Cassandra platform
Christos Diou
Kyriakos Chatzidimitriou
Postdoctoral research associates
CERTH/ITI
A multivariate platform for assessing the impact of
strategic decisions in electric power systems
1st CASSANDRA
Workshop
Coventry, 11 September, 2013
Outline
• Library-based scenarios in the alpha platform version
– Use pre-existing library components
• Measurement-based scenarios
– Model training to build models automatically
• Response models
– Consumer response to different incentives
• Consumer Social Network analysis
– Grouping of small-scale consumers into Consumer Social Networks
• Development status and next steps
About the platform
• Currently in alpha version, development is highly active
– Some functionality has not been integrated yet
• Open source platform, publicly available through GitHub
– http://github.com/cassandra-project
– Apache license 2.0
LIBRARY-BASED SCENARIOS
Cassandra platform
Login screen
Main screen
Main panel
LibrariesProjects and
entities
List of projects
List of projects
User library
User library
Cassandra library
Cassandra library
Appliances in Cassandra library
Appliances in Cassandra library
Creating a new project
Creating a new project
Adding a new scenario to our project
Installations can be added by drag’n’drop
from the user library
Persons, Activities, Appliances
Persons, Activities, Appliances
Activity models
Activity models (duration)
Activity models (start time)
Simulation parameters
Simulation parameters
Submit runs
Submit runs
Scenario 1: A mall
• What if …
– Roof-Top-Units (RTUs) are shut down 1 hour prior to mall closing time
taking advantage of thermal inertia in the sales area?
– Gradually start A/C units from 08:00-09:00?
– Set points from 21 to 24 degrees Celsius?
– Shut down all office A/Cs after 22:00 with manual override?
– Set the minimum fresh air from 20% to 5%?
– Change escalators from always on to escalators with motion sensors?
Baseline and test-case scenarios - Change a
set point
Baseline and test-case scenarios - Change
duration
Comparisons
SCENARIOS WITH MODEL TRAINING
Cassandra platform
Training module
Import installation measurements
Import installation measurements
Next step: Disaggregation
Training consumer activity models
Training consumer activity models
Export the models to the platform libraries
Models are visible in the user library
CONSUMER RESPONSE
Cassandra platform
Response models
Modify pricing scheme
Estimate consumer response
Modify pricing scheme
Models are posted to the platform
Models are posted to the platform
Example 1: Response from … to
Example 2: Response from … to
Complete change of habits
After 100 Monte Carlo runs of baseline and
response scenarios
…
Comparison
of runs
CONSUMER SOCIAL NETWORKS
(CSN)
Cassandra platform
CSNs
• Groups of similar consumers
– Multiple similarity criteria
• CSNs have potential:
– Increased market power of aggregated small-scale consumers
– Coordination of consumption activities at group level
– Targeted incentives at group level
CSN module
• CSN module: A tool for identifying links and grouping of consumers in
a meaningful way
– Existing social network connections
– Explicit attributes (e.g. working, non-working person, locality in the grid
topology)
– Implicit attributes (e.g. consumption similarity, peak similarity, behavioural
similarity)
• Early version implemented for experimentation
• Next version:
– More similarity criteria
– Estimation of group response to incentives
– GUI integration
The main graphical interface
of the CSN module
The network can be created
based on Installation Type,
Person Type, Average, Peak,
Similar, or Dissimilar
Consumption, etc.
CSNetwork based on person
type. Persons of the same
type are linked.
A network based on similar
consumption
Select a clustering algorithm
Adjust the clustering
parameters
Clusters appear in different
colors
The different consumer
groups appear
In summary, with Cassandra you can
• Simulate working scenarios/pilots
• Benchmark different energy efficiency solutions/products in simulation
before testing them in real-life
• Create detailed models that describe consumer behaviour
• Identify and evaluate optimal consumption schedules
• Estimate consumer response to a range of incentives
– Pricing schemes
– Consumer awareness
– Environmental impact
• Identify meaningful consumer groups and benchmark the application
of targeted incentives
So, what’s next?
• Beta release is expected before the end of 2013
• Further development and integration of response and CSN modules
• Integration of external modules with the platform
– thermal controllers, lighting models
• Evaluation of Cassandra in our three project pilot cases
• Evaluation of Cassandra in a limited number of NoI pilots (external
evaluation)
Thank you!
Questions?