Video Summary - Teoría de Colas - Líneas de espera

Main ideas and concepts (Teoría de Colas — Líneas de espera, Parte 1)

Queues appear in many everyday services, for example:

Core point: many services involve some form of waiting line before receiving service.

Waiting creates costs for both customers and companies.

Waiting causes customer cost (opportunity cost)
- People have limited patience.
- If the wait is too long, they leave and sales are lost.
Waiting also becomes a company cost
- Indirectly through lost customers and sales due to delays.
Some waiting is extremely costly
- In health services, people may wait days, months, or years for vital care.
- Other services can be “very fast,” but that typically requires paying for more resources.
Balance required
- Trade-off between:
  - Waiting costs (customer/opportunity cost, lost sales)
  - Service delivery costs (personnel, equipment, space, technology, etc.)
Queuing theory’s objective
- Model waiting-line systems mathematically to find:
  - Steady-state behavior
  - An appropriate service capacity (service rate)
- Often aims for minimum total cost (waiting + service).

A queueing system has two key parts:

Queue: the group waiting for service
Service installation / server: the service provider where service is performed

Customers arriving individually can represent people, cars, documents, suitcases, etc.

A typical queuing model includes:

Arrivals
- Customers/vehicles/documents enter the system and, if needed, join the queue.
Queue discipline
- The rule determining who is served first after waiting.
- Common forms mentioned:
  - FCFS (First Come, First Served / PEPS in Spanish)
  - Priority rules (e.g., pregnant people, elderly people, people with impairments)
  - Last-in, First-in (LIFO-like behavior, e.g., a stack of documents)
  - Random selection (noted as possible)
Service facility (server)
- Provides service to queued units.
Outputs
- Units leave after being served.
Key clarification
- The queue excludes the unit currently being served (the server is not counted as part of the queue).

Single queue, single server
- Simplest setup (example mentioned: waiting for a bus).
Single queue, multiple servers
- Arrivals join one line; the first available server serves the next unit.
- Examples: banks, call centers, ticket/number distribution systems.
Multiple queues, multiple servers (each server has its own queue)
- Examples: supermarkets (each checkout has its own queue), toll booths (each booth has its own queue).
Sequential queues with servers (multi-stage service)
- Customers move through service stages, possibly joining a new queue at each stage.
- Example: pharmacies (order/processing queue, then a payment queue).
- Also mentioned: university enrollments and other multi-step processes.

The speaker notes other combinations are possible (e.g., one queue with several servers, followed by sequential stages).

Waiting cost
- Cost to customers for waiting (opportunity cost of time).
- Sometimes measurable directly (example: trucks waiting at a dock at $20/hour; if 4 trucks wait on average → $80 waiting cost).
Service cost
- Cost to operate the service facility (personnel, space, equipment, technology, etc.).
- Often easier for companies to estimate.
Relationship with service rate
- Increasing service capacity/service rate generally:
  - Increases service cost
  - Decreases waiting cost
Optimization objective
- Choose the service rate/system configuration that yields the minimum total cost.

Inter-arrival time
- Time between consecutive arrivals.
Average arrival rate (λ)
- Average number of arrivals per unit time.
- Average time between arrivals is 1 / λ.
- Example:
  - If λ = 20/hour, then inter-arrival time = 1/20 hours = 0.05 hours = 3 minutes.
Probability distribution for arrivals
- Arrival counts per unit time are modeled as random.
- A discrete probability distribution is mentioned as frequently used (without naming it explicitly).
- Distribution shape changes with arrival rate:
  - More asymmetric at low rates
  - More symmetric at higher rates
Independence assumption
- The arrival behavior is treated as random/independent per time-step concept (not depending on previous arrivals in the way described).

Queue size / number of customers
- Number in the system = (waiting in the queue) + (being served)
Queue capacity
- Maximum number of customers that can wait in the queue.
- Often assumed infinite, but can be finite.

Service rate (μ)
- Expected number of units served per unit time.
Expected service time
- Equals 1 / μ.
- Example: cashier serves 25 customers/hour
  - Expected service time = 1/25 hours = 0.04 hours = 2.4 minutes
Service time variability
- Two cases mentioned:
  1. Constant service time (standard deviation = 0)
  2. Exponential distribution (standard deviation = average)
    - Implies a higher probability of short service times than long ones

Initial state
- When the system starts operating (e.g., business/restaurant opening).
Steady state
- Normal ongoing operating condition.
- Peak hours are treated as abnormal; analysis focuses on steady-state behavior.
Operating characteristics (performance indicators)
- Two main measured outcomes:
  - Average number of customers
    - Waiting in the queue and waiting in the entire system
  - Average waiting time
    - Time waiting in queue and time waiting in the entire system
- These are derived from expected values.

No specific instructor or host is explicitly identified in the subtitles.
Website: www.ade.com
Social platforms: a Facebook page and YouTube channel (names not provided in the subtitles).