The work became more credible when its limits became visible.

Our 2022 maintenance shutdown had accumulated eighteen proposed work packages. The calendar gave them a common starting point, but their preparation was uneven. Eleven had the relevant engineering scope, materials and specialist arrangements ready for the proposed sequence. Seven still depended on unresolved conditions. The original plan treated those dependencies as matters that would somehow be settled during the shutdown. That assumption put the available time at risk before the plant had even stopped.

I asked the engineering and site teams to distinguish work that could be executed from work that was still being prepared. After their assessment, we retained eleven packages and deferred seven into a separately owned programme. The revised plan included CNY 8.6 million in additional preparation, specialist support and rescheduling expense. Some colleagues preferred to keep all eighteen in the headline scope, arguing that a larger shutdown programme would demonstrate ambition and avoid another difficult scheduling discussion. I supported the smaller executable scope because a list of intended work was not a credible account of what the shutdown could deliver.

This was not an operating decision to waive a necessary maintenance requirement. Engineering and the relevant control owners assessed the consequences of each deferral and the conditions under which the plant could operate. Work that was essential to an approved operating condition could not be removed simply to protect a date. The seven deferred packages retained accountable owners, unresolved questions and the decisions required before a later schedule could be confirmed. Independent quality and safety responsibilities remained with their authorised functions.

The shutdown completed the eleven agreed packages. The other seven were reported as deferred, not included in a claim that the entire original programme had been delivered. That distinction gave the next planning discussion a more honest starting point. My mistake had been allowing a common date to make differently prepared work look equally ready. I would now require the preparation discussion before a proposed package enters the shutdown commitment, including the effect its unresolved conditions could have on the other work.

In 2010, while leading plant services, I relied too readily on an inventory report that described ninety-six selected mechanical spare items as available. Supervisors questioned the report after several requests required additional identification work. A physical and engineering review found sixty-two items matched the relevant current equipment and support records. Twenty-four related to equipment no longer supported by that stock arrangement. Ten could not yet be matched with sufficient confidence. The count had been accurate as a count of records, but misleading as an account of usable support.

We spent CNY 2.4 million on replacement procurement, identification work and the appropriate handling of obsolete stock. I supported that expense rather than continue presenting all ninety-six items as an operating reserve. Finance colleagues reasonably wanted to understand why another purchase was necessary when the inventory showed material already held. The answer required an item-level explanation of its intended use, condition and supporting evidence, not a general appeal for more resilience spending. Engineering colleagues retained responsibility for determining whether a part was suitable for a particular use.

The revised account kept three categories visible: sixty-two reconciled items, twenty-four obsolete items removed from the availability claim, and ten unresolved items excluded until their identity and suitability could be established. Replacement purchases were tracked separately rather than silently added to the original population. The exercise concerned mechanical support stock, not drug batches or product release. No inventory correction changed the controls governing work on equipment or the approvals required after that work.

The immediate result was a smaller but defensible statement of what the selected stores could support. I did not claim that every future breakdown would be shorter or that CNY 2.4 million had produced a measured saving. The more useful lesson was about the question behind the report. I had asked how many spare items were held when I needed to ask which maintenance situations those items could actually support. Afterward, the review linked stock to current equipment responsibility and preserved unresolved identification questions instead of hiding them inside an apparently complete total.

A 2016 site review began with a proposal to expand utility capacity. Production teams described recurring conflicts during periods when several activities drew on the same support system. The investment case assumed that the observed peaks represented unavoidable demand. I asked the site and engineering teams to examine the sequence behind those peaks before treating additional installed capacity as the only available response. The question was operational: which demands genuinely had to coincide, and which coincided because the planning process had never considered them together?

The review identified changes to the sequence of selected activities that could be considered without altering their approved technical conditions. Engineering, production and quality colleagues examined the relevant boundaries. We adopted the authorised sequence changes and funded CNY 5.2 million in scheduling work, limited supporting modifications and additional operating expense. This was not a unilateral COO instruction to change a manufacturing process. Nor did it cancel the need to evaluate a larger investment if the revised operating account still showed a durable capacity gap.

In two defined observation periods using the same internal planning measure, hours recorded above the agreed scheduling allowance fell from nineteen to seven per week. We stated both the population and its limitation. The measure described planning pressure on the selected utility arrangement; it was not a claim about product quality, a universal engineering threshold or proof that the sequence change alone explained every difference. The seven remaining hours stayed in the capital discussion. Commercial colleagues also needed to understand the expense and timing consequences of the new sequence, rather than receive a headline that implied unused flexibility had been free.

The decision gave management a better basis for the next investment choice. It separated a sequencing problem that could be addressed from a remaining requirement that needed further examination. I would now request that distinction at the start of a capacity proposal. My earlier approach had accepted a peak-demand picture without asking how the operating calendar had created it. Buying more capacity can be the right decision, but it should follow a clear account of the work the existing system is being asked to support.

The opening of an expanded operating area in 2023 created a familiar pressure to describe the project as finished. Construction and equipment milestones had their own records, but responsibility for maintaining the new area was less complete. We reviewed thirty-six maintenance and support responsibilities that needed an enduring owner. Twenty-seven had an agreed receiving team, records and support arrangement. Nine still relied on project personnel or temporary external help. A single completion label would have concealed that difference from the people expected to operate the area after the project team left.

I supported a CNY 4.7 million temporary support arrangement while the nine responsibilities were resolved. The expenditure was unpopular because it arrived after a substantial capital programme. Some project colleagues believed the receiving organisation should absorb the work immediately. The receiving managers argued that accepting a responsibility without the required knowledge, records or support would merely transfer the unresolved problem. I asked both groups to specify what remained missing and what evidence would allow the temporary arrangement to end.

The result was a named transition account for each of the nine responsibilities, with a receiving owner involved before the handover was represented as complete. This account did not replace technical acceptance or any independent quality decision. It concerned the practical work of sustaining the area after those decisions, including access to support and responsibility for follow-up. The twenty-seven assigned responsibilities were reported separately from the nine transitional ones. We did not treat temporary coverage as evidence that permanent capability already existed.

At the reporting date, six of the nine transitional responsibilities had moved to their agreed enduring arrangement; three remained under temporary support. The CNY 4.7 million was the approved transition allocation, not a statement that all of it had been spent or that the three open items had disappeared. I would begin this ownership discussion earlier in the project. The main correction was to stop allowing an opening milestone to answer a different question: who will keep the area usable when the people who built it are no longer present?

In 2004 I introduced a maintenance scorecard intended to give managers a consistent account of outstanding work. It combined requests from different sources into a single ageing list and made the oldest items prominent. The report looked disciplined, but supervisors found it difficult to decide what to do next. A minor request waiting for access could dominate the discussion, while a newer issue with a different operating consequence was buried in the same total. I had made those records easier to aggregate without making the decision easier to exercise.

The correction began with the supervisors who used the report. We separated the reason work was waiting, the responsible decision-maker and the consequence of further delay. Technical priorities remained with the authorised maintenance and operating owners; the scorecard did not invent a universal priority rule. We ran a fourteen-week trial in two production areas and spent CNY 680,000 on planning support, record correction and supervisor workshops. I accepted that cost rather than defend the original design because it had been my proposal.

During the trial, thirty-one requests were identified as waiting for an access or ownership decision that had not been explicitly assigned. Twenty-three obtained the missing decision within the trial period. Eight remained open and continued to be shown with their actual dependency. The outcome was not thirty-one completed maintenance jobs: obtaining a decision was only one stage in the work. We preserved that distinction, and we did not infer a company-wide reliability improvement from a small trial in two areas.

The revised discussion gave supervisors a more useful way to explain why work had not moved and what management could do about it. It also made my own responsibility clearer. A central report can create delay when its categories encourage people to debate the wrong question. I would test a new operating measure with its intended users before extending it across a site. Consistency matters, but a consistent account that cannot support an appropriate next decision is not yet a good management instrument.