Professional Documents
Culture Documents
Christian Johansson
Noda Intelligent Systems AB
Campus Grsvik 5
37175 Karlskrona, Sweden
E-mail: cj@noda.se
KEYWORDS
Industrial control, Interactive simulation, System dynamics, Energy, Interacting distribution and production.
ABSTRACT
Simulation is commonly used within the domain of district heating, both as a strategical decision support tool
and as an operational optimisation tool. Traditionally
such simulation work is done by separating the distribution models from the production models, thus avoiding
the intricacies found in combining these models. This
separation, however, invariably leads to less than satisfying results in a number of instances. To alleviate these
problems we have worked to develop a simulation tool
which combines the physical and financial dynamics
throughout the entire process of production, distribution
and consumption within a district heating system.
INTRODUCTION
Dynamic simulation of a district heating system can easily become overwhelmingly complex, due to the large
number of interacting components. A district heating
system may contain several production plants and literally thousands of buildings connected through a network
of hundreds of kilometers of pipes. The topology of the
network is usually described by a complex geometry including loops and numerous branches, and is geographically spread over a wide area resulting in large transport
times. Similarly the production plants and the consumer
endpoints are in themselves very complex entities.
All in all there is a large number of parameters that need
to be taken into continuous consideration. However, in
practice most of these parameters can not be determined
precisely, e.g. because they are not described in the construction plans or maybe because they are simply too difficult, or even impossible, to measure. In the ideal case, a
detailed computer model is available, which is validated
by comparison with measurements of high resolution in
time. Using measured values from a district heating system in the Swedish town of Gvle, we perform such a
comparison using our design and implementation of a
simulation tool. This simulation tool and its fundamental
design and capabilities are described in this paper.
Fredrik Wernstedt
Department of Systems and Software Engineering
Blekinge Institute of Technology
372 25 Ronneby, Sweden
E-mail: fwe@bth.se
hot tap water. The third part is the outdoor model, which
simulates the influence from the ambient environment.
The building model is composed of three components, a
energy demand component, a heating component and a
flow controller component. The energy demand component describes the building energy demand to maintain a
given indoor temperature at a specific outdoor temperature. The resistance, R, of the building is given by:
1
(1)
R=
U ext * Aext + e * n50 * Vair * air * C air
Valve
Control valve
(2)
T xi =
1
1
1+
TR x TC x
Q i + outi
TR x
T
+
x ( i 1)
TC x
(3)
(4)
CW
Pump
DH District heating
Measurement
HW Hot water
Cold water
0 = C*R
(5)
1982),
Tr (t ) Tu
= e 0
Tr ,0 Tu
(6)
where k =
Pv
UA + PL
1 k
(7)
(8)
21
50%
90%
20
19
18
17
0
12 16 20 24 28 32 36
40 44 48
Tim e [h]
X = 1 exp (u )du
(9)
Kitchen
Wash
/ 20
(11)
outd
+ Tr + Td
(12)
/5
(13)
Production
The purpose of production modelling is usually to optimize the operation in regards to production costs such as
fuel, wages, taxes and fees, possible network expansions
and so forth. Simulation modelling is a powerful tool
used by most district heating operators in order to plan
daily production as well as long term financial planning.
In addition, we also develop continuous production limitations for the distribution and consumption simulations.
Together, the production and consumption models form
the input for the distribution model.
There are several different types of production units and
they all have different constraints and operational limitations which influence their respective abilities to produce
energy. It is important to propagate these limitations into
the distribution and consumption calculations, since
these will be prone to error otherwise. Likewise, the production models will greatly benefit from continuous response from the distribution and consumption models
(Rossing and Johnsson 2005). This is one obvious case
were the combined production/distribution/consumption
model clearly shows its advantages over using separated
models.
Dhemos incorporates the use of pumps into the production model, and together with the output temperature
from the heating unit, their behavior form the practical
input data into the distribution model..
ing time t t , t is based on a series of rather fundamental relations on the form of:
0,02
Probabillities
1999); To = Tm + Tv * e (( i * s 4 ) mod 24 12 )
Bath
0,025
C ( t ) = t CU ( t )
0,015
0,01
i =1
(14)
0,005
0
0
12
16
20
24
Time
Distribution
v
perature front, vT . F is defined as, F = T .
v
(15)
c
4
1 + p p, p
+ 1
w c p , w di , p di , p
where is density, c is specific heat capacity,
(16)
is pipe
wall thickness and d is the inner pipe diameter. For example, if a DN 50 pipe is used the temperature front will
have propagated 83 metres as the water flow has moved
100 metres.
It is the temperature distribution that limits the computational performance during practical implementation of
the simulation models. This is due to the fact that pipes
have to be divided into a number of sections in order to
retain the energy status throughout the pipe. Dhemos
uses variable section sizes which vary dynamically
throughout the simulation in relation to the propagation
speed of the temperature front and thus also in relation
with the water flow. This means a somewhat higher
computational cost as opposed to using a fixed section
size, but it also means a higher degree of accuracy in the
simulation results.
SIMULATION OF GVLE DH SYSTEM
3000,00
F lo w [k g /s ]
F=
2800,00
2600,00
2400,00
2200,00
2000,00
0
12
15
18
21
24
Time [h]