When the bottleneck thesis is incomplete

The bottleneck thesis is useful when the objective is increased accepted serial flow under a compatible annual basis. It becomes incomplete when the decision concerns another duty or when the serial representation excludes a decisive mechanism. This chapter makes those challenges explicit. A strong thesis should identify the conditions under which it ceases to select the action, rather than protect itself by calling every adverse outcome another bottleneck after the fact.

Demand is the clearest structural challenge. If eligible demand is below every stage capacity, additional capability cannot increase the common flow in the current model. The changed-demand study demonstrates that reversal. The correct next question may concern customers, service quality, access or cost rather than another asset. Calling demand a bottleneck can be descriptively convenient, but it should not obscure that the relevant intervention is commercial and outside the app's capacity-only expansion tests.

Quality can change the objective from gross movement to accepted output. A higher rate may increase damage, contamination or rework. A stage that governs accepted quality can matter even when its gross annual capacity is high. The service map should therefore define eligibility and acceptance before the capacity comparison. The current app assumes uniform quantity and no loss, so it cannot calculate that quality tradeoff. A valid challenge requires a method and evidence representing the actual transformation or rejection mechanism.

Strategic reserve can justify capacity that does not increase normal annual flow. A sponsor may value continuity during an interruption, optionality for a future service or compliance with an access duty. Those reasons can be legitimate, but they need a stated objective and feasibility evidence. The annual minimum result does not evaluate them. The committee should avoid forcing every benefit into throughput and should avoid treating unquantified strategic language as automatic justification for any expansion.

A serial chain requires every tonne to pass through every represented stage. A parallel network allows route choices and may contain shared links. The constraint can then be a set of links or an allocation rather than the smallest individual stage. Treating alternatives as serial stages can make more options appear to reduce capacity, while summing all alternatives can overstate capacity when they share a decisive link. Both errors arise from a mismatch between topology and method.

When routes become parallel, specify origins, destinations, capacities and shared dependencies together with access conditions and assignment rules. Physical route options become service only through that allocation basis. The present app does not implement a network model, so renaming its serial stages cannot establish network optimisation. Select and implement a suitable method whose inputs, outputs and limits can be reproduced and reviewed, or retain the current calculation as a narrower serial screening result. The representation must change before its conclusions are widened to the new network decision.

Multi-product operations can also create allocation problems. Different cargo classes may consume handling time, space and access differently. A single tonne basis can conceal the tradeoff between them. A product mix change may alter effective capacity without any physical asset changing. The team should determine whether a homogeneous annual flow remains useful for the decision or whether explicit resource consumption and allocation are needed. The study offers no calculated mixed-product optimum.

Shared resources can make apparently serial stages interact. Receiving and reclaim may use the same handling equipment; loaded and empty movements may share infrastructure; documentary reviews may use a common service desk. Assigning separate independent capacities can overstate what can be achieved simultaneously. The model needs to represent the shared constraint or narrow its claim to a screening bound. The challenge is methodological, not a request to insert another arbitrary availability reduction until the result looks plausible.

The annual minimum can be unattainable when service opportunities do not align and buffers are finite. A stage can have ample annual capability but be unavailable during the only compatible downstream window. A yard can fill and block the upstream chain despite adequate annual receipt and loading rates. The remedy may involve chronology, batching or buffer deployment rather than another annual-capacity increase. A suitable time-based study should represent those mechanisms when they can reverse the decision.

Establish route authority, configuration safety and cargo acceptance before ranking flow alternatives. These noncompensable conditions define the feasible set: greater throughput cannot offset an unauthorised route, unsafe configuration or unacceptable cargo. The relevant actors must confirm eligibility and authority. The current model assumes an eligible fictional service and supplies no legal or technical admissibility decision. If a condition is unresolved, keep that option outside a commitment ranking until the required review resolves it rather than convert the condition into a compensable throughput penalty.

Cost and implementation can make the physical optimum unattractive. A large coordinated expansion may create more annual capacity but require expensive work, disruptive transition or unsupported customer demand. A smaller reversible change may be preferable under the actual economic basis. The canonical cases deliberately isolate physical response and operating denominator effects. They do not establish the best capital programme. The paper's recommendation remains to use the system constraint as part of the decision, not as its sole criterion.

Evidence can also be the binding decision issue. If the team does not know whether observed rail shortfall reflects capability, demand or destination blocking, a precise expansion calculation cannot resolve the uncertainty. A focused measurement may be more valuable than another scenario. The analysis should identify the observation that would distinguish the competing mechanisms and the consequence for the choice. This is a limit on what the current evidence permits, not a claim that every decision must wait for perfect information.

Suppose the fictional loading-only intervention is proposed to increase annual corridor quantity. The frozen result does not support that claim because rail remains binding. Now change the stated objective to completing a ready vessel transfer within a shorter permitted operating window. The same physical asset proposal may become relevant, but the annual app cannot establish its value. The correct response is a new service question and suitable evidence, not a defence that the original throughput model secretly represented the new objective.

This distinction permits serious disagreement without forcing a false binary. The annual result can be correct and the proposal can still deserve further study. The proposal can also be attractive under a local service objective while failing a wider cost or access condition. The committee should identify which proposition it is deciding and which evidence supports it. Changing the objective is legitimate when declared; changing it after a failed test while preserving the original success claim is not a transparent evaluation.

The reverse case also matters. A coordinated expansion can raise the fictional annual bound while worsening a service objective omitted from the model. It could create a difficult transition, increase exposed inventory or require an unacceptable access arrangement. No such consequence is asserted as an actual finding here. The point is that a positive physical output does not automatically dominate every feasible alternative. The adversarial review should test the omitted consequence most capable of reversing the choice.

The resulting review record should state whether the challenge changes the objective, feasible set, input basis or method. That classification tells the team what work is required. A corrected objective needs a revised service and evaluation plan. A changed feasible set needs relevant authority evidence. A changed input needs reproduction. A changed mechanism needs another implemented method or a narrower claim. This is more useful than adding a general caveat and leaving the recommendation unchanged.

A reviewer should challenge the thesis through an admissible counterexample. State the service, changed condition and expected effect, then explain why the annual bottleneck recommendation would be inferior or incomplete. A vague objection that reality is complicated is less useful than a specific case of demand limitation, tied constraints, shared topology, quality loss or incompatible windows. The review should distinguish a failure of the formula from a failure of its applicability to the decision.

The frozen slate supports several of these structural challenges. Demand limitation removes the capacity-expansion rationale for increased flow. Equal limits make isolated relief inadequate. An unavailable essential stage stops serial service and makes unit cost undefined. The coordinated case shows that the next constraint can govern the benefit. The cases do not cover network routing, stochastic queues or quality. Their absence is disclosed rather than filled with an unimplemented calculation.

The response should be proportionate. If the challenge identifies a basis error, correct and reproduce the model. If it identifies an omitted mechanism, select a method that represents it or narrow the recommendation. If it identifies an ineligible option, remove it from the feasible comparison. If it identifies an economic omission, obtain the missing appraisal evidence. A defence that simply adds caveats without changing the decision logic is insufficient when the challenge can reverse the choice.

A validity map should preserve the useful calculation after rejecting its original recommendation. If a hypothetical loader proposal is justified by berth-window recovery rather than annual quantity, retain the unchanged serial-flow result and move the window question to a dated operating study. If a parallel route is introduced, retain the serial calculation only for the route it represents and model the shared interfaces separately. If cargo rejection matters, retain gross movement as a handling measure and reconstruct accepted service. Each branch changes the object of inference rather than adding another disclaimer. The committee can then see exactly which conclusion survives, which is withdrawn and which new evidence would support the revised decision.