgment: the designer should consider reducing the moment factor where the load is primarily sustained at a temperature high enough for gasket creep and relaxation to matter — that is guidance, not a mandated numeric reduction, so read what your edition says rather than applying a default cut.

Division 1 vs. Division 2 vs. piping

UG-44(b) is a pressure-vessel rule. Neither B31.3 nor B31.1 requires it for piping-to-piping joints — they don't forbid it either, and many owners accept it as a recognized, Code-published method for B31 joints. Because no B31 clause names UG-44(b) directly, state the adoption in the calculation basis: for example, "Flange leakage evaluated per ASME VIII-1 (2025) UG-44(b), F_M per Table UG-44-1, adopted by reference for B31.3-2024 piping joints per Owner Spec XXX." Method 3 below has a stronger B31.3 hook of its own.

Method 3 — ASME VIII-2 Part 4.16: design the flange for the loads

When a standard flange fails both screens, or the flange isn't a standard flange at all, run the full flange design calculation with the external loads included. B31.3-2024 ¶304.5.1(b) is the clause that permits this route for piping: it lets a flange be designed to VIII-1 Mandatory Appendix 2 or to VIII-2 Part 4.16, using B31.3's own allowable stresses and temperature limits rather than the vessel code's. Since the 2025 edition, VIII-1's own Mandatory Appendix 2 sends you to Part 4.16 directly: ¶2-1(a)(4) states that the design requirements in ¶2-4 and Part 4.16 replace those formerly carried in Appendix 2's bolt-load, flange-moment and stress clauses (¶2-5 through ¶2-14, now deleted, invoked through UG-16(a) and Mandatory Appendix 46 per ¶2-1(a)(6)). Table 2-5.1 (gasket m and y) and Table 2-5.2 (gasket seating width — the b₀ → b → G rule) remain in Appendix 2 and are cross-used by Part 4.16.

What the calculation does

1. Geometry and materials. Flange dimensions (A, B, C, g₀, g₁, h, t), bolt count and size, gasket dimensions, and allowable stresses for the flange and bolt materials at design and ambient temperatures (Section II Part D — read the values from your own copy).
2. Gasket factors and G. Gasket factor m, seating stress y, and b₀ → b → G. m and y come from the gasket manufacturer, or the Code's own table where nothing better exists (see Gasket properties below).
3. Design bolt loads. The operating load is the hydrostatic end force from pressure, plus the gasket compression needed to seal — and this is where the external loads enter. F_A (the external axial force) and 4M_E/G (an axial-force equivalent of the external moment) are added inside the required bolt-area equation, A_m = max[(W_o + F_A + 4M_E/G)/S_bo, W_gs/S_bg]. The seating load is the bolt load needed to seat the gasket, independent of the external loads.
4. Bolt area check. Required area ≥ the larger of operating and seating loads divided by the bolt allowable at the matching temperature. This must be ≤ the actual root area provided.
5. Flange moments. Operating and seating moments from the hydrostatic end force, the gasket reaction and the pressure on the flange face, each at its moment arm. The external moment doesn't enter as a flat addition here either: the operating flange moment picks up M_oe = 4M_E·[I/(0.3846·I_p + I)]·[h_D/(C − 2h_D)] + F_A·h_D, which weights M_E by the ratio of the hub's stiffness to the flange ring's — I and I_p being the hub and flange-ring moments of inertia used elsewhere in the procedure.
6. Flange stresses. Longitudinal hub stress S_H, radial stress S_R, tangential stress S_T, and their combinations, each compared with its own allowable. Read the limits in your own copy — don't carry them from memory.
7. Flange rigidity index J. The step that actually addresses leakage. J measures how far the flange rotates relative to the rotation the design basis assumes; J > 1.0 means the flange rotates too much to keep the gasket compressed, even when every stress passes. A flange can pass stress and fail rigidity — the fix is a thicker ring or a longer hub, not a stronger material.

Why Part 4.16 rather than treating Appendix 2 as a separate route

Both are legitimate under B31.3 ¶304.5.1(b), and as of the 2025 edition they are functionally one procedure — Appendix 2 now defers its calculation to Part 4.16. What Part 4.16 offers over the equivalent-pressure methods:

• it builds the external axial force and bending moment into the operating bolt load and flange moment in one consistent procedure, instead of relying on an equivalent pressure;
• the rigidity check is part of the procedure, not an add-on;
• it pairs directly with the Division 2 Part 5 design-by-analysis route if you have to go to FEA.

When the Part 4.16 calculation still fails

In rough order of cost:

1. Reduce the load. Add or move a guide or stop close to the joint, change a support's stiffness (see our support stiffness guide), or re-route to add flexibility. The cheapest bolt load is the moment you never apply.
2. Raise the joint's capacity. Go up a flange class, move to a higher-strength bolting grade if the flange and gasket can take the extra assembly stress, or change the gasket — for example from sheet to spiral-wound or kammprofile — for a lower seating stress and better recovery.
3. Change the joint type. A welded joint, a compact flange, or a clamp connector, where the owner accepts it.
4. Design by analysis. FEA of the joint with gasket contact and bolt pretension, to VIII-2 Part 5 or EN 1591-1 criteria. No PiperCalc tool runs flange-joint FEA today — see Tools for this workflow below for what PiperFEA actually covers.

Method 4 — Alternatives an owner may specify

ASME Section III NC-3658.3

This clause comes from the nuclear Class 2/3 piping rules and is sometimes adopted by reference in owner specifications, including for relief-valve discharge flanges, as a moment-only screen against high-strength bolting. Because Section III is outside this guide's source review, treat the following as the commonly cited form to confirm in your own copy before use — the coefficients, the basis for A_b (root area vs. tensile stress area) and the exact yield-ratio cap wording all need that confirmation:

Sustained / operating: M_fs ≤ 3125 · (S_y / 36,000) · C · A_b

Occasional: M_fd ≤ 6250 · (S_y / 36,000) · C · A_b (i.e., 2× the sustained limit)

S_y is the flange material's yield strength at the case temperature (with S_y/36,000 not exceeding 1.0), C the bolt circle diameter, and A_b the total bolt cross-sectional area. NC-3658.3, as commonly cited, checks moment only — it has a separate pressure requirement and does not address axial force. Because the occasional limit is exactly twice the sustained limit, a specification that says "limit flange moments from PSV discharge to twice the NC-3658 values" is using the M_fd form; that reading ties to our PSV support guide.

EN 1591-1

EN 1591-1 is the European standard for flange joint calculation. It is understood to follow the gasket stress through assembly, operation and test against measured gasket properties, with an explicit leakage class — the most complete analytical method commonly available, and one that is gaining use on projects with European owners or licensors.

Run it: the free EN 1591-1 Bolted Flange Load-Balance Screening Calculator runs the force-balance screen.

FEA

FEA is the route for large-diameter, high-moment joints, for vessel girth flanges with non-uniform loading, or where thermal transients open the joint. Model it with gasket contact and bolt pretension, then evaluate gasket contact stress around the circumference against the gasket's minimum operating stress.

Worked example — NPS 6 Class 300 weld-neck, B31.3-2024 piping joint

Every value that isn't looked up from a standard is marked ASSUMED.

1. Gasket reaction diameter. N = (8.25 − 7.19)/2 = 0.530 in. b₀ = N/2 = 0.265 in, which is > 0.25 in, so b = 0.5·√0.265 = 0.2574 in. G = 8.25 − 2(0.2574) = 7.735 in.

2. Kellogg screen. π·G³ = 1,453.9 in³ and π·G² = 187.94 in². Moment term: 16 × 24,000 / 1,453.9 = 264.2 psi. Axial term: 4 × 2,000 / 187.94 = 42.6 psi. P_eq = 450 + 264.2 + 42.6 = 756.8 psig. Utilization: 756.8 / 635 = 1.192 — fails the Kellogg screen, with the moment term carrying 86% of the added load — typical at small NPS, where bending governs.

3. UG-44(b). Taking design pressure as P_D (no static-head correction assumed for this above-grade line, so the shortcut applies exactly): F_M,req = P_eq/P_R − 1 = 1.192 − 1 = 0.192. Read F_M for NPS 6 Class 300, ASME B16.5, from Table UG-44-1 in your own copy. If F_M ≥ 0.192 and the joint meets all four UG-44(b) conditions (MAWP within rating, PCC-1 Appendix O assembly, bolt allowable ≥ SA-193 B8 Cl. 2 at design temperature, and the equation itself), the joint is acceptable — record the F_M value, its table and edition in the calc. If not, continue.

4. NC-3658.3 cross-check — owner-specification method, confirm before use. This step only applies if the project specification names NC-3658.3 as an acceptable method, and every term below needs confirming against your own copy of Section III before it goes in a calc. Bolting: 12 × ¾ in (B16.5 Class 300 NPS 6 bolting; confirm against your own B16.5 table). Bolt circle C = 10.62 in. A_b = 12 × 0.302 in² root area = 3.624 in² (ASSUMED root-area basis — confirm against your Section III edition). S_y for A105 at 400 °F: read from Section II Part D Table Y-1; if S_y ≥ 36 ksi, S_y/36,000 is capped at 1.0 (ASSUMED here). M_fs,allow = 3,125 × 1.0 × 10.62 × 3.624 = 120,272 in-lb (10,023 ft-lb) — well above the 24,000 in-lb operating moment, if the form and every term above check out against your own copy.

5. If neither 3 nor 4 closes it: run VIII-2 Part 4.16 with M_E = 24,000 in-lb and F_A = 2,000 lbf, including the rigidity index — or reduce the moment. In this example the moment is thermal: a guide closer to the joint would likely cut it substantially. Check that in the model.

What goes in the calc: method, edition, G derivation, P_R citation (standard, group, class, temperature), F_M citation if UG-44(b) is used, and the assembly requirement relied on. A reviewer rejects "flange OK" with no method stated.

Gasket properties — where the numbers come from

Every one of these methods needs gasket data. Here is what exists and roughly how much confidence each source carries.

Practical guidance:

• Spiral-wound with an inner ring is the default for B16.5 raised-face joints in hydrocarbon service. The inner ring stops the windings buckling inward and protects against over-compression. B16.20 requires inner rings for some sizes and services — read your edition's requirements.
• Kammprofile (grooved metal with facing layers) has a lower seating stress and better recovery than spiral-wound — a reasonable upgrade for joints that are marginal on leakage.
• Sheet gaskets (fiber, PTFE, graphite) seal at low stress but creep. Watch the creep-relaxation behavior at temperature, and check the crush limit if you raise the bolt load to close a moment check.

Bolt properties and assembly — turning the calc into a sealed joint

Material

Match nuts to the stud grade and temperature (ASTM A194 2H for B7, 7/4 for L7, 8 for B8). Design allowable stresses come from Section II Part D, Table 3 — read them from your own copy at design and ambient temperature.

Areas

• Tensile stress area (ASME B1.1): A_s = 0.7854 · (d − 0.9743/n)², where d is nominal diameter and n is threads per inch. Use A_s to convert a target bolt stress into a bolt load.
• Root area: the Code flange calculations (Appendix 2 and Part 4.16) use root area for bolt stress. Don't mix the two in one calc.
• Studs 1 in and larger in B16.5 flanges are usually 8-UN (8 threads per inch); smaller studs are typically UNC.

Target assembly stress and torque

ASME PCC-1-2022 Appendix O gives a procedure for choosing a target bolt assembly stress between two limits: enough gasket stress to seal after relaxation and under the operating loads, and below the stress that crushes the gasket, yields the bolt, or over-rotates the flange. This is also the assembly UG-44(b) assumes.

Convert the target stress to torque with the nut-factor relation:

T = K · F_bolt · d / 12 (T in ft-lb, F_bolt in lbf, d in in)

F_bolt = S_target · A_s

K, the nut factor, depends on the lubricant and surface condition — get it from the lubricant supplier or from PCC-1 guidance rather than assuming a value. As an illustration only: a ¾ in B7 stud has A_s = 0.7854 × (0.75 − 0.09743)² = 0.334 in². At a target of 50 ksi (a value that would need to sit within the PCC-1 Appendix O range for the actual joint), F = 16,700 lbf; with K = 0.17 (illustrative), T = 0.17 × 16,700 × 0.75/12 ≈ 178 ft-lb. K is the least certain number in the chain — a change of ±0.03 moves the achieved bolt load by roughly ±18%, which is why calibrated tools, a measured lubricant nut factor, or bolt elongation/tensioning are worth the trouble on a joint that took UG-44(b) credit.

Sequence: a star pattern in multiple passes (roughly 30%, 60%, 100%), then circular passes until the nuts stop turning — see the free Flange Bolt Tightening Sequence Calculator. ASME PCC-1 Appendix A covers training and qualification of bolted-joint assembly personnel; specify it for joints that took credit from UG-44(b).

Checklist for the stress report

• Method named, with edition: Kellogg / UG-44(b) / VIII-2 Part 4.16 / NC-3658.3 / EN 1591-1
• Owner specification clause that selects or permits the method
• Loads taken at the flange node, resolved to F and resultant M, for every relevant case
• G derived (b₀ → b → G) from the gasket actually specified
• P_R cited: standard, edition, material group, class, temperature
• For UG-44(b): F_M value with table and edition; all four applicability conditions confirmed; PCC-1 Appendix O assembly specified
• For Part 4.16: stresses and rigidity index J reported
• Hydrotest case checked against the joint's test capability, not its rating
• Remedial action recorded where a check failed: load reduced, class raised, gasket changed

Tools for this workflow

PiperFLG runs the Kellogg screen with dimensions auto-filled from the embedded B16.5 tables, the full VIII-1 Appendix 2 / VIII-2 Part 4.16 bolt-load, stress and rigidity design, and an EN 1591-1 load-balance screen, alongside a parametric 3D model of the flange. External loads enter the full design the same way they enter the screen — through the Kellogg equivalent-pressure substitution — so it does not (today) report a UG-44(b) F_M,req value the way Method 2 above does by hand.

Flange-joint FEA with gasket contact and bolt pretension isn't offered by any PiperCalc tool today. PiperFEA is a related but different tool: finite-element checks of a nozzle-to-shell junction — a cylindrical shell, flat head, or elliptical head under pressure plus piping loads — not a bolted-joint model. It has no gasket-contact or bolt-pretension inputs.

FAQ

Is the equivalent pressure method in ASME B31.3?

No. B31.3-2024 requires you to consider the piping loads that reach a flange — ¶301.2 through ¶301.9 list them, including ¶301.5 dynamic effects, ¶301.6 weight and ¶301.7 thermal expansion — but ¶308 gives no flange-leakage equation of its own. The Kellogg equivalent-pressure method is industry practice from M. W. Kellogg's Design of Piping Systems, adopted by owner specifications rather than published in the piping codes. ASME VIII-1's UG-44(b) is the Code-published leakage-based method, written for pressure vessel flanges in Section VIII, and B31.3 ¶304.5.1(b) gives piping a direct hook of its own into the related VIII-1 Mandatory Appendix 2 / VIII-2 Part 4.16 flange design calculation, permitting it to be run with B31.3's own allowable stresses and temperature limits. Neither Kellogg nor UG-44(b) is a mandatory B31.3 requirement; both are recognized ways to close a gap the piping code deliberately leaves to the designer's judgment.

My flange fails Kellogg. Do I need a bigger flange?

Not yet. Work out F_M,req = P_eq/P_R − 1 and compare it with the UG-44(b) moment factor from Table UG-44-1 in your own copy — but check the pressure terms line up first: the shortcut is exact only when the design pressure used in the Kellogg run equals P_D, the vessel MAWP corrected for static pressure at design temperature, and the two need reconciling when they differ. Many standard weld-neck flanges that fail Kellogg pass UG-44(b), provided the joint also meets its other three conditions: the MAWP-versus-rating check, a PCC-1 Appendix O controlled bolt-up, and bolting with an allowable stress at or above SA-193 B8 Cl. 2 at the specified size and design temperature. If it still fails, or the flange or gasket sits outside UG-44(b)'s scope, try reducing the moment — often the cheapest fix — before upsizing anything, or run the full VIII-2 Part 4.16 design.

Which diameter is G?

The gasket load reaction diameter, derived from the b0 → b rules in ASME VIII-1 Mandatory Appendix 2, Table 2-5.2 (retained in the 2025 edition and cross-used by VIII-2 Part 4.16). It is not the gasket OD, not the raised-face OD, and not the bolt circle — using the bolt circle in its place shrinks the Kellogg moment term by the cube of the ratio between the two diameters, often a factor of 2.5 to 3.5, and can turn a failing joint into a passing one on paper. The derivation itself is short: find the basic seating width b0 from the gasket contact width N (b0 = N/2 for a flat ring on a raised face), take the effective width b (equal to b0 when b0 is 0.25 in or less, otherwise 0.5 times the square root of b0), and subtract twice that effective width from the gasket contact OD. For a spiral-wound gasket, N is the width of the sealing element itself, not the guide ring or any inner ring — take those dimensions from ASME B16.20-2023 for the actual gasket being purchased, not an assumed value.

Do I count compressive axial force as a credit?

Conservative practice is no — take compression as zero rather than subtracting it from the equivalent pressure, unless the owner specification explicitly allows the credit and you record that decision in the calc. The reasoning is directional: a compressive axial force helps sealing in the load case where it occurs, but a piping system that sees compression in one operating mode often sees tension in another as the thermal and support conditions change, and a flange qualified only against the compressive case can still leak under the tensile one. Treating compression as zero for the axial term keeps every load case screened against the worst credible tension, which is also the convention this site's free Kellogg and EN 1591-1 calculators use by default. If a specification does allow the credit — because the load case genuinely cannot reverse — say so explicitly in the calculation basis, including which case and why, so a reviewer can see the assumption rather than infer it from the arithmetic.

Why does my flange pass stress in Part 4.16 but fail rigidity?

Because stress and rigidity are different physical questions, and Part 4.16 checks both on purpose. The stress checks — longitudinal hub, radial and tangential stress against their allowables — tell you whether the flange material is strong enough not to yield or fail under the bolt loads and flange moments in play. The rigidity index J is a separate ratio comparing how far the flange actually rotates against the rotation the design basis assumes when it sets the gasket seating stress; J greater than 1.0 means the flange rotates enough to unload the gasket on the tension side even though every stress passes comfortably. A higher-strength flange or bolt material raises the stress allowables without making the ring or hub any stiffer, so switching materials to fix a rigidity failure is a common but ineffective move. The real fix is geometry: a thicker flange ring, a longer or thicker hub, or in some cases a design that redistributes the moment — exactly the load-reduction and capacity-raising options this guide's Method 3 section works through in order of cost.

Can I use m and y to predict leakage?

No — they are design-procedure constants embedded in the Appendix 2 / Part 4.16 bolt-load and seating equations, not measured tightness data, and using them to estimate how much a joint will leak at a given gasket stress is a misuse of what they were built for. For a genuine leakage-based answer, the closer tools are EN 1591-1, which works from measured gasket parameters tied to an explicit leakage class, or a UG-44(b) evaluation, which at least ties its acceptance criterion to the Code's own moment-factor table rather than a strength-only stress check. m and y still matter — every Appendix 2 / Part 4.16 run needs them to size the bolt load and the seating stress correctly — but treat their role as setting up a strength and seating calculation, not as a proxy for how tight the joint will actually run. Where leakage rate itself is the question an owner is asking, that is EN 13555 gasket-parameter and EN 1591-1 territory, not an ASME flange-design table.