AAE5301 Service Design and Fleet Management for Low-altitude Economy
The Complete Exam Bible · Semester 1, 2026/27AAE5301 Overview
- The Hong Kong Polytechnic University
- Semester 1, 2026/27
- Postgraduate Level 5
- Mathematical service design
What the subject asks you to design
AAE5301 treats low-altitude transport as a service and operations system. The captured teaching plan asks students to site facilities, design service networks, assign fleets, schedule flights and plan vehicle routes with mathematical models, then use decision support for UAV applications.
- Assessment Individual assignments, a group project and a final examination form the published grade.
- Pass condition Continuous assessment and the examination must each be passed; confirm the unpublished numeric thresholds on the subject Canvas page.
- Core method Translate service promises into variables, objectives, constraints and deployable decisions.
- Final details Use Canvas for the examination duration, sections and permitted materials when published.
How AAE5301 is assessed
| Component | Weight | Format |
|---|---|---|
| Individual assignments | 30% | Two assignments, each worth 15% in the current teaching plan |
| Group project | 20% | Service-network case study, analysis and group presentation |
| Final examination | 50% | Final examination; duration, sections and permitted materials are not stated in the captured files |
The current teaching plan publishes these three components and they total 100%. It states that students should pass both continuous assessment and the examination to pass the subject. The numeric pass threshold for either aggregate is not published in the captured materials. The first assignment and second assignment are each worth 15% within the 30% assignment total. Confirm the operative thresholds and final examination instructions on the subject Canvas page.
Assessment structure
Segment widths reproduce the published weights and total 100%.
Current AAE5301 dates
| Date | Item | Control |
|---|---|---|
| 5 October 2026 | First individual assignment | Due on or before this date in the current teaching plan and shown on Canvas. |
| 9 November 2026 | Second individual assignment | Due on or before this date in the current teaching plan. |
Dates are as published in the teaching plan and the Canvas assignment page for Semester 1, 2026/27. Confirm exact deadlines and submission settings in the live LMS.
What AAE5301 covers
Seven chapters move from optimisation and location through service networks, fleet decisions, routing and queueing to eVTOL and vertiport safety.
Optimisation Foundations for Low-altitude Services
objectives, decision variables, constraints, feasible sets, LP and mixed-integer model classificationFrom service question to mathematical decision
Optimisation starts by separating what an operator can choose from what the scenario supplies. Decision variables describe assignments, opened sites, vehicle counts or selected arcs.
- Decision variables and parameters
- Objectives with explicit units
- Constraints and feasible sets
- Binary, integer and continuous domains
- LP, IP, MIP and MILP classification
- Feasibility and sensitivity checks
Vertiport Location and Coverage Models
candidate screening, siting factors, coverage, p-median, p-center and hub flow decisionsLocation as a service promise
Vertiport siting affects access, route geometry, charging demand, network cost and safety. The course workflow begins with scenario and data, screens infeasible candidates, selects a model, and tests the result.
- Scenario and candidate screening
- Demand, safety, noise and access
- Set and maximal covering
- P-median and p-center objectives
- Opening and assignment variables
- Single and multiple hub flows
Service Networks and Tactical Decisions
service modes, strategic and tactical horizons, frequency, timetables and multimodal interfacesClassify the service before optimising it
Low-altitude services can cover vertiport-to-vertiport or first and last-mile legs and can run on schedules or in response to requests.
- Vertiport-to-vertiport service
- First and last-mile service
- Scheduled and on-demand modes
- Strategic, tactical and operational horizons
- Frequency and timetable decisions
- Time-space vehicle continuity
Fleet Sizing, Assignment and Repositioning
productive cycles, service targets, cost horizons, vehicle balance and static or dynamic redistributionCapacity, geography and time horizon
Fleet sizing converts demand, operating time, cycle time and a service target into an integer number of vehicles.
- Productive cycles per vehicle
- Service targets and integer rounding
- Strategic and tactical fleet costs
- Vehicle type assignment
- Static and dynamic repositioning
- Depot balance and conservation
Vehicle Routing for Drone Operations
TSP, CVRP, flow conservation, subtour elimination and pure or multimodal drone deliveryConnected service sequences
TSP creates one depot-returning tour, while VRP extends the structure to multiple vehicles, capacities, time windows, pickups, depots, transfers and batteries.
- TSP and depot-returning tours
- Arc variables and flow conservation
- Subtour elimination
- CVRP and cumulative load
- VRP variants
- Pure and truck-drone delivery modes
Queueing Models and Service Capacity
arrival and service rates, M/M/1, M/M/c, Erlang loss, blocking and linked service nodesRandom demand meets limited service
Queueing converts inter-arrival and service-time information into rates, utilisation, waiting and blocking.
- Poisson arrival rate
- Exponential service rate
- M/M/1 utilisation and waiting
- M/M/c parallel service
- M/M/c/c and Erlang-B
- Open and closed queueing networks
eVTOL Systems and Vertiport Safety
aircraft configurations, facility classes, TLOF, FATO, safety areas, OLS and obstacle-free volumeFrom mission to protected space
eVTOL and UAV definitions, propulsion configuration and energy source shape range, payload, turnaround and facility needs.
- eVTOL and UAV definitions
- Lift-plus-cruise and vectored thrust
- Aircraft D-value
- TLOF, FATO and safety area
- Landing point, pad and hub
- OLS and obstacle-free volume
The practical thread is consistent: define the service promise, choose decisions the operator can control, express cost or benefit through an objective, turn physical and service limits into constraints, and interpret the computed result in operational language.
How the seven chapters fit together
The guide begins with optimisation because every later model depends on variables, objectives, feasible sets and domains.
Location models decide where infrastructure opens and how demand is assigned. Service-network design decides whether movements are scheduled or on demand and whether aircraft cover a trunk or first and last mile. Fleet decisions translate demand and productive time into vehicle counts, assignments and repositioning. Routing creates depot-connected service sequences under capacity, time or energy limits.
Queueing adds random arrivals and service times, showing why spare average capacity can still create long waiting or blocking.
The final chapter works outward from eVTOL and UAV properties to TLOF, FATO, safety areas, obstacle limitation surfaces and obstacle-free volume.
What is hardest in practice
The mathematics is not difficult because the notation is long; it is difficult because one incorrect interpretation changes the model. Pooling vehicle time can hide an indivisible overload.
Degree constraints can admit a disconnected routing subtour. Treating an emergency request as waiting can select M/M/c when the system actually loses or redirects the call. Using a weighted location objective can leave the worst-served district far away.
Each chapter therefore ends by translating the result back into the physical service and checking the direction of change.
How assessment should shape revision
The current teaching plan allocates thirty per cent to individual assignments, twenty per cent to the group project and fifty per cent to the final examination. It also states that continuous assessment and the examination must both be passed.
The numeric thresholds for those two pass conditions are not published in the captured materials, so confirm them on the subject Canvas page. The captured materials also do not publish the final duration, sections, question types or permitted materials. Revision should focus on the model-building moves demonstrated in the lectures and first assignment while administrative details are confirmed on Canvas when released.
Worked example · free
Size a routine drone fleet from productive cycles
- 2Compute full cycles per drone from the daily operating time and complete cycle time.
- 2Convert the service target into an integer number of orders that must be completed.
- 4Round fleet size upward and verify that one fewer drone fails.
Key terms
- Decision Variable
- A quantity the model is allowed to choose, such as whether a site opens or how many drones move between depots.
- Feasible Set
- The collection of decisions satisfying every model constraint and variable-domain rule.
- Mixed-integer Program
- An optimisation model combining integer or binary decisions with continuous decisions.
- P-median Model
- A location model that selects a fixed number of facilities to minimise total demand-weighted assignment distance.
- P-center Model
- A location model that minimises the greatest distance between any demand point and its assigned facility.
- Hub Location
- A network decision that selects consolidation nodes and routes origin-destination flows through them.
- Fleet Repositioning
- The movement of idle vehicles from surplus locations to shortage locations before or during service.
- Flow Conservation
- A balance rule requiring movement entering and leaving a node to remain consistent with supply, demand or vehicle continuity.
- Subtour
- A closed customer cycle disconnected from the depot that can satisfy local routing degree rules.
- Blocking Probability
- The chance that an arrival finds all servers occupied in a no-wait loss system.
- TLOF
- The touchdown and lift-off area where the aircraft completes the immediate landing or departure manoeuvre.
- FATO
- The final approach and take-off area surrounding or incorporating the TLOF for the relevant manoeuvre.
- Obstacle Limitation Surface
- A protected approach, take-off or transitional surface that restricts obstacle penetration around a facility.
- Obstacle-free Volume
- A three-dimensional protected volume supporting vertical and multidirectional eVTOL movement above a vertiport.
AAE5301 FAQ
How is the subject assessed?
The current teaching plan allocates thirty per cent to individual assignments, twenty per cent to the group project and fifty per cent to the final examination. The two assignments are listed as fifteen per cent each. The three published components total one hundred per cent.
Do I need to pass the final examination separately?
Yes. The teaching plan states that students should pass both continuous assessment and the examination to pass the subject. Continuous assessment combines the assignments and group project. The numeric threshold applied to either aggregate is not published in the captured materials, so confirm both thresholds on the subject Canvas page.
What is the final examination format?
The captured materials confirm a final examination but do not state its duration, section structure, question types or permitted materials. The last teaching weeks are allocated to revision. Use the live Canvas instructions when they are released instead of assuming that the first assignment format will be repeated.
What should I write first in an optimisation answer?
Begin by defining decision variables with units and domains. Then write the objective, add each physical and service constraint, compute or compare solutions, and translate the result into an operational action. This order prevents parameters from being mistaken for choices and makes missing constraints easier to see.
How do I choose between p-median and p-center?
Use p-median when the promise is to minimise total demand-weighted access distance. Use p-center when the promise is to protect the worst-served demand point by minimising the maximum assignment distance. The two objectives encode different service priorities and can select different sites.
Why can a route satisfy flow rules and still be wrong?
Incoming and outgoing arc rules are local. They can admit a closed cycle among customers that never touches the depot. Subtour-elimination or other connectivity constraints are needed so every selected customer movement belongs to a valid depot-connected route.
When should I use a blocking model instead of a waiting model?
Use a no-wait loss model when a request finding all drones busy leaves the drone system or transfers immediately to another response. Use a waiting model when demand remains in a queue until a server becomes free. The operational handling of excess demand decides the model.
What does the group project reward?
The project brief rewards understanding of course and project material, a clear and comprehensive design logic, coverage of required content, time control, delivery quality and effective use of evidence such as data, models or visuals. Use new analysis and keep every network recommendation connected to the service scenario.
Questions students ask in each AAE5301 chapter
- Why should variables come before the objective?
- A symbol is meaningful only after it states what can be chosen, its unit and its domain. Defining variables first prevents fixed data from being treated as controllable and reveals whether a binary, integer or continuous decision is required. Chapter 01 · Optimisation Foundations for Low-altitude Services
- When is p-median the right model?
- Use p-median when a fixed number of facilities is given and the service goal is to minimise total demand-weighted access distance. It favours average performance and does not automatically protect the worst-served demand point. Chapter 02 · Vertiport Location and Coverage Models
- What is the difference between frequency and timetable?
- Frequency states how many services operate in a period. A timetable assigns departure and arrival times to specific movements. Frequency can meet aggregate volume while a poor timetable still creates missed connections or pad conflicts. Chapter 03 · Service Networks and Tactical Decisions
- Why do productive cycles round down?
- A partial cycle cannot complete an order inside the operating period, so productive capacity uses the lower integer. Required served orders and fleet count round upward because the service promise and vehicle count cannot be fractional. Chapter 04 · Fleet Sizing, Assignment and Repositioning
- Why are degree constraints not enough?
- One arrival and one departure at every customer can still create a closed cycle disconnected from the depot. Connectivity or subtour-elimination constraints are needed so selected arcs form valid service routes. Chapter 05 · Vehicle Routing for Drone Operations
- Why can waiting be long below full utilisation?
- Random arrivals cluster and service times vary. Average capacity can exceed average demand while temporary bursts create a line, so stability below one does not guarantee an acceptable waiting time. Chapter 06 · Queueing Models and Service Capacity
- How is an eVTOL different from a UAV?
- eVTOL describes electric vertical flight and can include passenger aircraft. UAV describes an aircraft with no onboard pilot, crew or passengers. The categories can overlap, but service purpose, payload and certification differ. Chapter 07 · eVTOL Systems and Vertiport Safety
How to study for the exam
Build one small service system across the whole guide. Start with four demand points and two candidate bases. Formulate a location model, assign demand, size a fleet from productive cycles, reposition vehicles, draw a depot-connected route, and choose a waiting or blocking model for random requests. Reuse the system only as a learning scaffold, changing the numbers and assumptions in each pass.
After every calculation, write the result in service language and name the constraint that drove it. For assessment revision, practise reconstruction from a blank page: variables and domains first, objective second, constraint families third, calculation fourth, operational conclusion last. Keep a short error ledger for units, rounding direction, pooled capacity, disconnected subtours and wrong queue capacity.
Those errors are more expensive than slow arithmetic because they change the model itself. Finish by checking Canvas for final examination administration, which is not specified in the captured lecture set.
