Skip to content

Proposals

What we propose, and what would show us wrong.

9 things no standard requires, which this publication argues for anyway. Each is stated once here rather than restated as a note wherever it happens to apply.

None of these is a requirement and none is numbered like one. They are named so that a reader can cite one, dated so that it is clear when it was first argued, and each carries what would refute it, because a proposal nothing could refute is a preference. Adopt them, argue with them, or ignore them.

A lifecycle ratio is unusable without three statements beside it

First argued 22 August 2026

A ratio of operating cost to capital cost carries what is inside each term, the assumed life and discount treatment, and the source of both, or it is not quoted.

Where the standards stop

ICMS 3 classifies and presents life cycle costs and carbon, and is free in full with its own Basis for Conclusions. Nothing requires a ratio quoted in a business case to say which ICMS categories it used, and widely quoted ratios such as 1:5:200 predate ICMS.

Why it earns its cost

The same building yields wildly different ratios depending on whether staff costs, renewals and end of life are inside the operating term, and over what period. A ratio without its three statements cannot be compared with any other ratio, which is usually why one is quoted.

What would show it wrong

Published ratios for comparable assets, from different authors, that agree closely without stating their basis, and keep agreeing once their bases are disclosed. That would show a working convention nobody writes down, which would make the three statements redundant rather than missing.

Sits beside: ICMS 3, International Cost Management Standard, life cycle costs and carbon; ICMS 3 Basis for Conclusions. Sources

Cite as: a-ratio-needs-its-three-statements, projects-advisors.com/proposals#a-ratio-needs-its-three-statements

Componentise to the level a decision actually reaches

First argued 25 August 2026

Beyond the separate depreciation of significant parts that IPSAS 45 requires, break an asset down only as far as a decision will actually use it: replacement, pricing, a payment mechanism or a criticality ranking. Stop there.

Where the standards stop

IPSAS 45 paragraph 41 requires each part with a cost significant in relation to the total to be depreciated separately, which sets the minimum breakdown, and paragraph 8 declines to prescribe a unit of measure. The standard does not say how far beyond that minimum a breakdown should go, so an exercise run without a stopping rule has no natural end.

Why it earns its cost

The question that stops the exercise is not how deep the asset goes but how deep any decision goes. A component nobody will replace separately, price separately or rank for criticality is a row that costs money to create and to maintain and changes nothing.

What would show it wrong

Evidence that components recorded below the level any decision used turned out to be needed: finer breakdowns that no replacement, pricing or criticality decision used at the time, but without which a later valuation, impairment test or audit could not have been done.

Sits beside: IPSAS 45, property, plant and equipment. Sources

Cite as: componentise-to-the-level-a-decision-reaches, projects-advisors.com/proposals#componentise-to-the-level-a-decision-reaches

Record the basis of a judgement, not only its result

First argued 28 August 2026

Where a standard requires a judgement but does not require evidence of how it was reached, record the basis beside the figure.

Where the standards stop

IPSAS 45 requires an amount, a useful life, a residual value and a depreciation method. Paragraph 48 requires the useful life and the residual value to be reviewed at least annually, and paragraph 58 the method. The paragraphs cited here do not require any of the four to carry the reasoning that produced it. Application guidance AG27 sets a capability test for when an asset is available for use and does not say how capability is to be evidenced.

Why it earns its cost

A review of a figure whose basis was never written down can confirm that the figure has not changed; it cannot test whether the reasons for it still hold, which is what the annual review is for. The basis costs a sentence when somebody knows it, and is hard to reconstruct a year later.

What would show it wrong

Evidence that reviews of judgements with a recorded basis reach revised or better-supported conclusions no more often than reviews without one: for example, audit findings on useful lives and residual values that are no less frequent where the basis was recorded.

Sits beside: IPSAS 45, property, plant and equipment. Sources

Cite as: record-the-basis-of-a-judgement, projects-advisors.com/proposals#record-the-basis-of-a-judgement

Evidence the traceability row by row, not only through a shared key

First argued 3 September 2026

Where two registers describe the same assets for different purposes, keep a row-level crosswalk that says which record in one corresponds to which record in the other, records where their values differ and which value is relied on for what, and is dated each time a person checks it.

Where the standards stop

ISO/TC 251’s guidance on the 2024 edition describes Clause 7.6, data and information, as clarifying how data and information differ and making its requirements clearer and more concise. It says nothing about a crosswalk. The row-level crosswalk is this publication’s proposal for making the correspondence between two registers demonstrable; its form and fields are not attributed to ISO.

Why it earns its cost

A shared key makes a lookup possible; it does not say how the two records relate. The financial register is coarser by design: one financial line can hold several maintainable items, and an item can sit in its own component or in the remainder. Their values can differ for good reasons, such as a depreciation life and a renewal interval set for different purposes. A crosswalk that states the correspondence, the accounting treatment and the reason for each difference lets a reviewer tell an intended difference from an error. A key alone does not.

What would show it wrong

Evidence that reviewers can trace items reliably between the two registers, and tell intended differences from errors, using the shared key alone, without a row-level crosswalk or recorded reasons. A test on real registers that times the lookups and counts misread differences, with and without the crosswalk, would show it.

Sits beside: ISO/TC 251, Updated Guidance on Asset Management System Requirements, ISO 55001:2024. Sources

Cite as: evidence-the-traceability-not-only-the-fields, projects-advisors.com/proposals#evidence-the-traceability-not-only-the-fields

A warranty record carries what makes cover fail, not only that cover exists

First argued 3 September 2026

A warranty register records, for each item, the event its warranty names as the start of cover and the date that event happened, the conditions the cover depends on, the route and notice period for claiming, where the item is, and, when a claim is refused, the ground the warrantor gave, because those decide whether the warranty can actually be used.

Where the standards stop

The Saudi national turnover procedure (national project manuals, Volume 16) requires a warranty log to be available at turnover and the entity to designate a warranty administrator. The contractor’s obligation to remedy defects after taking over, and the period it runs for, are set by the contract: FIDIC 2017 Clause 11 and the Saudi procurement regulations. A supplier’s or manufacturer’s warranty is a separate document with its own period and its own start. Neither the procedure nor those contract provisions requires the conditions, the claim route or the ground of a refusal to be recorded.

Why it earns its cost

A supplier’s warranty starts on whatever event its own terms name, which need not be taking over, so a register that dates every warranty from handover can show cover that has already lapsed, or cover that has not yet begun. A warranty conditional on quarterly service by an approved agent is also a maintenance requirement, and it sits in a document filed at handover while the maintenance contract that has to satisfy it is often written by somebody who never saw it. A refusal states which condition the warrantor relies on; recording that ground on the row is what lets the maintenance arrangements be checked against it.

What would show it wrong

Claim data, from warrantors or from owners, showing that refusals for an unmet condition are rare, which would mean the conditions are being met without anybody tracking them.

Sits beside: Tyson, FIDIC 2017: Clause 11, Defects After Taking Over, Howard Kennedy; Saudi Government Tenders and Procurement Law and its Executive Regulations, consolidated Arabic text, third edition. Sources

Cite as: record-what-makes-cover-fail, projects-advisors.com/proposals#record-what-makes-cover-fail

Capture what perishes at handover, on the day, or not at all

First argued 3 September 2026

The turnover package, and the register built from it, record the things that otherwise exist only in people: the values the systems were left running at, who to ring for each system, where what was built departs from what was approved and who decided, and where each item’s manuals and records went and who is now responsible for it.

Where the standards stop

ISO/TC 251’s guidance describes Clause 7.7, new in the 2024 edition, as requiring the organisation to determine the knowledge needed to operate its asset management system, to have processes for dealing with it and to keep it current. FIDIC requires as-built records, which show what is there and not that it was meant to be different, or why. The list above is this publication’s proposal for what to capture; neither source sets it out.

Why it earns its cost

All four cost little on the day and are hard to recover once the site team has dispersed and the subcontractors have moved on. Without the as-left setpoints the first fault investigation begins by re-commissioning; without the approved departures, every future modification reopens a decision somebody already made carefully.

What would show it wrong

Evidence that operators recover this material reliably after handover at acceptable cost, which would make capturing it at handover an optimisation rather than a necessity.

Sits beside: ISO/TC 251, Updated Guidance on Asset Management System Requirements, ISO 55001:2024; Tyson, FIDIC 2017: Clause 11, Defects After Taking Over, Howard Kennedy. Sources

Cite as: capture-what-perishes-at-handover, projects-advisors.com/proposals#capture-what-perishes-at-handover

Where the published life is a range, record which end you took and why

First argued 3 September 2026

Where an authority publishes a useful life as a range rather than a figure, the point chosen within it is a judgement and is recorded as one, with its reasoning and any evidence.

Where the standards stop

The Saudi Ministry of Finance publishes a useful life for 770 asset types and not one is a single figure; every one is a minimum and a maximum, printed beside a default life. Page 76 requires the entity to select useful lives within the published ranges, taking into account information from the manufacturer or its technical department, the nature of use and its ability to obtain the asset’s benefits, and to submit them for approval; it does not specify that the rationale for each selected point must be documented. The manual does not say what the default is for. IPSAS 45 paragraph 48 then requires the life to be reviewed annually.

Why it earns its cost

Across the 770 the maximum is a median 1.67 times the minimum, so the annual depreciation charge at the short end is a median 67% higher than at the long end and both are inside what the manual permits. A number that can move the charge by two thirds is a judgement large enough that its reasons belong beside it, and page 76 does not say that they must be.

What would show it wrong

A ministry instruction that fixes the point within each range, or requires the printed default to be used, which would make the choice a rule rather than a judgement. The second edition prints a default beside each range and still leaves the choice to the entity (page 76).

Sits beside: Saudi Ministry of Finance, comprehensive manual for the census and valuation of assets of government entities, Arabic; IPSAS 45, property, plant and equipment. Sources

Cite as: record-which-end-of-the-range-you-chose, projects-advisors.com/proposals#record-which-end-of-the-range-you-chose

Equipment still running does not tell you how long equipment lasts

First argued 3 September 2026

A service life is taken from equipment that has been replaced. Equipment still in service reports its age, which is a lower bound, and the two are never averaged together.

Where the standards stop

ASHRAE reports the two separately and correctly, in columns headed "currently in service" and "replaced". A life table that gives a figure, or a range, without saying which population it came from cannot be checked against that distinction.

Why it earns its cost

Every unit in the in-service column is still running, so its age is a lower bound on its life, not an estimate of it. Averaging the two columns mixes lower bounds with completed lives, so wherever the replaced units lived longer than the survivors have so far, the pooled figure falls below the observed life. In the rows of ASHRAE’s cooling set with five or more replacements, pooling understates the observed life by between 22% and 42%. The error has a mechanism, and anyone can check it against the published columns.

What would show it wrong

A dataset where the in-service and replaced populations are drawn on the same basis, in which case pooling is legitimate. ASHRAE’s are not, and any table that does not say which its figures came from cannot be checked either way.

Sits beside: ASHRAE Service Life and Maintenance Cost Database. Sources

Cite as: do-not-read-survivors-as-lifespans, projects-advisors.com/proposals#do-not-read-survivors-as-lifespans

A renewal cliff is a property of the assumption, not of the estate

First argued 3 September 2026

A renewal profile built by adding one life to one installation date shows a cliff that the estate does not have, so the dispersion of the lives is stated and applied, or the profile is not used for a funding decision.

Where the standards stop

The standards cited here do not set out how to build a renewal profile. ISO 55001 requires the activities that deliver the objectives to be planned and controlled, and says nothing about the arithmetic. ASHRAE publishes the dispersion, in the same table as the mean and in the same columns, and nothing requires anybody to use it.

Why it earns its cost

A hundred identical items on a point life produce a hundred renewals in one year and none in any other. At the standard deviation ASHRAE reports for the centrifugal chiller, 7.6 on a mean of 25.2, the same hundred items give about five in the peak year on a Weibull fitted to those two numbers, five on a normal and four treating every age in range as equally likely: the point calculation overstates the peak by 19.1, 19.7 or 26.3 times. The three agree because at this dispersion the shape hardly matters, which is the whole answer to the objection that the finding depends on a chosen curve. A bid for renewal funding built on the cliff asks for a spike nobody will fund and gets a number nobody believes.

What would show it wrong

Published replacement data for a real estate showing renewals genuinely clustering in the years a point calculation predicts. Fleet or utility asset data would show it either way, and the direction of the argument is checkable against any dataset that reports a spread at all.

Sits beside: ASHRAE Service Life and Maintenance Cost Database; ISO 55001:2024, asset management system requirements. Sources

Cite as: a-cliff-is-an-artefact-of-the-assumption, projects-advisors.com/proposals#a-cliff-is-an-artefact-of-the-assumption

Every claim on this site is marked by what it rests on. These rest on this publication’s own reasoning, set out above, rather than on any external requirement. The rest are on the standing page, sorted from the claims that rest on somebody else’s authority to the ones that rest on ours.