56 / 80
4
to the network without any benefit. When considering Ethernet for
use in substations or other utility sites, the hardware’s resistance to
EMI must be considered. EMI can adversely affect transmission of
data, especially over copper cables and within the units themselves.
Hardware manufacturer
It is also important to test functionality of a hardware manufacturer
before purchasing. Some Ethernet standards are still quite vague in
their exact requirements, or are very general in their requirements,
so a manufacturer could correctly and legally offer compliance to a
standard that turns out not to be the level of compliance required. A
current and relevant example of this is the IEC61850 [2] standard for
electrical substations. IEC61850 [2] is not a hard and fast standard, but
a collection of standards, best practices and suggestions (regarding
both the physical and logical aspects) for running a critical network
for substation automation. For this reason, some manufacturers will
state their hardware is IEC61850 [2] compliant when in fact it barely
complies. From a technical and legal standpoint, the statement of
compliance is correct and true. This leads to customers purchasing
hardware that is not going to meet their requirements, and thus huge
losses of capital time and investment.
Avoid vendor lock
Another important point when considering which manufacturer to use
is to avoid becoming vendor locked where possible. For instance, if
you select a certain manufacturer that offers a proprietary form of
redundancy you will then be obligated to use that same vendor for
any expansions or replacements in the future (unless that vendor
offers backwards compatibility with open standards). Switching to a
new vendor at that stage would require a lot of work and a possible
complete redesign of the network to cater for a different redundancy
mechanism. Using an open redundancy standard instead means that
in future you simply need any manufacturer that complies with the
original open standard. For instance, a power supplier could spend
a few years sequentially upgrading its entire control network across
the entire country. If a few years into this upgrade a new technology
becomes available that is not supported by the current hardware
manufacturer, and the company wishes/needs to use this functionality,
all the existing hardware would need to be replaced, even though the
majority of it is still new. Using open standards, the option would exist
to replace only the hardware on those sections of the grid that require
the new functionality, whilst interfacing with the existing hardware
where possible.
Hardware selection must be considered throughout the planning
phase. Hardware should be selected once the network is planned
and the requirements are clearly stated, otherwise the hardware
selection could limit your network design possibilities. One must
consider not only the present, but also future plans for the network
and the expansion thereof. If this is not properly considered in the
beginning, expansion at a later date could lead to wasted time, effort
and investment. For instance, if there are no spare ports, expansion
will require additional switches to up the port count on the existing
network. If the network is being expanded to add a single IP camera
at each substation, with no ports available one would need to consider
buying a new switch per substation, just to add a single connection
point. Similarly, if the IP structures have not been properly planned
and cannot cater for the upgrade, changing these would take a huge
effort and possibly include downtime (i.e. loss of production).
Topologies of the network
The next consideration is the topologies of the network, both the
physical and the logical. It is important to note that many of these
points would not be done in a linear fashion. For instance, topologies
will be affected by the manufacturer chosen and the functionality it has
available and will in turn affect things such as routing and redundancy.
So all of these points will need to be considered simultaneously to
provide the best possible overall solution. The physical topology of the
network refers, quite obviously, to the physical layout of the switches
54
ENERGY EFFICIENCY MADE SIMPLE 2015