Critical service boundaries and dependencies · Lesson 4

Prioritize critical functions using consequences and uncertainty

10 min read

Lesson objectives

  • Prioritize critical functions using consequences and uncertainty for a documented critical-service case.
Reading tools

Priority criterion Calling a service critical does not mean that every user receives an equal share when capacity is scarce. Priority reflects the consequence of service reduction for safety and essential functions, how quickly that consequence appears, the exposed users and their vulnerability, and whether a usable substitute exists. These dimensions assess the consequence of interruption, not the worth of people or organizations. An asset planner cannot independently set a clinical-use limit or invent a legal requirement. Obtain limits from competent service owners and verified operator obligations. If no approved limit exists, a curtailment decision must remain temporary and conditional, and the responsible authority must be contacted. This division of responsibility is part of resilience: a fast but unauthorized service decision may worsen harm despite accurate pump arithmetic.

Make a six-column matrix for each function: required minimum flow, time to impact, nature of impact, evidence for the limit, available substitute and activation time, and who can approve reduction. Do not assign arbitrary qualitative scores and present the resulting rank as objective. If weights are used, record who selected them and test sensitivity. Start with conditions that cannot be traded away, such as verified water quality or operating safety, before comparing options that satisfy them. A high-volume use may be deferrable while a lower-volume use is time sensitive. An overall percentage of demand served can hide that one essential function received less than its approved minimum.

Capacity comes from the dependency map, not the sum of named assets. If pumps produce 25 cubic metres per hour but a filter passes 18, allocate 18. Before dividing it, verify that flow can be routed to each user and that pressure and quality remain acceptable under the new pattern. Calculate the gap as the sum of proposed minimums minus usable delivered capacity. A positive gap means the plan cannot meet all of them. Seek authorized curtailment, a verified alternate source, or an explicitly approved emergency level. Handle uncertain capacity as a range or scenarios. Does the same allocation work at both 18 and 16? If the conclusion changes, identify a measurement or action required before committing to it.

Separate required minimum, forecast demand, and available capacity. A service may normally request 11 but have an approved temporary minimum of 8 for a specific disruption. The first number belongs to normal operation and the second to a bounded response plan; they cannot be substituted without recording the approving authority and duration. If a user has its own tank, verify usable stock, draw rate, quality, and routing before reducing its network allocation. The presence of nominal storage does not remove all demand from the grid. Short peaks also matter: a four-hour average flow may equal an allocated share while a critical function fails during a ten-minute peak. Record the measurement window, delivery observation, and pressure consequence at remote outlets for each allocation. When function-level metering is unavailable, name the estimate and the field test needed rather than marking an inferred output as confirmed.

Uncertainty concerns consequences as well as capacity. One function may tolerate a thirty-minute delay because it has validated local storage; another may deteriorate in ten minutes. Ask its owner for the time-to-impact and supporting evidence, then compare that period with actual switching time. A nominal alternate that takes four hours to start cannot protect a function whose impact starts after one hour. Specify the review point, cancellation measure, and notification owner for a temporary allocation. Priority then becomes a decision that can be revisited when facts change instead of a fixed ranking copied into every incident.

Test the decision with a concrete counterexample. In the example, capacity of 17 could supply 11 and 5 in aggregate. If a routing valve could deliver only 3 to the shelter, total capacity would be unchanged but the shelter's minimum of 5 would fail; the route would need repair or a verified local alternative. This explains why arithmetic on total shares alone is insufficient. If water quality fails, it is not legitimate to distribute 17 cubic metres per hour of unusable water and call the allocation equitable. Quality is a precondition. End curtailment through an authorized return-to-normal decision after measuring delivery, not merely after hearing that a pump motor has restarted.

Extended decision under changing facts At the decision meeting, show three evidence layers. First, service facts: who measured 17, at which point, and whether it includes only usable water. Second, user limits: who set 11 and 5, and whether either owner has an approved reduction or tested local stock. Third, implementability: can the valve route 16 as allocated, and how long does irrigation shutdown take? Whole-channel capacity does not automatically establish delivery at two remote points; loss and pressure may matter. A margin of one offers weak assurance if measurement uncertainty is two. Make a near-term field reading a condition of the allocation.

An estimated capacity of 17 with an operating range of ±2 gives 15–19. The two minimums sum to 16. At 19 the margin is three, at 17 one, and at 15 there is a shortage of one. Without a better measurement, do not call a one-unit estimated margin robust at every plausible value. Test flow at both outlets and arrange at least one unit of verified alternate supply at the low end, or seek an approved temporary level. Taking one from each minimum without authority is not a valid way to balance the arithmetic. The sensitivity identifies the evidence needed before a paper allocation becomes a reliable service choice.

Curtailment has an execution time. If irrigation starts automatically each hour, changing its schedule may not stop an active cycle for ten minutes. When storage lasts only five minutes, do not assume six units per hour of demand vanish as soon as the decision is typed. Record the control action and observed decline, then recalculate availability to the two essential functions during transition. A small buffer or an authorized manual stop may be needed. Written reduction matters only when delivered-service measurement shows its effect.

When a verified supplier adds three units per hour after two hours, recalculate from its actual start, not retroactively. If essential demand rises from 11 to 13 while the other minimum stays five, total need is 18: capacity 19 suffices while 17 does not. Reopen the decision with the changed fact, result, and next review time. The priority table responds to evidence instead of becoming a ranking copied unchanged across incidents.

Lesson summary

Service priority reflects the consequences and timing of lost function and available alternatives. Build an allocation respecting service owners' minima and usable delivery capacity. Review measurement margins, routing and quality, then identify reduction authority, escalation timing and allocation updates when facts change.