One Member Identity: Making Section Loss Roll Up Across 300 Structures

Six teams, six conventions, one inventory

A bridge owner with a few hundred steel structures does not employ the people who take the readings. Inspection is let as task orders to consultant engineering firms, often several at once by region, and the roster turns over on a multi-year procurement cycle. Each firm brings a competent team leader qualified under the national standards, its own report template, its own member nomenclature and its own habits about where on a girder web you put the probe when the report asks for remaining thickness at the bearing.

Every one of those reports is defensible on its own terms. Read one and you learn what the inspector saw and what they concluded. Read three about the same structure taken over eight years by three firms and you cannot compute a deterioration rate, because you cannot establish that any two of the readings describe the same square centimetre of steel. Read three hundred structures' worth and you cannot answer the network question at all, which is the question your asset management plan is supposed to be built on.

This is a fundamentally different failure from the one process plants have. A refinery's problem is that provenance fields are missing from an otherwise coherent record. A bridge inventory's problem is that the records are rich and internally coherent but mutually untranslatable. The remedy is not better inspection; it is an owner-defined identity and measurement schema that every deliverable is normalised into at the point of import, before the numbers are allowed to join the inventory.

What the number is actually for

In fixed equipment, a thickness reading feeds a corrosion rate, which feeds a remaining life, which sets the next inspection date. On a bridge the chain is different and shorter. A thickness reading feeds a net section, which feeds recomputed section properties, which feeds a load rating under the AASHTO Manual for Bridge Evaluation, which may feed a posting, a permit restriction or a repair prioritisation. The consumer is a rating engineer, not an interval calculation, and rating engineers need geometry rather than a scalar.

That changes what a good record looks like. Loss of two millimetres from a bottom flange at midspan and loss of two millimetres from a web at the bearing have completely different consequences: one attacks moment capacity, the other attacks shear and bearing capacity, and a web with holes at the bottom near a leaking joint may govern for web crippling before either. A record that stores the value without the location on the member cannot be used for any of these calculations, which means the inspection produced a report but not an engineering input.

There is a second consumer that people forget. Element-level condition states under the AASHTO element inspection manual are quantity-based: so many linear feet of girder in condition state 3. Thickness measurements are what justify moving quantity between condition states defensibly rather than by feel, and they are what makes the same judgement reproducible when a different team returns in two years. Tying readings to elements, not just members, lets the condition-state quantities and the measurements corroborate each other.

Pack rust and paint: two reasons a gauge lies on a bridge

Riveted built-up members are the signature structure of the older steel inventory, and they are where ultrasonic thickness is least trustworthy. A built-up girder or a truss chord is an assembly of plies clamped by rivets. Moisture works into the faying surfaces, corrosion products form between the plies, and because the oxide occupies several times the volume of the steel it consumed, the plies are forced apart. The result is pack rust: swelling, distortion, sheared rivet heads and a laminar plane that reflects ultrasound. A gauge placed on the outer face reports the outer ply and nothing beyond it, and the reading looks perfectly stable.

The honest treatment is to record what was actually measured. A reading on a built-up member needs a field stating which ply or which assembly was interrogated and whether the back face was verified. Where the section is packed, remaining section is better established by cleaning and physically measuring the remaining edge, with the measurement method recorded as caliper or rule rather than UT. That is not a downgrade; it is the correct instrument for the geometry, and it produces a number a rating engineer can defend.

Coating is the second trap and it is unavoidable on old steel. Historic systems on bridges are thick and often lead-bearing. Measuring through them with a single-echo gauge adds the coating to the wall; measuring echo-to-echo excludes it; removing them to reach bare steel triggers containment obligations under lead-in-construction rules and is a scoped, funded activity rather than something a team does on the fly. So the practical inventory contains a mixture of all three approaches. Recording which one was used on each reading is the only thing standing between that mixture and a set of trends full of phantom events.

Where the loss actually is

Section loss on steel bridges is not distributed; it is concentrated wherever water is delivered and held. The dominant driver is deck drainage. A failed or leaking expansion joint puts chloride-laden runoff directly onto girder ends, bearings, bearing stiffeners and the top of the abutment seat, and those few square metres will contain most of the measurable loss on the entire structure. Floor beam ends at the connection to the girder web are the second cluster, for the same reason. Scupper discharge and downspout failures create a third, usually on a fascia girder's outside face.

Truss and through-girder structures add their own geography. Gusset plates, particularly where debris collects in horizontal or shallow-angle connections, hold moisture indefinitely; the free edge and the region around the fasteners are where the section is lost, and gusset plate capacity became an explicit part of rating practice after the failures that prompted federal guidance on the subject. Pin and hanger assemblies, box girder interiors that collect condensation, and the interiors of closed members are all places where the loss is real and the access is poor.

In water, the pattern moves to the splash and tidal zones. Steel piling and pile jackets lose the most section in the alternating wet-dry band around mean high water, and readings can only be taken at low water, which constrains the inspection window to a few hours on specific days. Culverts, corrugated metal pipe and high-mast lighting or sign structure base plates round out the inventory, each with a characteristic loss location: the invert for culverts, the base plate and anchor region for poles. Predefining these locations as named measurement points on each structure type is what makes readings comparable across an inventory in the first place.

Exclusions that cost a permit

Bridge inspection exclusions are unusually expensive and unusually specific, which is a good reason to record them properly. The tide was too high to reach the splash zone measurement point. The under-bridge inspection unit was not available in the window, so the fascia girder underside was reviewed from the ground with binoculars rather than hands-on. Lane closure was permitted only overnight and the work ran out of window. Coating removal was not in the scope, so no bare-steel reading was possible. The member is encased in concrete. Traffic control was cancelled by the maintaining agency the morning of.

When those reasons are absent from the deliverable, the member simply has no reading for that cycle, and nothing distinguishes it from a member inspected and found sound. Over three cycles a member can accumulate no measurements at all while appearing in every report. That is a genuine safety exposure, and it is also a funding argument you cannot make: you cannot request budget for an under-bridge unit or for a coating removal package without a count of the measurements those constraints prevented.

Coded exclusions turn the constraint into a plan. If forty structures across the inventory have splash-zone points that were missed for tide in two consecutive cycles, the answer is to schedule those inspections against a tide table rather than a task-order calendar. If the recurring blocker is lead containment, the answer is to bundle the coating removal into a maintenance package covering the twenty worst girder ends. Neither conclusion is reachable from free text in a comments field.

The schema that makes three hundred structures comparable

Standardisation has to be defined by the owner and enforced by the system, because no individual consultant has either the mandate or the incentive to do it. The minimum viable schema has five layers: the structure identifier used in your inventory submittal; the span and its numbering rule referenced to ahead station so that span 1 is unambiguous; the element and member with an explicit numbering direction; a named permanent datum on that member, typically the centreline of bearing at a stated end; and the reading's longitudinal offset, transverse position and face.

Enforcement happens at import, and it has to be row-level. A deliverable arriving with three hundred readings should import the two hundred and seventy that resolve to a known member and datum, and hold thirty in a queue with a specific reason: unknown member, no datum, offset missing, coating treatment unstated. The consultant resolves the queue as part of deliverable acceptance. Done once at the start of a contract, this converts into a template the firm uses for the rest of the task order; done never, it converts into a permanent tax on the owner's engineers.

The schema also has to survive change. Members get replaced, spans get widened, structures get rehabilitated and renumbered, and a strengthening plate welded to a girder web changes the section for every future reading at that location. Each of those is an event on the member with a date, a work order and a new baseline, so that a reading taken before a retrofit is never trended into one taken afterwards. Retire history against the steel that carried it; never overwrite it.

Rolling up: deterioration rates, asset plans and posting decisions

Once identity is fixed, the inventory-level questions become computable and they are the questions that justify the programme. Which member types across the network lose section fastest, and does that correlate with joint type, deck drainage design, span configuration or age cohort? Which structures have a member on track to change a load rating before its next scheduled inspection, given a maximum routine interval of twenty-four months, or forty-eight under a risk-based interval? Which structures have members that have never actually been measured, only observed?

Those answers feed directly into a risk-based transportation asset management plan and into the deterioration models it depends on. Models calibrated on element condition-state transitions alone are coarse; models informed by measured section loss on the members that actually govern capacity are considerably better, and the difference shows up as better-targeted preservation spend. Joint replacement is cheap relative to girder-end repair, and the case for doing it early is made from exactly this data.

At the structure level the output is a rating and posting decision with a traceable measurement behind every reduced section property. When a rating engineer reduces a bottom flange by three millimetres, the record should show the reading, the technician, the certification held that day, the gauge, the coating treatment, the offset from the datum and the photograph. That is what makes the rating defensible years later, which on a public structure is not an administrative nicety. Atlantis configures this schema against your own structure numbering and inspection contracts, alongside ASNT Level III consulting, independent report validation and 3D laser scanning for as-built geometry. Affordable, accessible, fully customisable; request a demo or consultation at info@atlantisndt.com.

Request a consultation