NETWORKS AND HETEROGENEOUS MEDIA doi:10.3934/nhm.

2019011
American
c Institute of Mathematical Sciences
Volume 14, Number 2, June 2019 pp. 265–288

DERIVATION OF SECOND ORDER TRAFFIC FLOW MODELS

WITH TIME DELAYS

Michael Burger
Fraunhofer Institute ITWM
67663 Kaiserslautern, Germany

Simone Göttlich∗
University of Mannheim
Department of Mathematics
68131 Mannheim, Germany

Thomas Jung
Fraunhofer Institute ITWM
67663 Kaiserslautern, Germany

(Communicated by Seung-Yeal Ha)

Abstract. Starting from microscopic follow-the-leader models, we develop

hyperbolic delay partial differential equations to govern the density and velocity
of vehicular traffic. The proposed models can be seen as an extension of the
classical Aw-Rascle-Zhang model, where the reaction time of drivers appears
as an additional term in the velocity equation. We propose numerical methods
based on first principles and present a numerical study, where we focus on the
impact of time delays in comparison to undelayed models.

1. Introduction. Second order traffic models have been investigated since the
early 70’s (e.g. [27, 33]) and refined with the introduction of the Aw-Rascle-Zhang
(ARZ) model [3] and are still subject of current research. Traffic models in general
range from individual based microscopic to averaged macroscopic considerations,
see [9, 15, 19, 32] and the references therein. The motivation for the consideration
of second order traffic models is that on the one hand microscopic models may
explode for a large number of cars and, on the other hand, first order models are
supposed to be dependent on the traffic density only and hence lack of real-world
effects. In particular, macroscopic traffic flow models with delay terms have gained
special attention recently, see [30, 26]. While second order traffic models without
delay terms are well understood on a theoretical and analytical level [3, 9, 15],
models of this type including time delays are less considered so far [26], and if
considered, the delay is usually rewritten by a Taylor approximation [30, 26]. In
contrast, the explicit modeling of reaction times via delays is a well-known approach
in microscopic traffic modeling, see e.g. [6], and therefore the intention is that the

2010 Mathematics Subject Classification. Primary: 35L51, 90B20; Secondary: 65M06.

Key words and phrases. Traffic flow models, hyperbolic delay partial differential equation,
numerical simulations, microscopic to macroscopic, data fitting.
S. Göttlich acknowledges funding by the German Research Foundation under grant GO 1920/4-
1.

265
266 MICHAEL BURGER, SIMONE GÖTTLICH AND THOMAS JUNG

introduction of an explicit delay can make macroscopic models more reasonable.

More precisely, we expect similar effects as in microscopic delayed models such as
the developing of stop-and-go waves based on the reaction time.
Starting from car-following type models, our aim is therefore to develop and
study second order macroscopic models with an explicit dependence on the delay
term. For this purpose, we consider the explicit derivation of delayed second order
models from delayed microscopic models in three different ways following [4, 22, 30].
We will end up with a system of equations for the traffic density ρ in the sense of
mass conservation and a second equation for the velocity v with an additional delay
dependent term on the right-hand side. It turns out that for reasonably small delays
all proposed macroscopic delayed models coincide with the classical (or undelayed)
ARZ model
[3] which reads in Eulerian coordinates for some positive initial state
ρ0 (x), v0 (x) :
∂t ρ + ∂x (ρv) = 0,
∂t (ρw) + ∂x (ρvw) = 0, (1)
where w = v + P (ρ) and P (ρ) is a known pressure function.
For the numerical simulation of the delayed macroscopic models, we use finite
differences methods in a straightforward way and combine the method of lines with
solvers for delayed differential equations. We analyze numerically the delayed mod-
els and compare the solutions to the classical the ARZ model. A comparison to real
data and the corresponding fit of parameters (similar to [14]) is also presented.
This work is organized as follows: In section 2 we rigorously derive three sec-
ond order macroscopic models based on the delayed microscopic model description
according to [4, 22, 30]. The connection to the classical non-delayed ARZ model
is also shown. In section 3, we introduce and discuss numerical methods for the
solution of microscopic models with delay and second order delayed traffic models
using ideas from [5]. In section 4, we present simulation results for comparisons of
the different model level of descriptions and particularly stress on phenomena of
the delayed macroscopic models.

2. From microscopic to macroscopic models. In this section we derive macro-

scopic traffic models from delayed microscopic models of the car-following or follow-
the-leader-type using three different approaches, cf. [4, 22, 30]. The resulting models
are hyperbolic delay partial differential equations, where the time delay is explicitly
kept and appears as an additional term in the equation for the velocity. As we will
see, this new type of models gives rise to potential new dynamical behavior.

2.1. Microscopic model. The microscopic model under consideration dates back
to the 1950s and belongs to the class of car-following models, see e.g. [6]. In
particular, we consider a system of ordinary differential equations for the location
xi and speed vi of the vehicle i at time t ∈ R+
ẋi (t) = vi (t)
(vi+1 (t − T ) − vi (t − T ))
v̇i (t) = C , i = 1, . . . , N, (2)
(xi+1 (t − T ) − xi (t − T ))γ+1
with T > 0 the uniform constant reaction time and model constants C > 0 and
γ ≥ 0, cf. [4]. As initial data, we have to prescribe an initial history function on the
time interval [−T, 0] to get a well-posed solution of the problem starting at t = 0.
 TRAFFIC FLOW WITH TIME DELAY 267

Apparently, the acceleration at time t depends on the reciprocal distance between

vehicles and the difference in speed to the leading vehicle. Next, we define ∆X,
∆X ∈ R+ , as the length of a vehicle and the coefficient C = vref ∆X γ , where vref > 0
refers to the reference speed. We assume that the delay T is always positive and can
be therefore interpreted as a reaction time, in contrast to an anticipation T < 0.
Then, we rewrite (2) to
ẋi (t) = vi (t)
(vi+1 (t − T ) − vi (t − T ))
v̇i (t) = vref ∆X γ , i = 1, . . . , N. (3)
(xi+1 (t − T ) − xi (t − T ))γ+1
The microscopic model equations (3) are the staring point for the macroscopic
equations. We focus on three approaches, i.e. reverse spatial discretization [4],
coarse graining [22], Taylor method [30], to derive the corresponding macroscopic
delayed models. All approaches lead to second order traffic models that which
mainly differ in the delay term.
2.2. Reverse spatial discretization. The first approach we are interested in is
based on a derivation proposed in [4], where the classical ARZ model is derived from
(2). Here, we apply the same technique to the delayed equations (3) by identifying
the local density ρi with the free headway, i.e.,
∆X
ρi (t) := .
xi+1 (t) − xi (t)
We further define τi (t) := ρ−1
i (t) and wi (t) := vi (t) + P (τi (t)), where P (τ ) can be
understood as a function describing the anticipation of the road conditions in front
of the drivers: (
vref −γ
γ τ if γ > 0
P (τ ) :=
−vref ln(τ ) if γ = 0.
Then, the new variables τi and wi allow to equivalently reformulate system (3) as
ẋi+1 (t) − ẋi (t) vi+1 (t) − vi (t)
τ̇i (t) = =
∆X ∆X
d vref (vi+1 (t − T ) − vi (t − T ))
ẇi (t) = v̇i (t) + P (τi (t)) = + P 0 (τi (t))τ̇i (t)
dt ∆X τi (t − T )γ+1
vref (vi+1 (t − T ) − vi (t − T )) vref vi+1 (t) − vi (t)
= γ+1
− (4)
τi (t − T ) ∆X τi (t)γ+1 ∆X
since P 0 (τ ) = −vref τ γ+1
1
for γ ≥ 0. The system (4) can be interpreted as a semi-
discretization of the following delay partial differential equation for ∆X → 0:
∂t τ (x, t) = ∂x v(x, t)
 
∂t w(x, t) = vref ∂x v(x, t − T )ρ(x, t − T )γ+1 − ∂x v(x, t)ρ(x, t)γ+1 , (5)
where τ and ρ are the limiting expressions of τi and ρi . Note that the classical ARZ
model (1) in Lagrangian coordinates reads
∂t τ (x, t) = ∂x v(x, t), ∂t w(x, t) = 0. (6)
In fact, a direct computation shows that (5) is equivalent to
∂t τ (x, t) = ∂x v(x, t)
∂t v(x, t) = vref ∂x v(x, t − T )ρ(x, t − T )γ+1 (7)
268 MICHAEL BURGER, SIMONE GÖTTLICH AND THOMAS JUNG

applying the definitions of w and τ as well as information on P 0 (τ ).

Let us take a closer look at the right-hand side of the momentum equation (7).
Compared to the classical model, we now get the temporal change in speed depen-
dent on the density and the spatial derivative of speed from an earlier time. For
the interpretation of w, which is a Riemann-constant/invariant in the classical ARZ
model, we refer to [14], where w is used as the empty road speed. However, w is not
necessarily constant in the delayed model and changes due to the difference in the
product of density and the spatial derivative of speed. This means the empty road
speed, i.e. the speed a driver would have on an empty road, is not only dependent
on the driver but also on the traffic behavior around leading to a more cautious
actions.
So far, we have considered the Lagrangian macroscopic equations (5) and (7),
respectively. In a next step, we transform these equations into Eulerian coordinates
by introducing the new variables (x̂, t̂) with
∂x x̂ = τ, ∂x t̂ = 0, ∂t x̂ = v , ∂t t̂ = 1, (8)
or, exploiting t̂ = t,
∂t τ = ∂t̂ τ + (∂x̂ τ )v, ∂x v = (∂x̂ v)τ, ∂t w = ∂t̂ w + (∂x̂ w)v. (9)
In the following, we write t instead of t̂ since t̂ = t. Then, we obtain from (5) the
new equations
∂t ρ(x̂, t) + ∂x̂ (ρ(x̂, t)v(x̂, t)) = 0
∂t (ρ(x̂, t)w(x̂, t)) + ∂x̂ (ρ(x̂, t)v(x̂, t)w(x̂, t)) = vref ∂x̂ v(x̂, t − T )ρ(x̂, t − T )γ
− ∂x̂ v(x̂, t)ρ(x̂, t)γ


(10)
which are obviously closely related to the undelayed ARZ model.

2.3. Coarse-graining. Another approach to derive a delayed macroscopic equa-

tion is the so-called coarse-graining (CG) proposed in [12, 22]. Here, we focus on
[22] and highlight the most important steps. Although the method is different from
reverse spatial discretization (RSD) from subsection 2.2, we note that the resulting
equations are quite similar.
We start by defining the density ρ̂(x, t) and flux field q̂(x, t) of traffic flow as
follows:
X X
ρ̂(x, t) := δ(xi (t) − x), q̂(x, t) := ẋi (t)δ(xi (t) − x), (11)
i i

where xi (t) is the position of the i-th car at time t and δ denotes the δ-distribution.
We now coarse-grain these fields and obtain
Z
ρ(x, t) := Φ(x − x0 , t − t0 )ρ̂(x0 , t0 )dx0 dt0 ,
Z
q(x, t) := Φ(x − x0 , t − t0 )q̂(x0 , t0 )dx0 dt0 (12)

by applying the coarse grainingRenvelope function Φ(x, t) which has a peak at (0, 0)
and is normalized in the sense Φ(x, t)dxdt = 1. According to [22], we can derive
the following system of equations from (12) for the density ρ and flux q:
∂ ∂
ρ(x, t) + q(x, t) = 0
∂t ∂x
 TRAFFIC FLOW WITH TIME DELAY 269

∂ ∂
q(x, t) = ρ(x, t)hẍi (t0 )i(x,t) − [ρ(x, t)hẋ2i (t0 )i(x,t) ] (13)
∂t ∂x
with the bracketed average of a quantity fi (x, t) defined as
Z
0 0 1 X
hfi (x , t )i(x,t) = Φ(x − x0 , t − t0 ) fi (x0 , t0 )δ(xi (t0 ) − x0 )dx0 dt0 .
ρ(x, t) i
q(x,t)
By introducing v(x, t) = hẋi (t0 )i(x,t) = ρ(x,t) and Θ(x, t) = hẋ2i (t0 )i − v 2 (x, t) we
can rewrite the second equation in (13) to
∂ ∂ ∂
ρ(x, t)( v(x, t) + v(x, t) v(x, t)) = ρ(x, t)hẍi (t0 )i(x,t) − (ρ(x, t)Θ(x, t)). (14)
∂t ∂x ∂x
The first equation for the density is obviously the same as in (10) and the derivation
so far is independent on microscopic dynamics. However, the second equation for
the momentum involves a term hẍi (t0 )i(x,t) which is dependent on individual drivers’
behavior and therefore needs a careful discussion.
Let us first consider the general case of a car-following model of the form
ẍi (t) = B(∆xi , ∆ẋi , ẋi ).
Then coarse graining leads to the approximated momentum equation
∂ ∂
v(x, t) + v(x, t) v(x, t) ≈ B(h∆xi i, h∆ẋi i, v),
∂t ∂x
where we further approximate
1 ∂ −1 1 ∂ 2 −1
h∆xi i ≈ ρ−1 (x, t) + ρ (x, t) + 2 ρ (x, t)
2ρ(x, t) ∂x 6ρ (x, t) ∂x2
∂ 1 ∂2
h∆ẋi i ≈ ρ−1 (x, t) v(x, t) + 2 v(x, t).
∂x 2ρ (x, t) ∂x2
Plugging in and expanding around (ρ−1 (x, t), 0, v(x, t)) yields
B(h∆xi i, h∆ẋi i, v(x, t)) ≈B(ρ−1 (x, t), 0, v(x, t))
1 ∂ −1 ∂
+B1 ( ρ (x, t)) + B2 (ρ−1 (x, t) v(x, t)),
2ρ(x, t) ∂x ∂x
where Bi = ∂zi B(z1 , z2 , z3 )|(ρ−1 ,0,v) . We end up with an equation of type
∂t v(x, t) + v(x, t)∂x v(x, t) = B(ρ−1 (x, t), 0, v(x, t))
B1 B2
+ ∂x ρ−1 (x, t) + ∂x v(x, t). (15)
2ρ(x, t) ρ(x, t)
We remark that it is also possible to include terms of higher order leading to higher
order approximations as well as higher order derivatives. Since we are interested in
delayed models driven by (3), we can identify B as follows:
∆ẋi (x, t − T )
B(∆xi , ∆ẋi , ẋi ) = C
∆xi (x, t − T )γ+1
B1 = 0, B2 = Cρ(x, t − T )γ+1 , B3 = 0. (16)
The system of equations is then
∂t τ (x, t) = ∂x v(x, t)
∂t v(x, t) + v(x, t)∂x v(x, t) = vref ρ(x, t − T )γ ∂x v(x, t). (17)
270 MICHAEL BURGER, SIMONE GÖTTLICH AND THOMAS JUNG

In conservative form, we may write the system of equations

∂t ρ(x, t) + ∂x (ρ(x, t)v(x, t)) = 0
∂t (ρ(x, t)w(x, t)) + ∂x0 (ρ(x, t)v(x, t)w(x, t)) = vref [∂x v(x, t)ρ(x, t − T )γ
− ∂x v(x, t)ρ(x, t)γ ] (18)
which is (up to the term ∂x v(x, t)) directly comparable to (10).
For completeness, we also state the Lagrangian representation of (18). If we use
the inverse transformation from subsection 2.2 and write the equations dependent
on τ instead of ρ we end up with
∂t τ (x, t) = ∂x v(x, t)
∂t v(x, t) = vref ρ(x, t − T )γ+1 ∂x v(x, t). (19)

2.4. Taylor expansion. The last approach we present is based on a Taylor expan-
sion which was originally introduced in [30]. However, in contrast to the original
approach, we derive a second order delayed model instead of a first order model
only. Furthermore, we keep the explicit delay while the new model is derived from
the microscopic level and do not apply a diffusive approximation.
We start by expanding the delayed variables position xi and speed vi up to first
order. This leads directly to
xi (t − T ) = xi (t) − T ẋi (t) + O(T 2 )
and equivalently
vi (t − T ) = vi (t) − T v̇i (t) + O(T 2 ).
By using the microscopic description (3) we can characterize the following delayed
system:
xi (t − T ) = xi (t) − T vi (t) + O(T 2 )
 (vi+1 (t − T ) − vi (t − T )) 
vi (t − T ) = vi (t) − T C + O(T 2 ), i = 1, . . . , N.
(xi+1 (t − T ) − xi (t − T ))γ+1
(20)
The definition ∆xi (t) := xi+1 (t) − xi (t) then allows to approximate
∆xi (t − T ) ≈ ∆xi (t) − T ∆vi (t)
and
 ∆vi+1 (t − T ) ∆vi (t − T ) 
∆vi (t − T ) ≈ ∆vi (t) − T C − C .
(∆xi+1 (t − T ))γ+1 (∆xi (t − T ))γ+1
∆X
To derive macroscopic equations, we use again ρi (t) := xi+1 (t)−xi (t) and obtain
the conservation of mass by introducing τi = ρ−1
i in the limit ∆X → 0:
∂t τ (x, t) = ∂x v(x, t),
where τ and ρ denote the respective limit versions of τi and ρi . The second equation
for the velocity is derived as follows:
∆vi (t − T )
∂t vi (t) = C
∆xi (t − T )γ+1
1
≈C
(∆xi (t) − T ∆vi (t))γ+1
 TRAFFIC FLOW WITH TIME DELAY 271

 ∆vi+1 (t − T ) ∆vi (t − T ) 
× ∆vi (t) − T [C − C ]
(∆xi+1 (t − T ))γ+1 (∆xi (t − T ))γ+1
1
≈C 1 1
ρi (t)γ+1 − (γ + 1) ρi (t)γ T ∆vi (t)


× ∆vi (t) − CT ∆vi+1 (t − T )(ρi+1 (t − T ))γ+1 − ∆vi (t − T )(ρi (t − T ))γ+1 .

For simplicity, we omit the terms O(T 2 ) in the latter representation. Doing the
same scaling procedure and using the same definitions as above, we end up with
∂t τ (x, t) = ∂x v(x, t)
vref ρ(x, t)γ+1
∂t v(x, t) =
1 − (γ + 1)ρ(x, t)T ∂x v(x, t)
× ∂x v(x, t) − 2vref T (ρ(x, t − T )γ+1 ∂x v(x, t − T ))


(21)
in Lagrangian coordinates. However, the interpretation of the right-hand side is
now different compared to (7) and (19).
By the use of Taylor expansion, correction terms enter the model. This can be
interpreted as a correction towards the fact that drivers do not react immediately
but after some delay. These correction terms are, as in the previous models, depen-
dent on the product of density, the spatial derivative of speed and the (current and
past) state of the traffic. Note that in the case we ignore the terms depending on
T , we recover the classical ARZ model.

3. Properties of delayed macroscopic models. As already seen, the introduc-

tion of time delays lead to a right-hand side in the equation for the velocity that
includes different states of traffic. This can be interpreted as an anticipation of the
change in traffic based on the current and past state. Hence, we are able to model
the drivers ability to ‘extrapolate’ the traffic situation to change his/her driving
behavior.
Similar to the microscopic case, the introduction of explicit delays requires the
extension of the initial data time span to ensure well-posedness. For all states (ρ, v)
that appear in a delayed form, initial values at t = 0 are not enough and we need
initial histories. These are functions prescribing the states on at least the interval
[−T, 0].
We now aim to discuss some key properties of the delayed macroscopic models.
First, we show that for vanishing delays the models coincide with the classical ARZ
model. In this context, we also refer to the weak solutions of the ARZ model and
investigate the changes caused by the delayed case. Second, we comment on the
positivity of solutions. Closely related to this discussion is the question of stability.
We inverstigate the system of delayed differential equations (DDEs) resulting from
discretizing in space and derive properties to ensure stable solutions.

3.1. Convergence to the classical ARZ model.

Lemma 3.1. The macroscopic delayed models (7), (19) and (21) converge to the
undelayed ARZ model for T → 0.
Proof. The first equation of all proposed models is mass conservation and has no
explicit dependence on the delay T . We focus on the second equation in Lagrangian
coordinates and need to show that ∂t w(x, t) = 0 is satisfied.
272 MICHAEL BURGER, SIMONE GÖTTLICH AND THOMAS JUNG

Therefore, the velocity equation needs to be expressed in terms of w for all models
(7), (19) and (21). We use that w = v + P and get the following representation for
the models derived by reverse spatial discretization (7) and coarse-graining (19):
 
∂t w(x, t) = vref ∂x v(x, t − T )ρ(x, t − T )γ+1 − ∂x v(x, t)ρ(x, t)γ+1
and  
∂t w(x, t) = vref ∂x v(x, t)ρ(x, t − T )γ+1 − ∂x v(x, t)ρ(x, t)γ+1 ,
respectively. The velocity equation for the model derived by Taylor expansion (21)
reads
vref
∂t w(x, t) = 1 1
ρ(x,t)γ+1 − (γ + 1) ρ(x,t)γ T ∂x v(x, t)

× (∂x v(x, t) − 2vref T ∂x [ρ(x, t − T )γ+1 ∂x v(x, t − T )]) − vref ∂x v(x, t)ρ(x, t)γ+1 .
Then, in the limit T → 0, we end up with ∂t w(x, t) = 0 for all equations.
3.2. Comparison to weak solutions of the ARZ model. In this section, we
discuss the solutions to Riemann problems for the ARZ model (1) and explain
how the delayed models fit into this framework. The ARZ model is a system of
conservation laws and we have three basic types of weak solutions to the Riemann
problem, i.e. contact discontinuities, shocks and rarefaction waves, see [3, 15] for
an overview. We define u = (ρ, ρw)T and rewrite the fluxes for the ARZ model as
   
fρ ρv
f= = . (22)
fρw ρwv
For further investigations, we consider the ARZ model in the form
∂t u + ∇f ∂x u = 0,
where λi (u) and ri (u) are the eigenvalues and eigenvectors of ∇f , dependent of
the state vector u. After defining genuine nonlinear and linear degenerate eigen-
values, we can state the weak solutions. For the ARZ model, the eigenvalues and
eigenvectors are
 
1
λ1 = v − ρP (ρ), r1 = (23)
w
 
1
λ2 = v, r2 =
w + ρP 0 (ρ)
with λ1 being genuine nonlinear and λ2 being linear degenerate. While lifting this
analysis to the delayed models, the first problem occurs since ∇f (and its properties)
needs to be derived. However, except for γ = 0, a suitable flux function has not
been found yet since it would depend on the current and past state, respectively.
Now, let us focus on the weak solutions to Riemann problems with left state
ul = (ρl , ρl wl )T and right state ur = (ρr , ρr wr )T . The contact discontinuity belongs
to the linear degenerate eigenvalue and appears when λ2 (ul ) = λ1 (ur ). This means
vl = vr . The solution is then given by the initial date moving with vl = vr
u(t, x) = u0 (x − vl t). (24)
For the delayed models, we do not necessarily get the same solution for vl = vr .
However, if also the history is such that vl = vr for all [−T, 0], the solution looks
quite similar.
 TRAFFIC FLOW WITH TIME DELAY 273

For the shock solution, we consider the Rankine Hugoniot condition

f (ul ) − f (ur ) = s(ul − ur ).

This gives for every state a curve which can be connected to the latter by a shock.
The shock solution belongs to the genuine nonlinear eigenvalue. In the ARZ model,
a fixed ul can be connected by a shock to states u with wl = w. Furthermore, the
shock is admissible if ρl < ρr and looks like

u(t, x) = u0 (x − st). (25)

It remains unclear how the delayed models fit into this framework due to the missing
definition of a flux function and hence the evaluation of the Rankine Hugoniot
condition.
Similar to the shock solution, the rarefaction wave also belongs to the genuine
nonlinear eigenvalue λj . The solution is given by

ul
 x < λ1 (ul )t
u(t, x) = v( xt ) λ1 (ul ) < x < λ1 (ur ) , (26)

ur x > λ1 (ur )t


where v is such that v 0 (ζ) = rj (v(ζ)) and λj (v(ζ)) = ζ. For the ARZ model, it
holds v(ζ) = (ζ, wl )T . In the case of delayed models, no conclusions can be made
since eigenvalues and eigenvectors cannot be determined.
To illustrate the behavior of approximate solutions to the delayed model, we refer
to our numerical study in section 4, in particular figures 5, 6 and 7.

3.3. Positivity of solutions. The early proposed second order models, i.e. the
Payne-Whitham-model, suffered from the problem that in some Riemann problem-
s the density and velocity could become negative. This was first pointed out by
Daganzo [10] and the ARZ model has been developed to overcome this drawback.
Unfortunately, the delayed models we have derived, converge certainly to the classi-
cal ARZ model, but might lose the positivity of v and ρ due to the delay terms. This
can be seen by considering the momentum equation for each model and assuming
sufficiently smooth initial data.
For the first approach (reverse spatial discretization) we have to analyze equation
(7). We investigate the case that we have non-negative initial data and need to
ensure that the solution remains non-negative. However, the solution would become
negative in the case v(x, t) = 0 and ∂t v(x, t) < 0, i.e.,

vref ∂x v(x, t − T )ρ(x, t − T )γ+1 < 0

with vref and ρ(x, t − T )γ+1 are both positive by assumption. However, for the term
∂x v(x, t − T ) no statement is possible and hence the velocity can become negative.
For the coarse-graining approach (19) we have a different situation:

∂t v(x, t) = vref ρ(x, t − T )γ+1 ∂x v(x, t).

Since the term ∂x v(x, t−T ) does not appear here, but ∂x v(x, t) instead, the solution
could become negative only if ∂x v(x, t) < 0. So assuming non-negative data and
v = 0, we have at least for differentiable solutions that ∂x v(x, t) = 0 (since v = 0 is
a minimum) and thus v remains positive.
274 MICHAEL BURGER, SIMONE GÖTTLICH AND THOMAS JUNG

The last approach, the Taylor model (21), would become negative if v = 0 and
the term
vref
1 1 (∂x v(x, t)−2vref T ∂x [ρ(x, t−T )γ+1 ∂x v(x, t−T )])
ρ(y,t)γ+1 − (γ + 1) ρ(x,t) γ T ∂x v(x, t)
is smaller than 0. Using the same arguments as before, we can reduce the latter
expression to
−ρ(x, t)γ+1 ∂x [ρ(x, t − T )γ+1 ∂x v(x, t − T )] < 0
and can conclude that the speed can become negative.
Remark 1. Negative velocities also appear in the microscopic model (2). This is
the case when a driver is quite close to the vehicle in front and the driver in front
suddenly reduces the velocity within the given delay time, see e.g. [21] where this
effect is seen for a similar microscopic model. The non-physical solutions obtained
in this way are avoided by the choice of our numerical experiments.
3.4. Stability of discretized system. From theory we know the concept of linear
stability for DDEs, see [25]. To apply this idea to the PDE-type delayed models we
have derived, we need to discretize the equations in space. For the model based on
reverse spatial discretization (7), we get for γ = 0
v(xi+1 , t) − v(xi , t)
∂t τ (xi , t) =
∆x
1 v(xi+1 , t − T ) − v(xi , t − T )
∂t v(xi , t) = vref
τ (xi , t − T ) ∆x
exploiting τ (x, t − T )−1 = ρ(x, t − T ). The steady state is obtained by v ∗ constant
in time and space, i.e. v ∗ (xi , t) = v ∗ (xi+1 , t) = v ∗ (xi+1 , t − T ) = v ∗ (xi , t − T ), and
τ ∗ arbitrary. We disturb this steady state (τ ∗ , v ∗ )T by adding a small disturbance
η(t). For the linear stability analysis, we look at the linearized system around the
steady state, which is
−v(xi , t)
∂t τ (xi , t) =
∆x
1 −v(xi , t − T )
∂t v(xi , t) = vref ∗ .
τ ∆x
Inserting the disturbed steady state, we get
−η2 (xi , t)
∂t η1 (xi , t) =
∆x
1 −η2 (xi , t − T )
∂t η2 (xi , t) = vref ∗ .
τ ∆x
Since η2 is independent of η1 , we consider η2 first. Assuming the solution η2 (omit-
ting xi ) has the form η2 (t) = η0,2 eλ2 t , this gives
1 −η0,2 eλ2 (t−T )
η0,2 λ2 eλ2 t = vref .
τ∗ ∆x
This equation is equivalent to
1 −e−λ2 T
λ2 = vref .
τ ∗ ∆x
This type of equation can be solved for λ2 using the Lambert W −function, see [8],
1  −T vref 
λ2 = W ,
T τ ∗ ∆x
 TRAFFIC FLOW WITH TIME DELAY 275

Figure 1. Real parts λ2 for γ = 0.

and hence the solution for η2 (t). In a next step, η1 is given by integrating
η0,2 λ2 t
η1 (t) = − e + C.
∆xλ2
This means, we have stability in a linear sense if Re(W ( −T vref
τ ∗ ∆x )) < 0. For some
values of T and ∆x the real part is shown in figure 1. We observe that the system
is stable for small T . This result seems plausible since a large delay is known to
destabilize the traffic system in the microscopic case, cf. [21]. Furthermore, small
step sizes ∆x also lead to unstable systems.
4. Numerical results. First, we have to introduce suitable discretization schemes
for the microscopic and macroscopic delayed models. We make use of already well-
established numerical methods for undelayed models and discuss how the delay term
can be treated within this framework.
The microscopic model (2) is a system of DDEs. The numerical discretization of
the delay term can be done with method of steps, cf. [5]. The concept behind this
method is to separate the original initial history problem into several initial value
problems stated by ordinary differential equations (ODE) and then apply standard
schemes for ODEs.
To do so, we need to split our foreseen time interval into sub-intervals of smaller
size such that the delay is still included. If we look at the first interval, we have
all delayed values given by the initial history since the interval is smaller than the
delay. Therefore, we can plug in the history and get a classical ODE for this first
interval that needs to be solved numerically. For the second interval, we can plug
in the history and the solution of the first interval for the delayed parts and get
again an ODE. Iteratively, we use this strategy to solve the problem on the whole
time interval. A good interpolation of the solution is necessary since we might need
values in later iterations which do not correspond to the fixed time grid points.
We use the Matlab solver dde23 to simulate (2). This solver uses an explicit
Runge-Kutta (2,3) pair combined with the method of steps.
0 https://de.mathworks.com/help/matlab/ref/dde23.html
276 MICHAEL BURGER, SIMONE GÖTTLICH AND THOMAS JUNG

For the comparison with the microscopic model, we use a method-of-lines like
approach for the discretization of the delayed macroscopic model. We discretize
the model in space to get a system of DDEs and apply a forward finite differences
discretization to (7), (19) and (21) in Lagrangian coordinates as follows:
Discretization of (7):
v(xi+1 , t) − v(xi , t)
∂t τ (xi , t) =
∆x
v(xi+1 , t − T ) − v(xi , t − T )
∂t v(xi , t) = vref ρ(xi , t − T )γ+1 .
∆x
Discretization of (19):
v(xi+1 , t) − v(xi , t)
∂t τ (xi , t) =
∆x
v(xi+1 , t) − v(xi , t)
∂t v(xi , t) = vref ρ(xi , t − T )γ+1 .
∆x
Discretization of (21):
v(xi+1 , t) − v(xi , t)
∂t τ (xi , t) =
∆x
vref ρ(xi , t)γ+1
∂t v(xi , t) =
1 − (γ + 1)ρ(x, t)T v(xi+1 ,t)−v(x
∆x
i ,t)

v(xi+1 , t) − 2vref T ρ(xi+1 , t − T ) v(xi+2 ,t−T )−v(x

∆x
i+1 ,t−T )

×
∆x
!
v(xi , t) + 2vref T ρ(xi , t − T ) v(xi+1 ,t−T∆x
)−v(xi ,t−T )
− .
∆x
The resulting system can be solved with the method of steps again using the
Matlab solver dde23. All discretization procedures need an initial history function.
For simplicity, we choose a constant history function with the value of the states at
t = 0.
We remark that the different numerical schemes are used because they showed
to be better suited for the respective equation. Moreover, due to the potential
instabilities, we can not decrease ∆x arbitrarily, which in fact poses a problem.

4.1. Comparison of delayed models.

4.1.1. Microscopic versus macroscopic approaches. The first numerical experiment

we focus on is the numerical comparison between the microscopic model (2) and the
different derivations of delayed macroscopic models called reverse spatial discretiza-
tion (RSD) - eq. (7), coarse-graining (CG) and Taylor expansion (TE). Therefore,
we present the following scenario: We consider a street with constant initial densi-
ty in the macroscopic case which means initially uniformly distributed cars in the
microscopic situation, i.e. ρ0 = 0.1. For the initial speeds on the first half of the
road, the traffic is slow, i.e., v0 = 0.25, while on the second half the traffic moves
significantly faster, i.e. v0 = 0.5. In this way, we create a vacuum situation, where
the traffic in the second half moves away from the slower cars in the first section.
The spacegrid is equidistant with ∆x = 1 and the timestep is chosen by the DDE
solver.
 TRAFFIC FLOW WITH TIME DELAY 277

We simulate a time interval of 100s and the delay is T = 0.5. The boundary
conditions for the macroscopic models are introduced by ghost cells copying the
first and last cell values. The results are shown in Figure 2 and 3 for a snapshot of
the solution.

0.12

0.1
Density

Micro
0.08
RSD
0.06 CG
TE
260 265 270 275 280 285 290 295 300 305 310
Space

0.05 Micro
RSD
Flow

0.04 CG
TE
0.03

260 270 280 290 300 310

Space

Figure 2. Time evolution of density and flux of the delayed mi-

croscopic and macroscopic models for T = 0.5

Micro
0.1
RSD
Density

0.08 CG
TE
0.06

290 291 292 293 294 295 296 297 298 299 300
Space

0.05 Micro
RSD
Flow

0.04 CG
TE
0.03

290 292 294 296 298 300

Space

Figure 3. Zoom: Time evolution of density and flux of the delayed

microscopic and macroscopic models for T = 0.5

We see that the models resolve the vacuum solution well and are close to each
other.
If we now zoom in space and consider different values for ∆x, we observe the
convergence behavior as depicted in Table 1 and Figure 4.

4.1.2. Classical ARZ model versus macroscopic approaches. In section 3.2 we have
seen that weak solutions to the classical ARZ model are hard to interpret for the
delayed models. However, from a numerical point of view, a first idea of weak
solutions to the delayed model can be given. Here, we focus on the numerical
solutions to the RSD model and make comparisons to the three different types of
278 MICHAEL BURGER, SIMONE GÖTTLICH AND THOMAS JUNG

∆ x=5 ∆ x=1
0.105 0.105

0.1 0.1

0.095 0.095

0.09 0.09
Micro Micro

Density

Density
0.085 0.085

0.08 0.08

0.075
RSD 0.075
RSD
0.07 0.07

0.065 0.065

0.06 0.06
0 100 200 300 400 500 600 0 100 200 300 400 500 600
space Space

∆ x=0.5 ∆ x=0.1
0.105 0.105

0.1 0.1

0.095 0.095

0.09 0.09
Micro Micro
Density

Density
0.085 0.085

0.08 0.08

0.075
RSD 0.075
RSD
0.07 0.07

0.065 0.065

0.06 0.06
0 100 200 300 400 500 600 0 100 200 300 400 500 600
Space Space

∆ x=5.000000 ∆ x=1.000000
0.105 0.105

0.1 0.1

0.095 0.095

0.09 0.09
Micro Micro
Density

Density

0.085 0.085

0.08 0.08

0.075
CG 0.075 CG
0.07 0.07

0.065 0.065

0.06 0.06
0 100 200 300 400 500 600 0 100 200 300 400 500 600
Space Space

∆ x=0.500000 ∆ x=0.100000
0.105 0.105

0.1 0.1

0.095 0.095

0.09 0.09
Micro Micro
Density

Density

0.085 0.085

0.08 0.08

0.075
CG 0.075
CG
0.07 0.07

0.065 0.065

0.06 0.06
0 100 200 300 400 500 600 0 100 200 300 400 500 600
Space Space

∆ x=5.000000 ∆ x=1.000000
0.105 0.105

0.1 0.1

0.095 0.095
Density

Density

0.09
Micro 0.09 Micro

0.085 0.085
TE TE
0.08 0.08

0.075 0.075
0 100 200 300 400 500 600 0 100 200 300 400 500 600
Space Space

∆ x=0.500000 ∆ x=0.100000
0.105 0.105

0.1
0.1
0.095

0.095 0.09
Micro
Density

Density

0.085
0.09
Micro
0.08

0.085 0.075
TE
TE
0.07
0.08
0.065

0.075 0.06
0 100 200 300 400 500 600 0 100 200 300 400 500 600
Space Space

Figure 4. Convergence of the microscopic to the macroscopic

models at time t = 10 and fixed delay T = 0.5
 TRAFFIC FLOW WITH TIME DELAY 279

Table 1. Comparison of errors for different ∆x

∆x 5 1 0.5 0.1
RSD || · ||2 -Error 0.074 0.0759 0.0706 0.0335
RSD || · ||∞ -Error 0.038 0.0337 0.027 0.0288
CG || · ||2 -Error 0.0727 0.0746 0.0692 0.0308
CG || · ||∞ -Error 0.036 0.0305 0.0236 0.0288
TE || · ||2 -Error 0.0651 0.0683 0.0657 0.0486
TE || · ||∞ -Error 0.0223 0.0181 0.0236 0.0288

Contact Discontinuity
0.8
ARZ Ana
0.6 ARZ
Density

RSD
0.4

0.2
0 100 200 300 400 500 600 700 800 900 1000
Space

2
ARZ Ana
1.5 ARZ
Flow

RSD
1

0.5
0 100 200 300 400 500 600 700 800 900 1000
Space

Figure 5. Comparison for contact discontinuity: numerical RSD

and ARZ solutions vs. analytical ARZ

Shock Solution
0.8
ARZ Ana
0.6 ARZ
Density

RSD
0.4

0.2
0 100 200 300 400 500 600 700 800 900 1000
Space

1.8
ARZ Ana
1.6
ARZ
Flow

1.4 RSD
1.2
1
0 100 200 300 400 500 600 700 800 900 1000
Space

Figure 6. Comparison for shock solution: numerical RSD and

ARZ solutions vs. analytical ARZ

Riemann solutions, see (24) - (26), on the same grid sizes. The results are given in
figures 5, 6 and 7.
In a second experiment, we compare the delayed macroscopic models to the
classical ARZ model for a delay of T = 0.5 and T = 5, respectively, see figure 8 and
9.
280 MICHAEL BURGER, SIMONE GÖTTLICH AND THOMAS JUNG

Rarefaction Wave
0.8
ARZ Ana
0.6 ARZ

Density
RSD
0.4

0.2
0 100 200 300 400 500 600 700 800 900 1000
Space

1.5
ARZ Ana
ARZ
1
RSD
Flow

0.5

0
0 100 200 300 400 500 600 700 800 900 1000
Space

Figure 7. Comparison for rarefaction wave: numerical RSD and

ARZ solutions vs. analytical ARZ

0.1 0.07
0.065
0.09 0.06
Density

ARZ 290 292 294

0.08 RSD
0.07 CG
TE
0.06
250 260 270 280 290 300 310
Space

0.05 ARZ 0.034

0.032
RSD 0.03
0.028
0.04 CG 0.026
Flow

290 292 294 296

TE
0.03

0.02
250 260 270 280 290 300 310
Space

Figure 8. Comparison of macroscopic models for T = 0.5 with Zoom

Due to lemma 3.1, all delayed macroscopic models converge to the classical ARZ
model, cf. Figure 8. However, the situation changes for a larger delay, e.g. T = 5.
We see a slightly different behavior of the models at the point, where the density
changes from falling to rising, cf. Figure 9. There are additional sections with
different slopes in the delayed models. This is obviously an effect driven by the
time delay as the direct comparison with Figure 8 shows.
4.1.3. Taylor model (TE) versus convection-diffusion flow model [30]. The third
experiment compares our TE approach (21) to the first order convection-diffusion
flow model (TCHS) derived in [30]. Since the TE approach is inspired by the
derivation introduced in [30], we aim to point out how the two approaches differ
regarding the explicit tracking of the delay, cf. discussion in subsection 2.4.
The model in [30] is given by
∂t ρ + ∂x (ρV (ρ)) = −T ∂x ((ρV (ρ))2 ∂x ρ)
with V (ρ) = (1 − ρ) and reads in Langrangian coordinates
∂t ρ − ρV (ρ)∂x ρ + ρ∂x (ρV (ρ)) = −T ρ∂x ((ρV (ρ))2 ρ∂x ρ)
 TRAFFIC FLOW WITH TIME DELAY 281

0.1
0.07
0.09 0.065

Density
ARZ 0.06
0.08 292 293 294
RSD
CG
0.07
TE
0.06
265 270 275 280 285 290 295 300
Space

0.05
ARZ 0.035
RSD 0.03
0.04 CG
Flow

TE 290 291 292 293 294

0.03

260 265 270 275 280 285 290 295 300 305
Space

Figure 9. Comparison of macroscopic models for T = 5 with Zoom

This model is solved by a finite difference based method of lines discretization,

where the ODE solver is a forward Euler method.
For our comparison, we consider a street with a smooth initial density profile such
that we have a relatively high density in the beginning, i.e. ρ0 (x) = 0.7, a density
drop in the middle, i.e. ρ0 (x) = 0.5, and let the density rise again to ρ0 (x) = 0.6
in the end, cf. Figure 10. The time delay is T = 0.5s and T = 5s, as for smaller
delays the difference is not as pronounced. The total time horizon is 30s and the
discretization is ∆x = 0.5, ∆t = 0.01 in space and time, respectively. The results
can be seen in Figure 10 after 30s. The Taylor model (21) introduces some new
effects we do not see in the models TCHS and ARZ, in particular at x = 350 when
the density arises from 0.5 to 0.6 and vehicles start to queue. In the TCHS model,
all dependencies on the delay of higher order are excluded while the model keeping
the delay still leads to oscillatory effects. We also see that the oscillations increase
for larger delays.

TCHS
0.7 ARZ
TE Delay 0.5
TE Delay 5
Initial Profile
0.65
Density

0.6

0.55

0.5

0 100 200 300 400 500

Space

Figure 10. Comparison TE model and convection-diffusion flow

model for T = 0.5
282 MICHAEL BURGER, SIMONE GÖTTLICH AND THOMAS JUNG

4.2. Applications of delayed macroscopic models. For the next numerical

studies and applications, we introduce an alternative numerical discretization of
the macroscopic models based on the Eulerian coordinates representation. This is
useful since the scenarios we investigate are on a fixed road segment and so Eulerian
coordinates are better suited to tackle the boundary conditions. To set up a suitable
numerical scheme, we use known methods for hyperbolic partial differential (PDE)
equations coupled to a splitting technique, cf. [23]. This means, considering for
example the reverse spatial discretization (RSD), we split the model (10) into the
classical ARZ model solved in a first step and an additional delayed part solved in
a second step with a forward differences approach. We solve the hyperbolic PDE
part with the Lax-Friedrichs method using the index j as space index. Then, we
get for (10)
1
ρ(xj , t + ∆t) = (ρ(xj+1 , t) + ρ(xj−1 , t))
2
∆t
− (fρ (ρ(xj+1 , t), ρw(xj+1 , t)) − fρ (ρ(xj−1 , t), ρw(xj−1 , t)))
2∆x
1
ρw∗ (xj , t) (ρw(xj+1 , t) + ρw(xj−1 , t))
2
∆t
− (fρw (ρ(xj+1 , t), ρw(xj+1 , t)) − fρw (ρ(xj−1 , t), ρw(xj−1 , t)))
2∆x
ρw(xj , t + ∆t) = vref
 ρw(xj+1 ,t−T ) − P 1

ρw(xj−1 ,t−T ) 1


ρ(xj+1 ,t−T ) ρ(xj+1 ,t−T ) − ρ(xj−1 ,t−T ) + P ρ(xj−1 ,t−T )
× ρ(xj , t − T )
2∆x
∗ ∗
ρw (xj+1 ,t) 1

ρw (xj−1 ,t) 1


ρ(xj+1 ,t) − P ρ(xj+1 ,t) − ρ(xj−1 ,t) + P ρ(xj−1 ,t)

− ρ(xj , t) ,
2∆x
where fρ and fρw are as in (22). We also have to consider the CFL condition while
using the Lax-Friedrichs method. The eigenvalues are given in (23) and due to the
delay we can except the speed of information in the delayed case to be less or equal
than in the undelayed case.
Since the Lax-Friedrichs method is consistent, stable, monotone and l1 -
contractive, it is sufficient to provide a small enough time step. Now, we have
to look at the discretization. Let us start considering the method of lines. There,
we spatially discretize the equation which can be done in several ways, e.g. [17] or
[29]. We end up with a system of DDEs solved by the method of steps again. How-
ever, the method of steps uses ODE solvers [16]. Moreover, we have an influence
of the history which can not be ignored. Since the history function appears in the
ODEs that are solved within the method of steps, numerical difficulties may arise.
In the following, we introduce two examples to emphasize the impact of delayed
macroscopic models in the case of real-world applications.
4.2.1. Traffic lights. The first application we have in mind is a frequent traffic light
switching. We intend to compare the model (10) to the classical ARZ model (1).
The experimental setting is introduced in Figure 11, where at the end of the street,
i.e. at x = 1000, a traffic light is installed.
Initially, the density and the velocity are ρ0 (x) = 0.3, v0 = (x) = 0.4 for x ∈
[0, 1000]. The delay is fixed to T = 0.5s. The traffic light starts with a red phase
which is simulated by a boundary value of ρ(1000, t) = 1 and v(200, t) = 0. Then,
the traffic light switches 3 times from red to green and splits the simulated time
 TRAFFIC FLOW WITH TIME DELAY 283

ρ0 (x) = 0.3, v0 (x) = 0.4

0 1000

Figure 11. Traffic light scenario

horizon of 500s into 6 phases: the first 5 phases are 62.5s long and the last is 187.5s.
The discretization is ∆x = 5 and ∆t = 0.05. The numerical results in terms of ρ at
t = 500 are shown for the delayed and undelayed ARZ model in Figures 12 and 13.

500

450
0.8
400

350
0.7
300
Time

250 0.6

200
0.5
150

100 0.4

50
0.3
100 200 300 400 500 600 700 800 900 1000
Space

Figure 12. Density for the delayed model

500 0.9

450
0.8
400

350 0.7

300
0.6
Time

250

200 0.5

150
0.4
100

50 0.3

100 200 300 400 500 600 700 800 900 1000
Space

Figure 13. Density for the undelayed model

We see that the presence of the delay term leads to phenomena which can not be
observed in the classical ARZ model as for example the broader density plateaus
to resolve the stop-and-go behavior. The difference of the two models can be also
recognized for the speed v(1000, t), see Figure 14. The speed is zero in both models
when the traffic light is red. However, when the green phase starts, the speed in
284 MICHAEL BURGER, SIMONE GÖTTLICH AND THOMAS JUNG

the undelayed model increases with a steeper slope leading to a higher speed. Here,
the delayed model seems to be more realistic as sudden steep peaks of speed is not
what one would expect.

0.7
Undelayed ARZ
Delayed RSD
0.6

0.5

0.4
Speed

0.3

0.2

0.1

0
0 50 100 150 200 250 300 350 400 450 500
Time

Figure 14. Speed over time at the end of the road in the Traffic
Light Situation

4.2.2. Data fitting. As a second application, we analyze the performance of the

models regarding real-measured data. Therefore, we fit the classical (1) and delayed
(10) model to real-life data and compare the numerical results.
We start to investigate the fundamental diagram, i.e. the relation between traffic
density and traffic flow, for real data taken from the Minnesota Department of
Transportation. The data shows a reverse lambda shape, see Figure 15. At low
densities, there is a clear linear relation between flow and density. At medium
densities however, we get a wide spread of flows, making a functional relation hard
or even impossible to identify. We remark that in the classical ARZ model (1) the
artificial variable w can be interpreted as a way to get a whole class of functions to
cover this spread. The relation between density and flux gets more clear again for
high densities.
In a next step, we follow the approach in [14] to fit the classical and RSD model
to this data. We use data collected by the Regional Transportation Management
Center, a division of the Minnesota Department of Transportation (RTMC data),
in the same way as described in [14] and approximate the relation between den-
sity ρ and flow q by a function Q(ρ) that fits the fundamental diagram data in a
least squares sense. The function family we intend to optimize consists of three
parameters α, λ, p, cf. [14],
hp p p ρ
Q(ρ) = α 1 + (λp)2 + ( 1 + (λ(1 − p))2 − 1 + (λp)2 )
ρmax
r
ρ i
− 1 + λ2 ( − p)2
ρmax
and the resulting solution is plotted in Figure 15.
From Q(ρ), we can derive a data-fitted pressure function P (ρ) = U (0) − U (ρ),
where U (ρ) = Q(ρ)
ρ and U (0) = Q0 (0) are velocities. This way enables us to use the

0 http://data.dot.state.mn.us/datatools/
 TRAFFIC FLOW WITH TIME DELAY 285

2.5
data
function value
2

1.5
Flow cars/s

0.5

-0.5
-0.02 0 0.02 0.04 0.06 0.08 0.1
Density cars/m

Figure 15. Fundamental relation between density and flux and

fitted function

data in our models as follows: Plugging in the pressure function P (ρ) into model (7)
by exploiting the connection between the modeling parameter C in the microscopic
model (2) and P 0 we obtain

1
∂t = ∂x v(x, t)
ρ(x, t)
∂t v(x, t) = ρ(x, t − T )2 P 0 (ρ(x, t − T ))∂x v(x, t − T ). (27)

Now, we test the data-fitted model (27) with RTMC data to check the perfor-
mance. We follow again [14] and take a segment of street with no on- or off-ramps
and three measurement stations. The first and the third station deliver the bound-
ary values for our simulation, while the station in the middle is the reference point,
i.e. we will compare the simulation results with the measured data of the second
station. So we simulate the segment of road, use the real data as boundary values
and get a simulated traffic profile. This profile we compare at the point of the second
station with the real data. For a time interval of 600 seconds, i.e. 16:00-16:10 on a
workday, the results are plotted in Figure 16 for the fitted delayed and undelayed
ARZ model. The images represent the whole time sequence at the reference station
in the middle of the segment.
It seems that the undelayed ARZ model tends to avoid deviations while the
delayed model has stronger peaks. Beyond that, both models seem to fit the flow
better than the density, but this might be due to the specific example. For further
discussions on the data-fitted classical ARZ model, which might be useful for this
kind of experiment, we refer to [14] and [13].
Instead, let us look at the error between the real data and the simulated the
models. We therefore make use of the error defined in [14] which is

|ρmodel (x, t) − ρdata (x, t)| |vmodel (x, t) − vdata (x, t)|
E(x, t) = +
∆ρ ∆v
286 MICHAEL BURGER, SIMONE GÖTTLICH AND THOMAS JUNG

0.1
Reference
ARZ
0.08
RSD

Density
0.06

0.04

0.02
0 2 4 6 8 10 12 14
Time x 10
4

2
Reference
ARZ
1.5
RSD
Flow

0.5

0
0 2 4 6 8 10 12 14
Time 4
x 10

Figure 16. Comparison of the data-fitted delayed and undelayed

ARZ model to real data

for every position x and time t and normalization constants ∆ρ and ∆u. Due to
the measured data at only one reference point, the temporal average is given by
1 T
Z
E= E(x, t)dt.
T 0
The normalization factors ∆ρ and ∆v are the ranges of the density and speed
from the fundamental diagram. The errors we derive consider only the last 300s
of the simulations, although we simulate 600s in total. In the example above, the
error E for the undelayed model is 0.6404 and the error of the delayed model is
0.6034. To validate the impression that the delayed model performs better, we
study further examples. The data is again from workdays and the time is 16:00-
16:10. We investigated 10 randomly chosen weeks, i.e. 50 days. To compare the
errors of the simulations, we look at the following definition
Eundel − Edel
=4
(Eundel + Edel )2
where Eundel is the error E in the undelayed and Edel the error E in the delayed
simulation. This means, we rate the difference between the errors on the basis of
the mean of both errors. We evaluate this error for every day and observe a mixed
performance behavior, i.e. in some cases the undelayed model outperforms the
delayed model or vice versa. The extrema are that the undelayed model is in the
most extreme case better with  = −1.0203 while the delayed model is in the most
extreme case better than the undelayed model with  = 1.6160. So the undelayed
model outperforms the delayed model at best with 102.03% while the delayed model
outperforms the undelayed one with 161.6% at best. The mean over the 50 days is
 = 0.1102 which might indicate that the delayed model performs better.

Conclusion. Summarizing, we have derived a new type of macroscopic second

order traffic models including a delay from a microscopic follow-the-leader model in
three different ways. The proposed approaches are adapted from well-established
models known for the derivation of undelayed macroscopic traffic flow models, i.e.
reverse spatial discretization, coarse graining and Taylor expansion. We have also
 TRAFFIC FLOW WITH TIME DELAY 287

suggested different methods to numerically tackle the resulting equations. The

numerical results show that the delayed macroscopic equations lead to reasonable
results, in particular in comparison with the classical ARZ model.
Future work includes further theoretical investigations for the delayed PDEs that
have been already established for related problems, see for example [1, 2] and [31].
Furthermore, deeper investigations on suitable numerical methods for this kind of
delayed equations are necessary. From an application point of view, an extension
to traffic networks would be interesting.

REFERENCES

[1] A. Ashyralyev and D. Agirseven, Well-posedness of delay parabolic difference equations, Adv.
Difference Equ., 2014 (2014), 20pp.
[2] A. Ashyralyev and D. Agirseven, Bounded Solutions of nonlinear hyperbolic equations with
time delay, Electron. J. Differential Equations, 2018 (2018), Paper No. 21, 15 pp.
[3] A. Aw and M. Rascle, Resurrection of “second order” models of traffic flow, SIAM J. Appl.
Math., 60 (2000), 916–938.
[4] A. Aw, A. Klar, T. Materne and M. Rascle, Derivation of continuum traffic flow models from
microscopic follow-the-leader models, SIAM J. Appl. Math., 63 (2002), 259–278.
[5] A. Bellen and M. Zennaro, Numerical Methods for Delay Differential Equations, Oxford
University Press, Oxford, 2013.
[6] M. Brackstone and M. Mcdonald, Car-following: A historical review, Transportation Research
Part F: Traffic Psychology and Behaviour , 2 (1999), 181–196.
[7] R. E. Chandler, R. Herman and E. W. Montroll, Traffic dynamics: Studies in car following,
Operations Res., 6 (1958), 165–184.
[8] R. M. Corless, G. H. Gonnet, D. E. G. Hare, D. J. Jeffrey and D. E. Knuth, On the Lambert
W Function, Adv. Comput. Math, 5 (1996), 329–356.
[9] C. D’Apice and B. Piccoli, Vertex flow models for vehicular traffic on networks, Math. Models
Methods Appl. Sci., 18 (2008), 1299–1315.
[10] C.F. Daganzo, Requiem for second-order fluid approximations of traffic flow, Transportation
Research Part B: Methodological, 29 (1995), 277–286.
[11] N. Davoodi, A. R. Soheili and S. M. Hashemi, A macro-model for traffic flow with consider-
ation of driver’s reaction time and distance, Nonlinear Dynam., 83 (2016), 1621–1628.
[12] G. Emch, Coarse-graining in Liouville space and master equation, Helv. Phys. Acta, 37 (1964),
532–544.
[13] S. Fan and B. Seibold, Effect of the choice of stagnation density in data-fitted first- and
second-order traffic models, preprint, arXiv:1308.0393.
[14] S. Fan, M. Herty and B. Seibold, Comparative model accuracy of a data-fitted generalized
Aw-Rascle-Zhang model, Netw. Heterog. Media, 9 (2014), 239–268.
[15] M. Garavello and B. Piccoli, Traffic Flow on Networks, American Institute of Mathematical
Sciences (AIMS), Springfield, MO, 2006.
[16] S. Gottlieb and C.-W. Shu, Total variation diminishing Runge-Kutta schemes, Math. Comp.,
67 (1998), 73–85.
[17] A. Harten, High resolution schemes for hyperbolic conservation laws, J. Comput. Phys., 135
(1997), 259–278.
[18] D. Helbing, Verkehrsdynamik: Neue Physikalische Modellierungskonzepte, Springer Berlin
Heidelberg, 1997.
[19] S. P. Hoogendoorn and P. H. Bovy, State-of-the-art of vehicular traffic flow modelling, Pro-
ceedings of the Institution of Mechanical Engineers, Part I: Journal of Systems and Control
Engineering, 215 (2001), 283–303.
[20] W.-F. Jiang and Z. Wang, Developing an Aw-Rascle model of traffic flow, J. Engrg. Math.,
97 (2016), 135–146.
[21] E. Kometani and T. Sasaki, On the stability of traffic flow (Report-I), J. Op. Res. Japan, 2
(1958), 11–26.
[22] H. K. Lee, H.-W. Lee and D. Kim, Macroscopic traffic models from microscopic car-following
models, Physical Review E , 64 (2001), 056126.
[23] R. J. LeVeque, Finite Volume Methods for Hyperbolic Problems, 31st edition, Cambridge
University Press, 2002.
288 MICHAEL BURGER, SIMONE GÖTTLICH AND THOMAS JUNG

[24] R. J. LeVeque, Numerical Methods for Conservation Laws, 2nd edition, Birkhäuser Verlag,
Basel, 1992.
[25] F. Maghami Asl and A. Ulsoy, Analysis of a system of linear delay differential equations,
Journal of Dynamic Systems, Measurement, and Control, 125 (2003), 215–223.
[26] D. Ngoduy, Generalized macroscopic traffic model with time delay, Nonlinear Dynam., 77
(2014), 289–296.
[27] H. J. Payne, Models of freeway traffic and control, Mathematical Models of Public Systems,
Simulation Council Proceedings, (1971), 51–61.
[28] RTMC Data Set, Available from: http://data.dot.state.mn.us/datatools/.
[29] C.-W. Shu, TVB uniformly high-order schemes for conservation laws, Math. Comp., 49 (1987),
105–121.
[30] A. Tordeux, G. Costeseque, M. Herty and A. Seyfried, From traffic and pedestrian follow-
the-leader models with reaction time to first order convection-diffusion flow models, SIAM J.
Appl. Math., 78 (2018), 63–79.
[31] C. Travis and G. Webb, Existence and Stability for Partial Functional Differential Equations,
Transactions of the American Mathematical Society, 200 (1974), 395–418.
[32] M. Treiber and A. Kesting, Traffic Flow Dynamics, Springer, Heidelberg, 2013.
[33] G. B. Whitham, Linear and Nonlinear Waves, John Wiley & Sons, Inc., New York, 1999.

Received December 2017; revised December 2018.

E-mail address: [email protected]
E-mail address: [email protected]
E-mail address: [email protected]