Access, Not Inspection Hours, Is What Sets the Turnaround Critical Path

The schedule is mostly access, and access is mostly guessed

Turnaround schedules are presented as work: replace these trays, retube this exchanger, inspect these welds. Underneath, the durations are mostly access. A weld that takes an inspector forty minutes to examine sits behind a scaffold that takes a four-man crew two days to erect, a permit that takes a morning, insulation removal that takes another crew, and a dismantle that has to happen before the unit boxes up. Multiply that across several hundred points and the access work is the schedule.

AP-Networks benchmarking is blunt about where this goes wrong. More than two-thirds of turnarounds miss cost or schedule by 10 percent or trip after startup, 40 percent miss by more than 30 percent, and nearly half change their start date at least once. The consistent finding across that data set is that overruns originate upstream in planning rather than in the execution crews who get blamed for them.

Access estimates are the most under-planned part of that upstream work, because they have historically been unmeasurable in advance. A planner writes a scaffold allowance based on the last turnaround and a walkdown that could not physically reach half the points in question. The estimate is a memory. A model turns it into a measurement, and that single substitution is the entire argument for using a twin here.

Measuring the route before anyone buys the steel

A digital twin built from a laser scan of the operating unit carries dimensions. Scanning of congested industrial environments lands in the 1 mm to 3 mm accuracy band, which is far tighter than any access decision requires. From that model, a planner measures the things that determine scaffold quantity: elevation to the work point, clear ground footprint under it, headroom, tie-in structure, and the distance to the nearest access route wide enough to carry materials.

These measurements are otherwise obtained by sending someone up. In an operating unit that means a permit, a harness, and a person occupying the exact position that is dangerous enough to need a scaffold in the first place. Measuring in the model removes the pre-shutdown access trip that was needed to plan the shutdown access, a recursion that quietly consumes weeks of front-end schedule at most sites and appears in no plan.

The output is quantified. Bays, lifts, boards, tie-ins and a footprint drawing per location, generated while the unit is still running and priced by the scaffold contractor as a defined scope rather than a rate-based allowance. Whether the contractor's design is buildable remains a question for a competent designer. The quantities they are designing against are no longer invented, and neither is the count of locations.

Sorting every inspection point into four access classes

The most valuable single pass is classification. Every inspection point in the turnaround scope gets assigned to one of four access classes: existing permanent platform, rope access, drone, or temporary scaffold. Most sites have never done this deliberately and default to scaffold, because scaffold is what the contract covers and what the last turnaround used. Our digital twins work treats this pass as the first deliverable, because every point that moves out of the scaffold column takes erection and dismantle duration off the schedule with it.

The classification rules are concrete. Points already reachable from a permanent platform need nothing, and are routinely scaffolded anyway because nobody checked. Single readings at height with an anchor structure above suit rope access, which IRATA members apply across tens of millions of working hours a year on precisely this kind of short-duration, single-trade task. Large-area screening on shells and stacks suits contact UT from a drone.

Points that need repeated visits, several trades, insulation removal, or a stable working platform for a phased array setup stay on scaffold, and should. The honest position is that scaffold is the right answer for a large fraction of turnaround inspection scope. The gain sits entirely in removing the fraction where it was never the right answer, and that fraction is only visible when someone checks every point against the geometry.

Sequencing so a tower is erected once and used by four trades

The expensive failure in scaffold planning is not quantity. It is building the same tower twice. Inspection needs the platform on day three, insulation on day six, welding on day nine and painting on day fourteen, but the four scopes were planned by four coordinators who each raised their own scaffold request. The tower goes up, comes down, and goes up again, and the second erection was pure waste that nobody logged as waste.

In the model, requests attach to locations rather than to work orders. Four scopes pointing at the same nozzle become one access requirement with four occupancy windows, and the planner sees immediately that the tower must stand from day three to day fourteen. That single view is where the volume reductions come from. Becht reports typically around 30 percent scaffold volume reduction on turnarounds where this discipline is applied, occasionally reaching 50 percent.

Sequencing also exposes conflicts. A tower standing until day fourteen may block the crane pick planned for day eight, or the laydown route for exchanger bundles. Those collisions are geometric, which means the model can find them and a spreadsheet cannot. Finding them in planning costs an afternoon of re-sequencing. Finding them during execution costs a shift, and shift losses in a turnaround are counted in units nobody wants to write down.

Scope challenge before the freeze, with quantities attached

Scope freeze is the moment a turnaround stops being negotiable, and AP-Networks data shows scope still growing 19 percent on average after it, against roughly 7 percent for top-quartile performers. Access quantities produced after the freeze can only react to what was frozen. Produced before it, they change what gets frozen. A point costing a fifteen-metre tower and eleven days of occupancy is worth challenging. The same point costing two hours of rope access is not.

That challenge is a risk conversation, and it belongs with the people who set the inspection scope. Where the scope came from a risk-based inspection study, the question is whether the mechanism and consequence at that point justify the access cost. Where it came from a fixed interval inherited from a previous decade, the question is considerably sharper. Either way, the conversation improves when the access cost is a number instead of an intuition.

Sites without a defensible basis for their turnaround scope tend to discover it here, when someone asks why a line is being inspected at all. An NDT programme audit and gap assessment is a considerably cheaper place to find that out than the week before a freeze date. The access model exposes the weakness rather than creating it, and exposing it early is the point.

The scaffold register as a live object rather than a spreadsheet

During execution, the scaffold register is normally a spreadsheet of tag numbers, erection dates and hand-back status, maintained by one coordinator who becomes a single point of failure. It answers questions about tags. It cannot answer which towers are still standing in this area and who is using them today, which is the question that actually controls dismantle. Dismantle crews get released late because nobody can prove a tower is finished with.

Attached to geometry, the register becomes a live map: erected, in use, released, dismantled, all shown in place. A coordinator sees the towers standing in a unit, the scopes still attached to each one, and the ones with no remaining occupancy. Releasing those early recovers rental days and clears congestion for the trades still working, and congestion relief is worth more than the rental in a dense unit.

This is also where the record for the next turnaround gets created. A register that captured what was actually built, where it stood, and how long it stood becomes the baseline estimate for the following cycle. That replaces memory with measurement in the one place where turnaround estimating has always been weakest, and two cycles of that data is worth more than any single planning exercise.

Safety arithmetic the walkdown cannot do in advance

The safety case for reducing scaffold is not rhetorical. OSHA's rulemaking for its scaffold standard estimated that of roughly 510,500 construction injuries and illnesses annually, 9,750 were scaffold-related, and that at least 79 of about 924 annual occupational fatalities were associated with work on scaffolds. The agency puts the exposed population at about 2.3 million construction workers, some 65 percent of the industry.

Erection and dismantle are the highest-exposure phases, and they are the phases a scaffold nobody needed still requires in full. Removing an unnecessary tower removes the erection, the dismantle, the working-at-height hours of the crew that built it, and the objects that can fall from it while it stands. That is a hazard elimination, which sits at the top of the control hierarchy, rather than a control applied to a hazard accepted by default.

Rope access shifts rather than removes the exposure, and its record is well documented. IRATA International members reported more than 33 million working hours in 2024 across over 700 companies, with 20 reports in the fatal, major or over-seven-day categories. Drones remove human exposure at height entirely for the scope they can cover, which is screening and hard-to-reach single points rather than weld examination.

What this needs from your inspection and ERP data

Access planning needs the inspection scope before it can classify anything, which means the twin depends on whatever produces your point list. Where that list lives in an inspection software database with locations that reconcile to the model, classification is a day's work per unit. Where it lives in several spreadsheets held by different coordinators, reconciliation becomes the project, and fixing that is not optional before the model earns anything.

Downstream, the access plan has to reach procurement and scheduling. Scaffold quantities become a contract scope, occupancy windows become schedule constraints, and rope or drone scope becomes a resourcing question about who is certified, current and available on the dates concerned. Certification currency is a hard gate on both, which is why it belongs in a system such as NDT personnel certification tracking rather than on a wall chart.

Access planning is the digital twin use case with the shortest path to a number a plant manager recognises, which makes it the right place to start rather than the third phase of a platform programme. If you want the boundary between the twin and your inspection data management system drawn before anyone captures geometry, read digital twin vs IDMS first, then get in touch with the unit you would pilot on and the date of your next shutdown.

Request a consultation