Lund, undated, Cowiconsult 1985, Hart and Ponsness 1992). Most of the computer-
aided design methods that have been developed are proprietary and thus any
optimization that they profess to make is not open to inspection or modification. An
optimal design model that is open to inspection and modification is therefore
needed. The objective of this work is the development of such a model.
OPTIMIZATION IN DISTRICT HEATING SYSTEM DESIGN
Determination of pipe sizes is one of the major decisions that the designer of a
district heating system faces. Other critical decisions that must be made are heat
source, distribution media, distribution temperatures and load management strat-
egy. Many of these factors have been addressed by previous studies. The emphasis
of this work will be on determining pipe size.
Currently, pipe sizes are usually determined on the basis of simple criteria, such
as maximum pressure loss or maximum flow velocity. A number of investigators
have addressed the issue of pipe size determination, trying to improve on these
simple criteria. Aamot and Phetteplace (1976) presented a method that relies on
establishing the ratio between the heat losses and the pumping cost and then finds
the lowest cost pipe diameter by minimizing the sum of capital, heat loss and
pumping costs. Their work only addressed a single pipe segment and did not
include the effect of varying load over the yearly cycle. Szepe and Calm (1979)
presented a model for single pipe segments that neglected heat losses and time
varying loads, but used geometric programming theory to achieve additional
insight into their simplified problem. In later work, Phetteplace (1981) included the
effect of annual load variations, but only single pipe segments were addressed.
Frederiksen (1982) provided a detailed analysis of the heat generating station and
the consumer's systems, but simplified the transmission network to a single supply
and return pipe.
A number of investigators have addressed multiple pipe networks. Of course, a
great deal of work has been done for water distribution systems where the problem
is much simpler owing to the lack of heat losses and load variation with temperature
as well as mass flow rate. Marconcini and Neri (1979) described a model that
calculates the flow rate, pressure and temperature in networks of steam pipes. They
discussed the effect that pipe diameter has on operation, but did not offer any
methodology for selecting diameter values.
Stoner (1974) discussed models that are capable of modeling either steam or water
networks. Although the models do not determine optimum diameters, he gave a
procedure for achieving an optimal design by sensitivity analysis, but did not dis-
cuss how this process would be accomplished for networks of more than one pipe.
Zinger et al. (1976) described a computer program for calculating flows and
pressure levels in branched networks of hot water pipes. Their program accounts for
pressure drops in the consumers' equipment and throttling devices placed in the
network. Diameters are assumed to be known and they did not discuss how to
Morofsky and Verma (1979) developed a feasibility analysis and costing tool for
district energy systems, not intended for detailed design. They found the appropri-
ate pipe sizes by finding those that absorbed all of the available pressure difference.
They started the search for pipe size at the smallest available discrete pipe diameter
and then calculated pressure losses. If the pressure losses were more than the
available pressure difference, they increased the pipe size to the next discrete size
and repeated the calculation. They proceeded in this fashion until they reached a
discrete pipe diameter that did not result in pressure losses greater than the available