Now that you have determined the set of EnPIs that strikes the right balance between
your business objectives and your budget, the next step is to create a measurement road
map and to begin connecting existing and new measuring devices to your energy
management system to track, report, and forecast EnPIs. It is best to start at the devices
themselves. The devices are often transmitters sending either physical 4–20 mA analog
signals or register values via a connected network. It is important to keep in mind that
these transmitters are often in place to measure instantaneous flows used in control and
SCADA systems. When trying to measure energy or work, however, it is more
appropriate to capture totalized flow rather than instantaneous flow measurements.
Many devices will provide both instantaneous and totalized values, but often at a price
premium. Totalizing meters (e.g., utility gas, water, and electric meters) are the most
accurate way to aggregate energy usage/consumption. Such meters typically provide
pulse outputs that increment with the delivery or consumption of a fixed amount of
energy. The next best option is to totalize energy and production flows in programmable
logic controllers (PLCs) and distributed control systems (DCSs). However, the farther up
the data chain from the source meter, the larger the error that can be introduced into the
energy integration calculation.
Once you determine the “what” and the “how” of data collection, you need to
consider the frequency of data collection. As mentioned at the outset, monthly data is not
granular enough to provide actionable information about what is happening in your
operations each week, day, or hour. Many plants find that hourly or 15 min data
collection provides the best balance between data storage cost, system performance,
and granularity of measurements to really understand how the EnPIs are tracking through
each week, day, or hour. Keep in mind that in order to properly control a process, many
control systems will scan and update real-time measurement as frequently as in
millisecond intervals. For EnPIs, it is not necessary to capture totalized flow at this
frequency. EnPIs provide good results at lower data capture rates. We find that most
plants choose the 15 min data capture interval as it aligns with the 15 min demand interval
period that most electric utilities use to calculate demand charges.
Once the meters and measurement instruments are connected to networks, control
systems, data loggers, and/or data historians, many energy management systems (EMSs)
can capture and reuse these data to record, track, report, and forecast EnPIs. It is also
important to make sure that the energy management system you select and use to track
EnPIs provides a means to capture manual data. Fuel sources such as oil and coal are
typically delivered in bulk batches and rarely have automated metering systems. In
addition, you may still have a few production processes that are tracked on clipboards.
Do not ignore these valuable data just because they exist in a manual format. Embrace
them and use them to the extent you can to increase the completeness of energy
measurement and balance calculations.
Finally, a word about the importance of creating “virtual meters.” Meters and
instruments at the “top” or supply side of a given utility will often capture and measure
total energy supplied to the plant or system. These high-level meters include energy used
in loads (both metered and unmetered), energy lost in distribution systems, and energy
DATA COLLECTION 353
lost in system and machine inefficiencies. Rarely will plants have every energy system
metered to every source and load. Therefore, “bottom-up” EnPIs and “top-down”
EnPIs will almost never reconcile perfectly. It is a good practice to have your
energy management system capture the losses and unmetered loads in a “virtual meter.”
Virtual meters are created in an EMS by mathematically combining physical meters.
Virtual meters provide useful insights into whether to invest in more meters and/or
address efficiency losses in systems. Virtual meters also act as good proxies to identify
failed instruments.