Traffic Flow Problems

Nicodemus Banagaaya
Supervisor : Dr. J.H.M. ten Thije Boonkkamp

October 15, 2009

/centre for analysis, scientific computing and applications

Outline

Introduction
Mathematical model derivation
Godunov Scheme for the Greenberg Traffic model.
Numerical experiments
Higher Order Effects.
Shock structure
Conclusion

/centre for analysis, scientific computing and applications

Outline

Introduction
Mathematical model derivation
Godunov Scheme for the Greenberg Traffic model.
Numerical experiments
Higher Order Effects.
Shock structure
Conclusion

/centre for analysis, scientific computing and applications

Outline

Introduction
Mathematical model derivation
Godunov Scheme for the Greenberg Traffic model.
Numerical experiments
Higher Order Effects.
Shock structure
Conclusion

/centre for analysis, scientific computing and applications

Outline

Introduction
Mathematical model derivation
Godunov Scheme for the Greenberg Traffic model.
Numerical experiments
Higher Order Effects.
Shock structure
Conclusion

/centre for analysis, scientific computing and applications

Outline

Introduction
Mathematical model derivation
Godunov Scheme for the Greenberg Traffic model.
Numerical experiments
Higher Order Effects.
Shock structure
Conclusion

/centre for analysis, scientific computing and applications

Outline

Introduction
Mathematical model derivation
Godunov Scheme for the Greenberg Traffic model.
Numerical experiments
Higher Order Effects.
Shock structure
Conclusion

/centre for analysis, scientific computing and applications

Outline

Introduction
Mathematical model derivation
Godunov Scheme for the Greenberg Traffic model.
Numerical experiments
Higher Order Effects.
Shock structure
Conclusion

/centre for analysis, scientific computing and applications

Outline

Introduction
Mathematical model derivation
Godunov Scheme for the Greenberg Traffic model.
Numerical experiments
Higher Order Effects.
Shock structure
Conclusion

/centre for analysis, scientific computing and applications

Introduction
Definition
Traffic flow

Figure: copyright British library Board

/centre for analysis, scientific computing and applications
Mathematical model derivation
Consider the traffic flow of the cars on a highway with only one
lane. Let,
ρ (x, t) be the density of the cars (in vehicles per kilometre),
x denote any point along the highway,
t denote the time.
v (x, t) denote the velocity of cars.
Q (x, t) = ρ (x, t) v (x, t) denote the number of cars pass
through x at time t.( flux of the vehicles given by vehicles
per unit time).
The number of cars which are the interval (x1 , x2 ) at time t is
Z x2
ρ (x, t) dx. (1)
x1

Assumption: ρ and Q are continuous functions.

/centre for analysis, scientific computing and applications
Mathematical model derivation
Consider the traffic flow of the cars on a highway with only one
lane. Let,
ρ (x, t) be the density of the cars (in vehicles per kilometre),
x denote any point along the highway,
t denote the time.
v (x, t) denote the velocity of cars.
Q (x, t) = ρ (x, t) v (x, t) denote the number of cars pass
through x at time t.( flux of the vehicles given by vehicles
per unit time).
The number of cars which are the interval (x1 , x2 ) at time t is
Z x2
ρ (x, t) dx. (1)
x1

Assumption: ρ and Q are continuous functions.

/centre for analysis, scientific computing and applications
Mathematical model derivation
Consider the traffic flow of the cars on a highway with only one
lane. Let,
ρ (x, t) be the density of the cars (in vehicles per kilometre),
x denote any point along the highway,
t denote the time.
v (x, t) denote the velocity of cars.
Q (x, t) = ρ (x, t) v (x, t) denote the number of cars pass
through x at time t.( flux of the vehicles given by vehicles
per unit time).
The number of cars which are the interval (x1 , x2 ) at time t is
Z x2
ρ (x, t) dx. (1)
x1

Assumption: ρ and Q are continuous functions.

/centre for analysis, scientific computing and applications
Mathematical model derivation
Consider the traffic flow of the cars on a highway with only one
lane. Let,
ρ (x, t) be the density of the cars (in vehicles per kilometre),
x denote any point along the highway,
t denote the time.
v (x, t) denote the velocity of cars.
Q (x, t) = ρ (x, t) v (x, t) denote the number of cars pass
through x at time t.( flux of the vehicles given by vehicles
per unit time).
The number of cars which are the interval (x1 , x2 ) at time t is
Z x2
ρ (x, t) dx. (1)
x1

Assumption: ρ and Q are continuous functions.

/centre for analysis, scientific computing and applications
Mathematical model derivation
Consider the traffic flow of the cars on a highway with only one
lane. Let,
ρ (x, t) be the density of the cars (in vehicles per kilometre),
x denote any point along the highway,
t denote the time.
v (x, t) denote the velocity of cars.
Q (x, t) = ρ (x, t) v (x, t) denote the number of cars pass
through x at time t.( flux of the vehicles given by vehicles
per unit time).
The number of cars which are the interval (x1 , x2 ) at time t is
Z x2
ρ (x, t) dx. (1)
x1

Assumption: ρ and Q are continuous functions.

/centre for analysis, scientific computing and applications
Mathematical model derivation
Consider the traffic flow of the cars on a highway with only one
lane. Let,
ρ (x, t) be the density of the cars (in vehicles per kilometre),
x denote any point along the highway,
t denote the time.
v (x, t) denote the velocity of cars.
Q (x, t) = ρ (x, t) v (x, t) denote the number of cars pass
through x at time t.( flux of the vehicles given by vehicles
per unit time).
The number of cars which are the interval (x1 , x2 ) at time t is
Z x2
ρ (x, t) dx. (1)
x1

Assumption: ρ and Q are continuous functions.

/centre for analysis, scientific computing and applications
Mathematical model derivation
Consider the traffic flow of the cars on a highway with only one
lane. Let,
ρ (x, t) be the density of the cars (in vehicles per kilometre),
x denote any point along the highway,
t denote the time.
v (x, t) denote the velocity of cars.
Q (x, t) = ρ (x, t) v (x, t) denote the number of cars pass
through x at time t.( flux of the vehicles given by vehicles
per unit time).
The number of cars which are the interval (x1 , x2 ) at time t is
Z x2
ρ (x, t) dx. (1)
x1

Assumption: ρ and Q are continuous functions.

/centre for analysis, scientific computing and applications
Mathematical model derivation

Figure: Derivation of the conservative law,where a = Q(x1 , t)

,b = Q(x2 , t)

/centre for analysis, scientific computing and applications

Mathematical model derivation

Z x2
d
ρ (x, t) dx = Q (x1 , t) − Q (x2 , t) .
dt x1

Then we have,
Z x2  
∂ρ ∂
+ Q(x, t) dx = 0 (2)
x1 ∂t ∂x
∂ρ ∂
+ (ρv ) = 0. (3)
∂t ∂x

/centre for analysis, scientific computing and applications

Mathematical model derivation

Z x2
d
ρ (x, t) dx = Q (x1 , t) − Q (x2 , t) .
dt x1

Then we have,
Z x2  
∂ρ ∂
+ Q(x, t) dx = 0 (2)
x1 ∂t ∂x
∂ρ ∂
+ (ρv ) = 0. (3)
∂t ∂x

/centre for analysis, scientific computing and applications

Mathematical model derivation

Since x1 , x2 ∈ R, t1 , t2 > 0 are arbitrary, we conclude that

∂ρ ∂
+ (ρv ) = 0, (4)
∂t ∂x
with Initial
ρ (x, 0) = ρ0 (x), ∀x ∈ R
Let flux Q = Q(ρ) , then Q(ρ) = ρV (ρ).
Thus equation ( 4) can be written as
∂ρ ∂ρ
+ c(ρ) = 0, where c(ρ) = Q 0 (ρ).
∂t ∂x

/centre for analysis, scientific computing and applications

Mathematical model derivation

Figure: Flow density curve in the traffic flow

Traffic models
Lighthill-Whitham-Richards model
 
ρ
v (ρ) = vmax 1 − , 0 ≤ ρ ≤ ρmax . (5)
ρmax
/centre for analysis, scientific computing and applications
Mathematical model derivation

Greenberg model
v(ρ) = a log ρmax
ρ , 0 < ρ ≤ ρmax .

where a is in kilometres per hour.

we are going to solve this model numerically using Godunov
scheme .

/centre for analysis, scientific computing and applications

Godunov Scheme for the Greenberg model

Godunov Scheme for Nonlinear Conservation laws

consider the initial value problem

∂u ∂f (u)
+ = 0, x ∈ R, t > 0, (6a)
∂t ∂x
u (x, 0) = u0 (x), ∀x ∈ R. (6b)

Let us introduce a control volumes or cells Vj as follows:

h 

Vj = xj− 1 , xj+ 1 , xj+ 1 = xj + xj+1 , j = 0, ±1, ±2, . . . ,
2 2 2
(7)

/centre for analysis, scientific computing and applications

Godunov Scheme for the Greenberg model

Associated with ujn is the function ū(x, t), defined as the

solution of the following initial value problem with piecewise
constant initial at t = t n :
∂ ū ∂f (ū)
+ = 0, x ∈ R, t > t n , (8a)
∂t ∂x

ū x, t n = ujn , x ∈ Vj (j = 0, ±1, ±2, . . . ). (8b)

To compute the numerical flux;


u n , if x < xj+ 1 ,
n j
ū(x, t ) = n ,
2
uj+1 if x > xj+ 1 .
2

/centre for analysis, scientific computing and applications

Godunov Scheme for the Greenberg model

The solution of this Riemann problem is a similarity of the form

x − xj+ 1
ū(x, t) = uR (η; ujn , uj+1
n
), η= 2
. (9)
t − tn
Since η = 0 for x = xj+ 1 , the computation of the numerical flux
2
we simply find

F (ujn , uj+1
n
) = f (uR (0; ujn , uj+1
n
)). (10)

Thus the Godunov scheme is given by

4t  
ujn+1 = ujn − F (ujn , uj+1
n n
) − F (uj−1 , ujn ) , (11)
4x

with the numerical flux as defined in (10).

/centre for analysis, scientific computing and applications
Godunov Scheme for the Greenberg model

And the stability condition of the method is given by

4t
max |f 0 (ujn )| ≤ 1, (j = 0, ±1, ±2, . . . ).
4x

Implementation of the scheme to the Greenberg model

x ∗ t at ρ(x ∗ , t ∗ )
x∗ = ,t = = , q(x ∗ , t ∗ ) = , ρ > 0. (12)
L L/a L ρmax

/centre for analysis, scientific computing and applications

Godunov Scheme for the Greenberg model

Where L the characteristic distance along the highway. Thus q

satisfies the conservation law
∂q ∂f (q)
∗
+ =0 (13)
∂t ∂x ∗
with the flux function f (q) = −q log q, q > 0.
The corresponding Riemann problem has the following
piecewise constant initial condition
(
ql , if x < 0,
q(x, 0) =
qr , if x > 0.

/centre for analysis, scientific computing and applications

Godunov Scheme for the Greenberg model
The Riemann Problem has two kinds of solutions;
b(q) = f 0 (q) = − log q − 1,
Case 1: b(ql ) > b(qr ) =⇒ ql < qr .
(
ql if x/t < s,
q(x, t) =
qr if x/t > s.

f (ql )−f (qr )

where s = ql −qr .
Case 2: b(ql ) < b(qr ) =⇒ ql > qr .

ql
 if x/t < b(ql ),
x
q(x, t) = e−(1+ t ) ifb(ql ) < x/t < b(qr ),

qr if x/t > b(qr ).


/centre for analysis, scientific computing and applications

Godunov Scheme for the Greenberg model

In order to implement the Godunov scheme lets us consider the

Riemann problem given below

∂ q̄ ∂
+ (−q̄ log q̄) = 0, x ∈ R, t > t n , (14)
∂t ∂x

q n if x < xj+ 1 ,
j
q̄(x, t n ) = n
2
qj+1 if x > xj+ 1 .
2

For the similarity solution of this problem we also distinguish

two cases.

/centre for analysis, scientific computing and applications

Godunov Scheme for the Greenberg model

If qjn < qj+1

n ,

(
qjn if η < sjn ,
qR (η; qjn , qj+1
n
)= n
qj+1 if η > sjn .

qj+1 log qj+1 −qj log qj

where sjn = qj −qj+1 .
if qjn > qj+1
n ,

 n
qj
 if η < b(qjn ),
n n
qR (η; qj , qj+1 ) = η if b(qjn ) < η < b(qj+1
n ),

 n
qj+1 n ).
if η > b(qj+1

/centre for analysis, scientific computing and applications

Godunov Scheme for the Greenberg model

For the numerical flux

F (ujn , uj+1
n ) = −q (0; q n , q n ) log q (0; q n , q n );
R j j+1 R j j+1
If qjn < qj+1
n , then

(
−qjn log qjn if sjn > 0,
F (qjn , qj+1
n
) = n log q n
−qj+1 j+1 if sjn < 0.

If qjn > qj+1

n , then

if qjn < e−1 ,

 n n
−qj log qj

n
F (qjn , qj+1 ) = e−1 n
if qj+1 < e−1 < qjn ,

< e−1 .
 n n n
−qj+1 log qj+1 if qj+1

/centre for analysis, scientific computing and applications

Numerical experiments
In this case the initial condition for q is given by
(
0.1 if x < 0.2,
q(x, 0) =
1 if x > 0.2.

/centre for analysis, scientific computing and applications

Numerical experiments
In this case the initial condition for q is given by
(
0.5 if x < 0.8,
q(x, 0) =
0.1 if x > 0.8.

/centre for analysis, scientific computing and applications

Higher Order Effects
q = q(ρx , ρ).
Assumptions:
q = Q(ρ) − νρx , v = V (ρ) − νρ ρx .
There are two additional effects one may wish to include the
theory: Diffusion of waves, and Response Time. To incorporate
these effects,
ρt + c(ρ)ρx = νρxx .

 
Dv 1 ν
= vt + vvx = − v − V (ρ) + ρx . (15)
Dt τ ρ

where τ - measure of the response time. The equation (15)

is to be solved together with the conservation equation.

/centre for analysis, scientific computing and applications

Higher Order Effects

If equation (15) and the conservation equation are linearized for

small perturbations about ρ = ρ0 , v = v0 = V (ρ0 ), by
substituting
ρ = ρ0 + r , v = v0 + w,
and retaining only the first powers of r and w, we have
 
0 ν
τ (wt + v0 wx ) = − w − V (ρ0 )r + rx ,
ρ0
rt + v0 rx + ρ0 wx = 0.

/centre for analysis, scientific computing and applications

Higher Order Effects

The kinematic wave speed is

c0 = ρ0 V 0 (ρ0 ) + V (ρ0 );

hence V 0 (ρ0 ) = −(v0 − c0 )/ρ0 . Introducing this expression and

then eliminating w, we have
2
∂2r

∂r ∂r ∂ ∂
+ c0 =ν 2 −τ + v0 r. (16)
∂t ∂x ∂x ∂t ∂x

/centre for analysis, scientific computing and applications

Higher Order Effects

The effect of the finite response time τ is more complicated but

can be approximated as follows.
∂ ∂
≈ −c0 . (17)
∂t ∂x
If this approximation is used in the right hand side of (16) , the
equation reduces to

∂r ∂r h i ∂2r
+ c0 = ν − (v0 − c0 )2 τ . (18)
∂t ∂x ∂x 2

/centre for analysis, scientific computing and applications

Shock structure
We need a steady profile solution of (15) and the conservation
equation with

ρ = ρ(X ), v = v (X ), X = x − Ut,

where U is the constant translational velocity.

Then,

−Uρx + (v ρ)x = 0 (19)

and may be integrated to

ρ(U − v ) = A, where A is a constant. (20)

/centre for analysis, scientific computing and applications

Shock structure
Equation (15) becomes

τ ρ(v − U)vx + νρx + ρv − Q(ρ) = 0. (21)

Since v = U − A/ρ,

A2
 
ν − 2 τ ρx = Q(ρ) − ρU + A. (22)
ρ

For τ 6= 0, the possibility that ν − A2 τ /ρ2 may vanish introduces

the new effects.
We are interested in the solution curves between ρ1 at X = +∞
and ρ2 at X = −∞. For traffic flow c 0 (ρ) = Q 00 (ρ) < 0, so
ρ2 < ρ1 and the right hand side of (22) is positive for
ρ2 < ρ < ρ1 .
/centre for analysis, scientific computing and applications
Shock structure
If ν − A2 τ /ρ2 remains positive in this range, then ρx > 0 and we
have a smooth profile as in the figure below. In view of (20), the
condition for ν − A2 τ /ρ2 to remain positive may be written
p p
ν > (v − U)2 τ, that is v − ν/τ < U < v − ν/τ . (23)

Figure: Continuous wave structure

/centre for analysis, scientific computing and applications

Conclusions

The traffic model is based on first order approximation, and

hence the original assumptions are not good
approximation.
Include higher order effects and shock structure in order to
have a better solutions.

/centre for analysis, scientific computing and applications

Thank you

/centre for analysis, scientific computing and applications