=Paper=
{{Paper
|id=None
|storemode=property
|title=Overview on Energy Data Reporting
|pdfUrl=https://ceur-ws.org/Vol-923/paper11.pdf
|volume=Vol-923
}}
==Overview on Energy Data Reporting==
https://ceur-ws.org/Vol-923/paper11.pdf
Overview on Energy Data Reporting
Tiago Cardoso and Paulo Carreira
{tiago.cardoso, paulo.carreira}@ist.utl.pt
Instituto Superior Técnico
Abstract. Energy Management Systems (EMSs) are tools that monitor
building energy consumption enabling more informed decisions towards
rational usage of energy to be made. Current EMS applications provide
high report diversity, allowing their users to gain significant insight to
understand how their building is performing. However, to the best of our
knowledge, there is no literature available on the different reports and vi-
sualizations types available. Report usability towards the user may vary
depending on the type of data displayed and how it is displayed. This
article tries to define and classify existing visualizations and reports. A
reference data model together with an architecture is presented, both
avoid having either lack of energy reports and also rigid reporting op-
tions, by enabling the generation of new reports.
1 Introduction
Energy Management Systems are tools able to monitor facility operations through
the gathering of building data related to environment and equipment operations.
Gathered data is used to generate reports that will help increase user aware-
ness on how building operations are performing. Through reports, EMS aims
at guaranteeing maximum operation efficiency, reducing energy usage, without
adversely affecting the building occupants comfort. Adequate data presentations
enable EMS to: Improve the level of building management, acting as centralized
management system; Provide a pattern of energy consumption, through the cap-
tured energy data, allowing unexpected consumptions to be identified; Identify
peak electrical demand, that might be responsible for additional costs.
In order to achieve its purpose, an EMS performs three fundamental opera-
tions, (i) gathering all energy related data , (ii) interpretation of collected data,
(iii) data presentation to the users in the form of reports [11]. Currently there
is not an agreed architecture describing an EMS, actually, most of the systems
follow a monolithic approach, unsuited with the idea of having an extensible
and flexible system. Allied to the fact that most of the current systems are ven-
dor specific developed as close projects, current EMS systems have a narrow
scope on reports, providing only a set of rigid reports. This inability cripples
current solutions ability to maintain users informed through the use of reports
that transmit raw data instead of knowledge.
The remaining of this document is organized as follows: Section 2 describes
the main energy data reporting dimensions. Section 3 overviews which oper-
ations are desirable to perform according to dimensions described in previous
section. Section 4 presents the report visualizations types. Section 5 describes
the required data sources necessary to support the report generation with focus
on a reference architecture, data model presented which design was oriented to-
wards flexibility and extensibility and an EMS implementation based on them.
Finally Section 6 will summarize this article and present the conclusions.
2 Energy Data Reporting
Reporting is the process of presenting collected data to the user. It is crucial
that information is transmitted effectively from the EMS to the user in order to
support his decisions [8, 13]. Report effectiveness and quality can be measured
in terms of: Interactivity with the user; Time window needed for user read and
understand the report; the Usability the report might have to the user and
energy manager tasks; Data exportation capability allowing data to be captured
and transformed freely by the user. In order to better understand report usage,
we must study which report dimensions are available:
Time dimension allows the system to identify, through the use of a time-stamp,
when the data was acquired.
Device and device type dimension is required to identify where the data is
being collected from. It is through the device and device type that location
dimensions and properties being measured are obtained.
Location dimension enables the user to known which locations are associated
to the acquired data.
Organizational data dimension allows the system to relate which locations
are associated to which department or individual.
Measured property dimension enables the user to select which properties/
measurement type should be presented in the report.
Energy tariff dimensions enable the system to correlated measurements from
energy meters to the tariff in practice at the time of the measurement.
3 Reporting Operations
3.1 Filtering and aggregation
Report information must be filtered, otherwise the user could be overwhelmed
with information presented towards him. Filtering is achieved through the ma-
nipulation of the data on report dimensions:
Time dimension allow users to select which time period will be analysed. EMSs
gather data since the moment they are turned on, so reports need a time
dimension filtering option otherwise report data could not be analysed in
adequate time. Report data can also be aggregated by time, allowing users to
compare energy consumption on different time periods (days, weeks, years).
Device and device type dimension is needed to identify where the data is
being collected from, allowing to select only the information gathered by
specific devices. This option enhances system capability to debug a problem
when the building is not performing accordingly to expectations [8].
Location is a dimension obtained from the devices installation location or moni-
toring range. Location can act either as a filtering and aggregation dimension
enabling the user to select which data devices should be presented.
Departmental organization is a dimension obtained from the relation be-
tween locations and departments. This dimension can be seen as a degen-
erated dimension from location, clustering locations through ”belongs to”
relationships and offering filtering and aggregation operations.
Measured property dimension enables the user to select which properties be-
ing monitored should be presented in the report. Some static data properties
such as production line schedules might be available to selection due to their
relevance to energy consumption.
3.2 Energy Profiling
Energy profiling consists on tracing an energy consumption pattern based on
past consumptions or behaviours. Consumption is highly related to time, usually
buildings have patterns that reflect daily scheduled operations or occupants rou-
tines. Nevertheless some significant deviations on energy consumption when an-
alyzing collected data shall be found. Finding the sources of those deviations and
how they influence energy consumption, will improve forecast effectiveness [10].
Those deviations can be found in data being extracted from equipment across the
building. Determine those sources is a process denominated data normalization
and in energy its main drivers are [2, 17]:
Weekdays influence building operations and building occupancy. Due to their
influence on building energy consumption, information must be correlated
according to the week day when the consumption occurred [16].
Temperature influence energy consumption due to its impact on HVAC sys-
tems, buildings major energy consumers. Systems use two set point tem-
peratures, one to start cooling the building and other warm it up. Data
normalization is usually performed according to cooling degrees and heating
degrees.
Lighting equipment are the second largest energy consumer in our buildings [5].
Some buildings try to conserve energy by setting lights according to time
schedules. In this case such event will be capture with consumption pattern
according to time. However more advanced systems perform according with
outside brightness. In those systems brightness levels need to be captured.
Humidity affects HVAC efficiency to cool and warm the buildings environment
having direct impact on consumption.
Building occupation has an impact in the usage of most systems on the build-
ing [14]. The number of people have an direct influence on the effectiveness
of HVAC system, because air needs to be recycled more often to reduce
the amount of CO2 and body heat production. Furthermore, the increase of
building occupation might relate to the use of additional electrical equipment
such as computers.
Production schedules might have a huge impact on energy consumption. In
some cases, due to the heavy machinery being used during those periods,
this factor overshadows other factors regarding energy consumption.
Equipment status gathers the mode on which systems are operating. For in-
stance, on HVAC systems lowering the set-point might bring energy savings,
in spite of affecting occupants comfort [8]. Some energy policies are defined
according to the settings on which equipment operate.
3.3 Energy Forecast
Energy forecasting is the process of predict energy consumption based on past
events. After data normalization process takes place, is possible to make accu-
rate consumptions forecast according to the expected conditions. Through the
forecast process energy peaks can be estimated and actions to avoid them can
be deployed, allowing to save associated costs to be saved.
3.4 Cross Operations
A complete solution must offer the possibility to cross the previously stated
operations. The raw energy consumption data does not provide any insight on
how our building is performing, we need comparative views. By comparing ex-
pected consumptions against the real consumption, performance indicators can
be found [7]. These will offer an better insight on how current energy policies are
affecting the energy consumption. Due to the fact that forecasting the energy
consumption results directly from assigning values to the measures used in the
profiling operation, forecasting shares profiling dimensions.
4 Report data visualization types
EMS solutions must generate visualizations according to user needs. While per-
forming his tasks, the user is interested to either analyse acquired consumption
data or observe data regarding current building state. According to user needs
and requirements, three different visualization types can be identified.
4.1 Historical Data Analysis
This approach is used when the user requires access to information regarding
gathered data. However, a single query might return hundreds of thousands of
points to be evaluated [12]. A data table presenting all gathering points could not
be evaluated in adequate time. Charts and dashboards allow EMSs to display
all several information data at once. EMS might use charts such as line charts or
stream charts to perform analysis over time [13]. Dashboard can be used showing
performance indicators, informing the user how energy was consumed during a
time period, relative to expected consumption or past consumptions.
4.2 Real Time Monitoring
Real Time Monitoring enables users to see information about energy usage,
equipment status and environment conditions on real time. Presenting the last
read values from the devices across the building captured provides the current
building state. However, metering data presents only values accumulated over
time. This information alone unable to tell the user how the building is perform-
ing without information about the previous reading, therefore, some data has to
be manipulated before being presented to the user.
4.3 Hybrid view
Hybrid view combine the best properties of both real time monitoring and his-
torical data analysis. The user is looking for a correlation between gathered
measures with the expected ones. The gathered values from environment sen-
sors can feed the forecasting model (discussed in 3.3) to provide a prevision on
real time consumption. A visual indication on how current consumption relates
to the expected consumption can be presented, providing an insight on how the
building is performing at a glimpse increasing system usability [13].
5 Energy Data Integration
5.1 Data Acquisition
Data gathering is the process responsible for acquiring data that might con-
cern energy consumption. Gathered data might come from three different source
types: energy meters, environment sensors, equipment status sensors [11]. Data
retrieved from meters enable the EMS to gain insight on building energy con-
sumption. Environment data will allow the EMS to correlate energy consump-
tion to monitored conditions. Collecting data regarding equipment status enables
correlations between consumption and operational status to be found.
Energy Metering Meters are cumulative reading devices, designed to measure
the amount of consumed energy [1]. Their installation location might dictate the
insight possible to be obtained.
In order to achieve a better resolution regarding energy consumption, some
solutions might install meters on locations that were already monitored by other
meters, this approach is named sub-metering [6]. Bigger insight on energy
consumption can be achieved following this approach, allowing unexpected con-
sumptions to be diagnosed and its root to be found.
Environment conditioning monitoring Energy consumption is highly re-
lated with external factors such as weather and building occupation, so they
must be monitored and acquired in order to be correlated with energy consump-
tion. Environment can be monitored through sensors, which provide an instant
reading over the measure being monitored. These relations between environ-
ment and energy consumption enable EMS to profile consumption and forecast
consumption based on the expected conditions.
Equipment status monitoring Equipment status data enables the EMS to
obtain information about which devices were operating and under which con-
ditions when a consumption pattern occurred. Therefore, collecting these data
enable the EMS to establish a relation between energy consumption and device
operation status. Usually energy policies are equipment settings definitions, such
as HVAC set points, collecting this data will allow to identify which measures
provided better savings.
With this knowledge the energy manager might decide the best path to
achieve a desirable condition, e.g. lowering building temperature using the less
amount of energy.
5.2 Data Quality
To ensure that reports are precise and accurate, faulty data must be identified
and its integration into the repository avoided. Faulty data can be identified as
incorrect, inconsistent or noisy. These type of data are determined through a
data quality process, in which several quality dimensions must be evaluated [3]:
Accuracy dimension relates to the precision of the data being captured. All
equipment have a known precision, however due to equipment malfunction
or lack of maintenance its precision can be affected, leading to incorrect data.
Timeliness dimension reflects data quality according to how long is it valid.
For instance, on real time monitoring data must be updated constantly.
Volatility dimension reflects data expiration date. Meters and sensors do not
store data. EMS needs to keep acquiring devices to obtain their value, oth-
erwise the time gap between reading intervals will have to be interpreted has
missing data and be estimated leading to incorrect results.
Consistency measure evaluates if the equipment readings are non-conflicting.
For instance, in a room with two temperature sensors, if they measurements
are inconsistent it is impossible to say what the real temperature was.
Completeness dimension evaluates if all required information is available. Col-
lecting data from a meter which monitoring range is unknown is an example
of incomplete data. In this example sub-metering operations could not be
tracked.
Duplicate dimension evaluates if gathered data was already acquired and added
to the system repository. Acquiring duplicate values can lead to misleading
information being presented on reports (energy peaks that didn’t occur).
Acquired data that does not meet those measures must be identified as faulty
data and the system must correct them. On EMS the mostly common issues
are related to network failure, duplicate data, misread information and human
error while inserting data manually. Having identification, EMSs have to either
discard them or by correct faulty data. On EMS where data quality analysis is
performed, error rate is expected to be in order of approximately 1-5%, although
error rate can go up to 30% where the process is non-existent [15]. After extracted
data is analysed and corrected, it might need transformations to fit system data
model. [9]. The entire process of data extraction, cleaning and transformation is
also known as the process of Extract Transform and Load (ETL) [12].
Data Mining Data mining is the process of data exploration aiming to find
patterns in data in order to ”extract” knowledge. [9] These patterns can aid
determine if data is being retrieved correctly, measuring data quality. If some re-
trieved data is outside the range of expectable value according to other measure-
ments, data can be seen as an ”outlier” and be discarded or being simply market
as having highly inaccurate. Furthermore, this process can be particularly use-
ful for retrieving consumption patterns, identifying what are the most frequent
causes of consumption peak responsible for high price tariffs and high billings.
After retrieving an energy consumption pattern, the energy manager might pre-
vent those peaks by managing the energy load. Is through Data Mining process
that energy consumptions variables are found, allowing energy consumption and
energy profiling to be made [4].
5.3 Centralized Model
The EMS architecture solution proposed is designed oriented towards flexibility
and extensibility. This solution aims to gather data from several heterogeneous
sources that might be added or removed from the system, while implementing a
platform to develop new reports.
Fig. 1. Proposed architecture.
The proposed solution is a three layered architecture presented in Figure 1.
Data Acquisition Layer oversees data acquisition from meters and sensors, In-
tegration Layer is responsible for collecting, storing and granting access to the
gathered data and Application Layer is responsible for the data visualization.
Real-Time Data Acquisition module allows the addition of new devices through
the implementation of a new Device Data Gathering Driver. These drivers are re-
sponsible for implementing the communication protocol, enabling them to collect
data from the devices they are designed to. They are connected to the real-time
data acquisition module through its exposed interface. For instance, if data from
deviceA is required, the solution only needs to know that the deviceA is reachable
through driverX. If a new device is created, to use it we simple develop a new
driver and add it to the solution by linking the device to the driver.Data will be
integrated in the solution data repository by the Data Integration module. This
module is also responsible for dealing with data quality issues, identifying data
errors and correcting them. Unexpected data, that is, data that is not coherent
with the energy profile or forecast, will be marked as being inaccurate for fur-
ther analysis. The frequency on which it occurs might trigger a warning about a
required maintenance check. A Service Layer will be placed over the remaining
solution providing access to gathered data through services exposure. This mod-
ule serves as an abstraction layer between the data model and the applications
demanding its data. Third party applications, represented as EMS Reports can
be deployed using Service Layer exposed services. The underneath architecture
serves as framework to the EMS applications deployed on top. They will be re-
sponsible for showing data and information to the user. This conception allows
new data visualizations to be added to the deployed solution, offering another
expansion capability without impacting the remaining solution.
5.4 Data Model
The data model is the conceptual design of a data warehouse where information
collected from multiple data sources is stored into a coherent repository [9].
Having all information stored in a DW, makes information available even when
is collected on different locations, enabling data analysis and data mining to be
performed at a single point. All collected data will be stored in it supporting
system reports generation. The schema model is represented in the Figure 2.
In this model there are three data gathered types needed to be modelled:
(i) meter data, (i) environment data and (iii) equipment maintenance. Each
measurement will be identified according to the time in which the measurement
was performed, the device that performed the reading and which measurement
(property) is being captured. Measurements performed by energy meters might
have a tariff model associated to it, enabling energy costs to be obtained from
a set of readings. Each measurement will have a data quality dimension associ-
ated, indicating if the measurement appears to be correct, doubtful according to
expected results, or has been programmatically estimated to overcome missing
data. Through the device dimension, we can access information regarding the
device measurements range, an important aspect once a single device might be
monitoring several locations. A location in this model can be seen as a tree, where
each node is a distinct location. This hierarchical view allows sub-metering to
be performed and tracked by the system. The tariff model relates to the meter
data fact table, through the TariffModel dimensional table which might present
several values.
Fig. 2. Conceptual view of the data model.
5.5 Prototype Deployment
The final prototype will be deployed on Tagus Park installations using Tagus
Park monitoring equipment and energy meters provided by QEnergia. QEnergia
energy meters will be the S-Energy Manager 1 controller that can be connected
to several devices collecting their readings. A driver will be implemented to col-
lect data from QEnergia devices, as well as other available sensors and meters
available at the laboratory. The real time data acquisition module will periodi-
cally retrieve information from the assigned drivers. Before being stored into the
system data model, collected data will be analysed and its quality evaluated.
Collected data must meet all quality measures (discussed in the Section 5.2), so
data must be compared against data retrieved from other devices to check its
veracity and only then can be stored into the system. Gathering data from sev-
eral sources allows more interesting data correlation to be found allowing data
analysis and energy forecast to be performed and evaluated more accurately.
1
http://www.saia-energy.com/14-0-Energiemanager.html
Static data referring to Tagus Park class rooms and other areas as well as
locations assigned to each department (organization data), will be loaded into
the system. The energy cost plan in use at Tagus Park will be modelled into the
system.
To test EMS extensibility, two applications will use the solution service layer.
One application will use real time monitoring visualization, testing application
capability to send events back to the applications connected the it. The second
application will implement an interface that will allow to set energy price tariffs,
and extract summarized reports regarding energy costs.
In the end the prototype development will result in a tool able to monitor the
building operations, helping reducing the energy costs associated to them. Notice
that although the final result will consist on an ordinary EMS application, this
will be able to be expanded. New drivers can be developed in order to support
new devices or even retrieve data from other sources such as available online
web services to gather weather data. Due to the developed data model, the
system will be able to support data from several types of devices relating energy,
building data, organizational data and even equipment operations. On top, new
visualizations can be developed and added to the existed solution, offering an
expansion possibility to add new reports, visualizations and mechanism that will
inform and aid both energy managers and building occupants.
6 Conclusion
In this paper we discussed the main features presented by current EMS systems,
as well as their limitations. Energy Managers are fully aware on the potential
offered by these systems towards energy savings. Nevertheless they are unaware
on the fact that most of these systems only offer a narrow view over the available
data. Most developed solutions are able to present a lot of data, but with little
knowledge over the collected data. The system must be able to provide, effec-
tively, an insight on energy consumption. A related issue relies on the fact that
few systems are able to perform data analysis and detect problems with data.
Faulty data must be handled by these systems, warning users about a potential
lack of accuracy when presented.
Through the presented proposal, we believe those problems can be over-
comed. Presented data model enables gathered data from multiple sources to be
stored under the same repository, enabling the system to collect data regarding
energy consumption, equipment condition and environment status. To enhance
system’s flexibility, new data sources can be added to the system through the
implementation of data gathering drivers. Data quality issues can be found with
data analysis and a quality measurement can be presented on the report. The
solution service layer allows third-party software to use the underneath solution
as a framework, enabling it to retrieve gathered data by the solution. New data
reports can be added easily with minimal impact to the remaining solution. Pro-
totype evaluation will be performed by system deployment, testing the solution
capability to add new data sources and expandability. We expect to assist dis-
semination of EMS systems in buildings, through the developing of a state of the
art system that will bring together features that will help to monitor and pre-
dict the cost of energy consumption in a single tool, enabling energy managers
to make informed decisions to save energy.
References
1. Y. Agarwal, T. Weng, and R.K. Gupta. The energy dashboard: improving the
visibility of energy consumption at a campus-wide scale. Embedded Network Sensor
Systems, SenSys, pages 55–60, 2009.
2. Ammar Ahmed, Joern Ploennigs, Karsten Menzel, and Brian Cahill. Multi-
dimensional building performance data management for continuous commissioning.
Advanced Engineering Informatics, 24(4):466 – 475, 2010.
3. C. Batini and M. Scannapieca. Data quality: Concepts, methodologies and tech-
niques. Springer-Verlag New York Inc, 2006.
4. B.L. Capehart and T. Middelkoop. Handbook of Web Based Energy Information
and Control Systems. Fairmont Pr, 2011.
5. F.M. Courtel, D. Duguay, Y. Abu-Lebdeh, and I.J. Davidson. The potential for
demand responsive lighting in non-daylit offices. Journal of power sources, 202:269–
275, 2012.
6. S. Drenker and A. Kader. Nonintrusive monitoring of electric loads. Computer
Applications in Power, IEEE, 12(4):47–51, 1999.
7. S. Few and Safari Books Online. Information dashboard design. O’Reilly, 2006.
8. A.E. Guntermann. Are energy management systems cost effective? Industry Ap-
plications, IEEE Transactions on, pages 616–625, 1982.
9. H. Jiawei and M. Kamber. Data mining: concepts and techniques. San Francisco,
CA, itd: Morgan Kaufmann, 5, 2001.
10. Ricardo Sosa Liliana Argotte, Manuel Meja-Lavalle. Business intelligence and
energy markets: a survey. Intelligent System Applications to Power Systems, page
pages 74?84, 2009.
11. X. Ma, R. Cui, Y. Sun, C. Peng, and Z. Wu. Supervisory and energy management
system of large public buildings. In Mechatronics and Automation (ICMA), 2010
International Conference on, pages 928–933. IEEE, 2010.
12. D.Y.R. Nagesh, J.V.V. Krishna, and S.S. Tulasiram. A real-time architecture for
smart energy management. In Innovative Smart Grid Technologies (ISGT), 2010,
pages 1–4. Ieee, 2010.
13. D.Y.R. Nagesh, A. Sowjanya, and S.S. TulasiRam. Real time decision support
for energy management. In Proceedings of the World Congress on Engineering,
volume 1, pages 16–22, 2008.
14. Luis Prez-Lombard, Jos Ortiz, and Christine Pout. A review on buildings energy
consumption information. Energy and Buildings, 40(3):394 – 398, 2008.
15. T.C. Redman. The impact of poor data quality on the typical enterprise. Com-
munications of the ACM, 41(2):79–82, 1998.
16. S. Rojchaya and M. Konghirun. Development of energy management and
warning system for resident: An energy saving solution. In Electrical Engi-
neering/Electronics, Computer, Telecommunications and Information Technology,
2009. ECTI-CON 2009. 6th International Conference on, volume 1, pages 426–429.
IEEE, 2009.
17. J.C. Van Gorp. Enterprising energy management. Power and Energy Magazine,
IEEE, 2(1):59–63, 2004.