|
Load Management
is usually necessitated by cost factors, or by
demand outstripping utility company’s supply capabilities. There are two
possible situations. First, slowly increasing power demand brings the load to a
preset threshold, and second, load has to be brought down from some high value
to a preset threshold as quickly as possible. After that, in both cases, all
that is left to do is to maintain load as near that threshold as possible for as
long as possible (or necessary). The usual approach is to disconnect customers
in the order assigned by the company, and keep them disconnected as long as
needed. In cases when disconnect times are long the company may opt for load
rotation, that is reconnecting disconnected customers after some elapsed
time while disconnecting others.
The purpose of load management systems is to automate these
procedures, and to run them as swiftly and efficiently as permitted by local
circumstances. While maintaining load below a set limit is not difficult to do,
care must be taken that the threshold is straddled as closely as possible. As
for the two cases mentioned above, the second one is harder to handle. It is
critical in that if load can not be brought down fast enough the company pays
penalties. On the other hand, if initial shed cycle overshoots the desired high
limit by too large of an amount, the company loses revenue. What follows is a
discussion of some of these issues, and how NTMC systems attempt to
solve them.
It must be understood that even though these issues are
discussed separately as if they happen one at a time, they actually occur
concurrently. Any observed demand curve is always the result of many competing
factors, including actual demand at the time, current management cycle activity,
previous management cycle activity, and states of the individual load points at
the time.
One-Way Load Management
Load management is called one-way when devices used to control
loads are unable to report back to the controller whether the respective loads
are running at the time of management. Most switches in the market today are one
way. This has important implications when load has to be shed quickly from a
high value to a preset limit (second case of previous section).
 Ideally,
if every switch was connected to an active load, total load could be dropped by
the desired amount in a single management cycle, between two regularly scheduled
polls for demand data. This situation is shown in Fig. 1. Vertical dashed lines
are poll times, the orange line represents the threshold, and the blue line the
measured load.
 In
reality, what happens is that some percentage of load points are connected to
customers not consuming any power at that time. Since the one-way load
management system has no way of knowing that while shedding load, we get the
situation depicted in Fig. 2. Clearly, after the first management cycle it
becomes clear that not enough load was shed, so another shed cycle is executed.
This can go on for several poll cycles resulting in very slow system response
times. Since load stays above the high threshold longer, the utility company
ends up paying penalty fees which directly affect their bottom line.
NTMC attempts to remedy this situation by providing a
configuration parameter called Load Rating Bias
Factor. An experienced operator sets this number to the (estimated)
percentage of loads actually running at the time. This makes the load management
system rescale load point ratings by the same factor. As a consequence more load
gets shed during the first cycle.
While this approach works in improving system response time,
it is imprecise. Its accuracy relies on operator’s powers of observation.
Furthermore, situation changes from day to day. Therefore, a better solution is
needed.
Two-Way Load ManagementTM
Load management is called two-way when devices used to control loads are able to
report back to the controller whether the respective loads are running at the
time of management. NTMC systems do have this capability.
When a two-way load management system enters into a shed
cycle, each outgoing control code is not only confirmed, but also followed by,
in the same confirmation message, the kW number of the target load at the
time of the disconnect. Any customers not
running at the time of the command are not switched off, additional loads are immediately added to the shed
list, and the system continues to turn loads off as long as it takes to bring
total load down to the desired level in one management cycle (situation depicted
in Fig. 1).
In addition, by using Telescada NeXGen RTUs, NTMC can, in a single
poll, get not only the latest energy consumption numbers (latest dial readings),
but also instantaneous demand. This can be used to pre-poll load points just
before shed cycle is to begin. One poll only is necessary to know who is running
just before load reduction is to begin. This further increases load shed
efficiency.
Active Cycle Demand Correction
When a system polls for demand data after an active management cycle the
observed demand reading is too high. This is because the reading is the average
load between the current and the previous polling cycles. Since load was shed
during that time, and since the shed process is not an instantaneous event, the
observed demand reading is biased upwards. This may result in additional load
shedding, and thus in an unwanted system  overreaction
during the second management cycle, as shown in Fig. 3. To recover, the system
may attempt to release some load which may or may not work depending on whether
total load fell below release threshold, and whether individual load points are
self starting when power is restored. In any case all this results in loss of
revenue for the company selling electricity.
To compensate for this NTMC systems perform Active
Cycle Demand Correction. Demand reading observed after every active
management cycle is corrected downwards by an amount calculated from several
factors including amount of load shed during previous cycle, amount of time it
took to accomplish that, and the system polling period. This eliminates the
overshoot depicted in Fig. 3, returning the situation to the one sketched in
Fig. 1 thereby increasing company revenue.
Average Demand
Extrapolation
As already mentioned it is important to  straddle
load threshold as closely as possible. One can, for example, shed just enough
load to bring it below the threshold when the latter is exceeded. During periods
of increasing demand this results in load profiles such as the one depicted in
Fig. 4 where it is clear that, on the average, one is holding total load just
above the threshold. This may result in penalty fees which directly affect the
bottom line of the company. For this reason, when demand is rising, NTMC
systems employ the Average Demand Extrapolation algorithm. The latter
calculates how much load to shed in order to keep the future average load below
the high threshold, resulting in situations more like the one shown in Fig. 5.
More
on how LMEX, the NTMC
load management module, implements above can be found here.
| 
