The asset manager presents the monthly maintenance report to the operations committee. MTBF of 640 hours, MTTR of 9 hours, availability of 98.3 percent. The committee looks at the indicators and finds no cause for alarm. On the next slide, the number of incidents logged in the quarter is up twenty-three percent from the previous one. Two committee members disagree: for one, the indicators show a healthy operation and the incident bump is statistical noise; for the other, the rise anticipates a deterioration that has not yet fed into the weighted MTBF. The asset manager has no way to settle the disagreement with the numbers on the table, because the aggregate metric collapses the spatial dimension, and the spatial dimension is exactly where the signal sits.
Two failures four days apart and one hundred meters apart are not the same as two failures four days apart and forty kilometers apart. To the MTBF formula they are. A basic spatial analysis distinguishes them in seconds and reveals what the average cannot: whether there is a cluster, whether there is a line, whether there is a contractual boundary separating one side from the other. The quality of the committee report improves as much as the ability to phrase the right spatial question improves.
What a spatial analysis of failure looks at
An elementary spatial analysis over a corrective history is not an academic exercise. It is five concrete queries that an asset manager can ask their platform to run within the time window of a pre-committee meeting. Hotspot detection identifies the points where failure density clearly exceeds the expected mean — a Getis-Ord query or a simple grid-cell aggregation usually suffices. Density clustering — DBSCAN applied over coordinates and timestamp — groups failures that occur close in space and time and exposes what would read as a series of independent events in a tabular view. Spatial autocorrelation — Moran's I or Local Moran — quantifies how much a point resembles its neighbors: a high positive value indicates clustering, a negative value indicates unusual dispersion. Buffer aggregation against assets of known criticality crosses the spatial pattern with the business model: three failures that looked isolated turn out to sit inside a thirty-meter radius of the same transformer. And comparing time bands shows whether the spatial pattern is stable or drifting; a cluster that appears in the northeast during the first half and migrates to the southeast during the second tells a different operational story from a cluster that stays anchored in place. These five queries share an important property: they do not require introducing new metrics, only looking at existing ones with a dimension that until now was absent from the report.
What the data infrastructure requires
For none of these five queries to force the asset manager out of their working environment, three conditions have to hold on the base. The first is geolocation of every intervention at signature time in the field, not the theoretical geolocation of the linked asset; the difference between the two shows up in the analysis and adds nuance to the patterns. The second is a data structure that supports querying by criticality, contractor, intervention type and time window without exporting to an intermediate spreadsheet. The third is a spatial engine — PostGIS is today the de facto standard — able to resolve the analysis in seconds over portfolios with tens of thousands of historical events.
At Maptainer, these three conditions close by construction: the signed capture of an intervention includes the GPS position at signature time, the data structure supports cross-filtered queries from the asset manager's interface, and the PostGIS backend resolves the spatial analysis with no intermediary tool. The committee stops waiting two days for a post-hoc report and receives the analysis inside the meeting itself. The delta is not marginal: the difference between showing a heat map on screen while the committee is still in the room and sending it two working days later is the difference between influencing the next decision and documenting a decision already taken.
Three patterns that recur
After three or four spatial audits, an asset manager starts to recognize three patterns that show up more often than the rest. The first is the cluster around a single asset: when several apparently independent correctives group at a short distance from a common element, the root cause usually sits in that element, not in the correctives it has produced. The second is the linear pattern along a network segment: failures that align on an electrical line, an irrigation canal or an underground trace point to an upstream infrastructure issue that surfaces as downstream correctives. The third is the contractual boundary pattern: failures cluster precisely at the boundary between two responsible contractors, and the operational conclusion is that there is a coordination process between them that has not been closed. The three patterns are actionable inside the meeting, without a follow-up study. Each one redirects the operational conversation toward a concrete decision: replace an upstream asset, reinforce a network segment, close the coordination process between contractors at the identified boundary. All three decisions are measurable, and their impact shows up in the next committee's spatial analysis.
The shift in the committee's conversation
When the asset manager walks into the committee with a spatial analysis instead of only the aggregate mean, the conversation changes subject. The meeting stops debating whether MTBF moved or not; it discusses which root cause explains the west-sector cluster and which action on transformer T-12 returns two tenths of availability in the next quarter. The asset manager moves from justifying variance to proposing concrete interventions. That reading is the one the finance function accepts as the basis for CAPEX allocation and the one the operations committee expects when the report is properly prepared. The difference is not in having more data. It is in asking the same data a spatial question that until now was not being asked. And once the committee has heard that kind of answer twice in a row, the meeting no longer accepts the aggregate-only report as a valid deliverable. The bar rises quietly, and the operations function rises with it.