We focus on capacity planning and admission control of people accessing a system consisting of paths, where the flow varies along spatial, temporal, or event-based dimensions. People arrive during a time window, are admitted, travel through specified paths, and leave the system within a provided admission time guarantee. The decisions include choosing capacity for disruption plans and reactive capacity, adaptive path adjustment through redirection, and admission control in response to realized demand scenarios. We first develop a continuous-time model to choose optimal capacity for disruption plans corresponding to an expected admission time guarantee. The next model determines the optimal upfront versus reactive capacity decisions given demand scenarios. For each disruption plan, a mathematical programming model chooses optimal capacity by path segment, redirection, and admission control over time to manage realized demand under an admission time guarantee. We quantify the value of flexible capacity deployment and: (a) provide the optimal capacity by disruption plan to ensure an admission time guarantee, (b) provide the optimal long term capacity and reactive capacity, and (c) provide the optimal adaptive routing and admission control decisions to minimize capacity costs to ensure an admission time guarantee. We apply the model to data from the Kumbh Mela at Ujjain, a bathing festival in India that attracts close to 85 million people over a 25-day period, to provide insights for managers. The models can assist managers of food festivals, music events, religious walks, and other events involving managing people flow across paths during specific time windows.