Traffic Flow Theory — Greenshields Model & Level of Service

1. Traffic Flow Variables — q, u, k Revisited

Traffic flow theory describes the collective behaviour of vehicles on a road using three macroscopic state variables:

A traffic flow model specifies the relationship between any two of these three variables — the third is determined from q = uk. The key model to specify is the speed-density relationship u(k), from which q(k) and q(u) are derived.

2. Greenshields Linear Model

Greenshields (1935) proposed the simplest possible speed-density relationship: a straight line decreasing from free-flow speed u_f at zero density to zero speed at jam density k_j.

2.1 Physical Justification

As density increases, drivers perceive shorter headways and reduce speed to maintain safety. The linear model assumes this reduction is proportional to density — each additional vehicle per km causes the same reduction in average speed. While this is a simplification (in reality, speed drops more sharply near jam conditions), the Greenshields model fits field data reasonably well at low to medium densities and has the important advantage of yielding simple closed-form expressions for capacity and optimal conditions.

3. Derived q–k and q–u Relationships

3.1 Flow-Density Relationship

3.2 Speed-Flow Relationship

3.3 The q–u–k Relationship Diagram

The three pairwise relationships (u–k, q–k, and q–u) form a consistent system. Key insight: for any given flow q < q_max, there are two possible states:

• Uncongested (free-flow) branch: High speed, low density — stable operation
• Congested branch: Low speed, high density — unstable, prone to stop-and-go

Traffic can be on either branch for the same flow rate — a fundamental implication for traffic management and incident response.

4. Maximum Flow, Optimum Speed, Optimum Density

Physical meaning: Maximum capacity is achieved when vehicles are travelling at exactly half their free-flow speed and half the road is “used up” by vehicles. Driving faster (lower density) or slower (higher density) both reduce the throughput below the maximum.

5. Other Traffic Flow Models

5.1 Greenberg’s Logarithmic Model

5.2 Underwood’s Exponential Model

5.3 Model Comparison

6. Microscopic Flow Models — Headway and Following Distance

6.1 Safe Following Distance

Microscopic models describe the behaviour of individual vehicles — particularly how a following driver maintains a safe gap from the lead vehicle.

6.2 Jam Density from Vehicle Length

7. Traffic Shockwave

A traffic shockwave is the boundary between two different traffic states (different q, u, k values) that propagates along the road. It occurs when a sudden change in traffic conditions — an accident, a signal turning red, a sudden speed reduction — creates a boundary between a dense slow region and a less dense fast region.

8. Level of Service (LOS)

Level of Service is a qualitative measure of traffic operating conditions — it grades the quality of traffic flow from A (best) to F (worst). LOS is used to evaluate existing road performance and set design targets for new roads.

8.1 LOS for Basic Freeway Segments (HCM Framework)

8.2 LOS for Two-Lane Rural Highways (IRC context)

For Indian two-lane highways (IRC:64), LOS is determined by the volume-to-capacity (v/c) ratio using PCU flow:

8.3 Design LOS Standards

• National and State Highways (rural): LOS C or better (v/c ≤ 0.77)
• Urban arterials and expressways: LOS C or D (v/c ≤ 0.90)
• Intersection approaches: LOS D acceptable under peak hour conditions

9. Worked Examples (GATE CE Level)

Example 1 — Maximum Flow from Greenshields Model (GATE CE 2022 type)

Problem: A road has a free-flow speed u_f = 80 km/h and a jam density k_j = 160 vehicles/km. Using the Greenshields model, find (a) the optimum speed, (b) the optimum density, and (c) the maximum flow (capacity).

Answer: u_o = 40 km/h; k_o = 80 veh/km; q_max = 3200 veh/hour

Example 2 — Speed at a Given Density (GATE CE 2021 type)

Problem: Using the Greenshields model with u_f = 60 km/h and k_j = 120 veh/km, find the space mean speed when the density is 30 vehicles/km, and compute the corresponding flow rate.

Answer: u = 45 km/h; q = 1350 vehicles/hour

Example 3 — Finding Free-Flow Speed and Jam Density from Data (GATE CE type)

Problem: Traffic observations on a highway show that at a density of 20 veh/km the speed is 70 km/h, and at a density of 60 veh/km the speed is 30 km/h. Using the Greenshields model, find u_f and k_j.

Answer: u_f = 90 km/h; k_j = 90 veh/km; q_max = 2025 veh/hour

Example 4 — Greenberg Model Application (GATE CE type)

Problem: For a road following Greenberg’s model with u_o = 35 km/h and k_j = 140 veh/km, find (a) the speed at k = 40 veh/km, (b) the maximum flow, and (c) the optimum density.

Answer: u(k=40) = 43.8 km/h; q_max = 1803 veh/hr; k_o = 51.5 veh/km

Example 5 — Two Possible States for a Given Flow (GATE MCQ type)

Problem: For a Greenshields model with u_f = 100 km/h and k_j = 200 veh/km, find the two possible densities at which q = 4000 vehicles/hour, and determine which state is the stable (uncongested) operating condition.

Answer: k₁ = 55.3 veh/km (u = 72.3 km/h, uncongested, stable); k₂ = 144.7 veh/km (u = 27.6 km/h, congested, unstable)

Example 6 — Shockwave Speed (GATE CE type)

Problem: Free-flowing traffic (State 1) has q₁ = 1800 veh/hr and k₁ = 30 veh/km. An accident reduces traffic to State 2 with q₂ = 600 veh/hr and k₂ = 120 veh/km. Find the speed and direction of the resulting shockwave.

Answer: w = –13.33 km/h — queue propagates upstream at 13.33 km/h

10. Common Mistakes

Mistake 1 — Using q_max = u_f k_j/2 Instead of u_f k_j/4

Error: Writing q_max = u_f k_j/2 (forgetting that both u and k are at half their maximum values).
Root Cause: q_max = u_o × k_o = (u_f/2) × (k_j/2) = u_fk_j/4. Multiplying two halves gives one-quarter — not one-half. A common error is taking only one factor of 1/2.
Fix: Derive it explicitly: q_max = u_o × k_o = (u_f/2) × (k_j/2) = u_fk_j/4. The factor is 1/4, not 1/2. Alternatively, substitute k = k_j/2 into q = u_fk(1–k/k_j) and confirm q = u_f(k_j/2)(1–1/2) = u_fk_j/4.

Mistake 2 — Thinking There is Only One Density for a Given Flow Rate

Error: Given a flow q < q_max, computing only one density using q = u_fk(1–k/k_j) and missing the second solution.
Root Cause: The flow-density relationship q = u_fk(1–k/k_j) is a quadratic in k — for q < q_max, it has two real roots (one < k_o, one > k_o). Students often find only the smaller root (uncongested state) without noticing the quadratic gives two answers.
Fix: Set up q = u_fk(1–k/k_j) as a quadratic in k, solve using the quadratic formula, and always interpret both solutions: smaller k = uncongested (stable, high speed); larger k = congested (unstable, low speed).

Mistake 3 — Using Time Mean Speed Instead of Space Mean Speed in the Greenshields Model

Error: Treating the “speed” u in the Greenshields model as time mean speed when comparing model predictions with field data measured as average of spot speeds (TMS).
Root Cause: The Greenshields model uses space mean speed u_s (correct for q = u × k). Field spot speed studies measure instantaneous speeds, giving TMS. TMS ≥ SMS — using TMS in the model overestimates the flow.
Fix: For traffic flow theory calculations, always use SMS in the fundamental equation and the Greenshields model. If TMS is given, convert: u_s = u_t² – σ²/u_t (approximate) or use the harmonic mean directly from individual speed data.

Mistake 4 — Wrong Sign for Shockwave Speed Direction

Error: Computing |w| = |Δq/Δk| and not keeping track of the sign — concluding the queue moves forward when it actually moves backward.
Root Cause: When a congestion forms behind an incident, the change in flow Δq = q_congested – q_free is negative (flow drops), and Δk = k_congested – k_free is positive (density increases). w = Δq/Δk < 0 → the shockwave moves upstream (against traffic). Taking the absolute value loses this crucial directional information.
Fix: Always compute w = (q₂–q₁)/(k₂–k₁) with sign. Positive w = shockwave moves with traffic; negative w = moves against traffic (upstream). The sign determines whether a queue grows upstream (backward shockwave, dangerous) or dissipates (forward shockwave).

Mistake 5 — Confusing Optimum Density k_o in Greenshields vs Greenberg Models

Error: Using k_o = k_j/2 (Greenshields result) when solving a Greenberg model problem.
Root Cause: Each model has different optimum conditions. Greenshields: k_o = k_j/2. Greenberg: k_o = k_j/e. Underwood: k_o is a model parameter. Mixing models and results gives wrong answers.
Fix: Identify the model being used first. Greenshields: all “halves” (k_o = k_j/2, u_o = u_f/2, q_max = u_fk_j/4). Greenberg: k_o = k_j/e; q_max = u_ok_j/e. Derive the optimum conditions afresh for each model from dq/dk = 0.

11. Frequently Asked Questions

Q1. Why does the Greenshields model predict two possible traffic states for the same flow rate?

This dual-state phenomenon is a fundamental feature of traffic flow, not just a mathematical artefact of the parabolic q–k curve. For any flow rate below capacity (q < q_max), traffic can operate in either a high-speed/low-density state (free-flow or uncongested) or a low-speed/high-density state (congested). The two states exist on either side of the maximum of the q–k parabola. In practice, the uncongested state is stable: if a small perturbation increases density slightly, the resulting decrease in speed sends a corrective signal upstream, and traffic returns to free-flow. The congested state is unstable: small perturbations grow into stop-and-go waves that propagate upstream. This instability explains why urban motorways can suddenly transition from steady flow to stop-and-go congestion — a single braking event on the congested side of capacity can trigger an upstream-propagating shockwave. Traffic management strategies (variable speed limits, ramp metering, incident detection) try to keep traffic on the stable uncongested branch and prevent the transition to the congested regime.

Q2. What are the limitations of the Greenshields model in describing real traffic flow?

Greenshields’ linear u–k relationship has several well-known deficiencies when compared with field data. First, it overestimates speed at high densities — real traffic tends to reduce speed more sharply than linearly as density approaches the jam value, forming a more concave u–k curve. Second, it predicts that at very low densities (k → 0) speed approaches u_f gradually, whereas real drivers quickly reach their desired speed and maintain it over a wide low-density range (a flat plateau in the u–k curve). Third, the model gives a symmetric q–k parabola with maximum at exactly k_j/2, whereas observed data typically show the maximum at lower densities (k_o ≈ 0.2–0.35 k_j). Despite these limitations, the Greenshields model remains widely used for educational purposes and analytical work because it yields simple closed-form results for capacity, optimum conditions, and shockwave speeds. More accurate but mathematically complex alternatives include the Northwestern two-regime model, the Van Aerde model, and the LWR continuum theory with more realistic fundamental diagrams.

Q3. How is traffic flow theory used in practice for Indian highway planning?

In Indian practice, traffic flow theory is applied through the Level of Service framework (IRC:64) and traffic simulation for corridor analysis. The fundamental equation q = uk and the q–k parabola inform the design of capacity improvements: if current flow is approaching q_max (LOS E), widening from two to four lanes approximately doubles capacity. Traffic flow theory also underpins the design of toll plazas (how many toll lanes are needed to prevent queue buildup), weaving sections on expressways (where merging and diverging flows interact), and ramp designs on urban flyovers. For the Bharatmala Pariyojana expressway corridor projects, NHAI requires traffic demand forecasts using four-step travel demand models, with the fundamental flow equation used to assign traffic to individual highway links. Microsimulation (VISSIM) is used for complex interchange design on NHs with heavy mixed traffic. The LOS concept is being integrated into India’s Highway Capacity Manual (under development by CSIR-CRRI), which will replace the current IRC:64 tables with more comprehensive Indian-specific capacity models.

Q4. How does traffic flow theory explain why adding capacity sometimes makes congestion worse (Braess’s paradox)?

Braess’s paradox (1968) is a counterintuitive result in network traffic theory: adding a new road (or increasing capacity on one link) to a network can sometimes increase average travel times for all users. It occurs when individually rational route-choice decisions collectively produce a worse outcome. Each driver chooses the route that minimises their own travel time — but if many drivers switch to the new road, the resulting congestion on the new link may exceed the benefit it offered. This is not captured by simple link-capacity analysis (which assumes traffic distributes evenly) but emerges from network equilibrium models (Wardrop’s user equilibrium principle). In the q–u–k framework, the instability of the congested branch plays a role: a new road may attract enough traffic to push the network onto the congested branch of multiple links simultaneously, increasing overall delay. Real-world examples include cases in Seoul, Stuttgart, and New York where removing or closing roads temporarily improved traffic flow. For Indian planning, this means that simply widening an inner-city corridor can attract induced demand that fills the additional capacity and may worsen network-level conditions unless demand management measures (parking pricing, public transport investment, staggered work hours) are implemented simultaneously.