What can we offer?
Crystalball is a Building Energy Management System that runs either in the cloud or locally on off-the-shelf hardware. It can be deployed in buildings of any size to manage equipment such as inverters for Solar Panels, Heat Pumps, Electric Vehicles, Stationary Batteries and large electricity consumers within the building.
What does it do?
Gives oversight of energy usage across the building
Controls the interaction of key equipment (e.g. Heat Pump running, Electric Vehicle charging, etc.)
Automatically optimises the energy cost of the building based on demand constraints and forecasted sun irradiance (PV production if available and passive solar gains for the building) and forecasted ambient temperature and electricity tariffs.More Info
What is unique about it?
Crystalball optimises energy cost based on local weather forecast together with user behaviour and usage preferences. It has a physical model for predicting room temperature and controlling heating, hot water usage and hot water boiler loading, the same applies also with EVs and stationary batteries.More Info
What is the current status?
The Crystalball product is already being deployed for multi-apartment buildings as well as for single homes. It has already been integrated with a wide variety of equipment covering many interface protocols.
Feel free to sign in to get a preview of Crystalball capabilities.Sign in
How could it be incorporated within a project?
Crystalball can be used as a stand-alone product. Additionally it could be used within a larger project covering multiple buildings and/or communities and integrated with other software such as CSEM’s Maestro.
Recording and displaying building information
Crystalball gives a building owner an overview of their energy consumption. The building could be a single home dwelling, could be a multi-apartment building or could be a commercial premise.
The typical overview dashboard available to the building owner is shown below and is the case for a single home dwelling.
At its simplest level of usage, Crystalball lets the building owner see the status of their equipment and parameters such as the temperature of their room and hot water. The building owner sees the overall level of current electricity consumption and also has a record of the amount exported.
The choice of data sampling rate is important and needs to be carefully selected to get the right balance between system needs and costs of data storage. Crystalball allows varying data sampling rates to be used and we have strong experience in understanding the needs based on the characteristics and influence on the system. For example, heating systems have a high thermal inertia and need only relatively low data sampling rates.
Automatically controlling building equipment
Equipment such as inverters for Solar Panels, Heat Pumps, Electric Vehicles are usually supplied with their own controller acting only on the individual piece of equipment. It is left to the building owner to decide how to use the equipment in terms of interaction with each other.
Without any central, overall controller, this leaves the building owner with 2, poor, usage options:
- Set each equipment to run on its own automated program in a hope that the equipment operates in an acceptable manner.
- Regularly make manual interventions in switching equipment on and off to ensure that basic conditions are met (e.g. car adequately charged).
To address these poor options, there are several controllers on the market that act centrally. These offer a better strategy:
- Maximise self-consumption. These controllers sometimes have the benefit of a single user interface to most covered equipment.
Crystalball also acts as a “central brain” that controls all equipment in an integrated way. It also provides a much more powerful optimisation strategy:
- Automatic optimisation of energy cost based on forecasting and user needs and constraints.
The user has the ability to input their needs into a single system interface as illustrated below. This particular screen is for Electric Vehicle usage. There are 2 cars from different manufacturers in this particular household and input of planned trips in a single interface allows each car to be charged in an optimum way to ensure sufficient charge for each journey with an adequate reserve margin embedded.
Automatic optimisation of energy cost
In order to take decision to minimize the cost operation, and consider forecast in the decisions, a control approach implementing Model Predictive Control (MPC) has been chosen. MPC uses a model of the system to predict the cost of operation for any operating strategy: it uses this model to formulate an optimization problem to minimize the cost of operation over a pre-defined period of time (typically one to several days ahead). Currently, we operate Crystalball on a 3-day ahead basis.
To make a plan for the future, the MPC algorithm needs to make a forecast of the external factors affecting the system. The key influential parameters to predict are:
- Solar irradiance
- Outside temperature
- Electric consumption
- Hot water consumption
- Electric car usage
Temperature and solar irradiance are collected from a weather forecast provider.
Due to the criticality of the PV production, a dedicated forecasting code has been implemented within the Crystalball system.
Simpler approaches have been used for the other quantities.
Due to its computational cost (completing calculations takes in the order of minutes), the MPC algorithm cannot be executed very fast. However, large variations, in particular of the PV production can occur in a matter of seconds. As a consequence, a fast control loop is required to absorb these fast variations. This has prompted the development of a hierarchical control architecture. A local and relatively simpler controller can adjust the inputs to the different subsystems, while the MPC calculates a new operation plan every 15 minutes. This plan is a baseline around which the local controller can move in order to adapt to real-time conditions measured at the site.
An example of how optimisation of energy cost can bring benefits over simple self-consumption optimisers is set out below.
Saturday morning, high PV power being produced (higher than the minimum charge rate of EV).
Current commercial controllers optimising self-consumption:
Will always charge the car to ensure maximum self-consumption.
Will use cost functions to consider the following option and execute it if cheaper:
Export what PV production there is now at a high tariff even though it could be used and then charge the EV from 1300 when the export tariff is lower.
Clearly there is a potential for reducing costs here depending on: PV produced, forecast of PV to be produced, amount of extra electricity needed to charge car to the needed level.
A similar example can be given with the stationary battery. Scenario in winter months when not enough energy can be stored to cover all the energy needs of the night and the next morning. A self-consumption optimizer would use all power during the night while Crystalball will limit the power usage from the battery during low tariff such as to have remaining energy the next day during high tariff.
Crystalball has 3 key features that ensure high quality optimisations:
- The ability to incorporate full physical models of the building (losses, thermal inertia etc.)
- It is easy to add additional cost functions to the model (e.g. flexible tariff schemes)
- It is also possible to add physical models of other equipment based on the existing library of CSEM