132 lines
4 KiB
Markdown
132 lines
4 KiB
Markdown
# Smart grid
|
|
|
|
## Not smart grid
|
|
|
|
- Electric grid, production, delivery and consumption has to occur
|
|
instantaneously and in perfect balance
|
|
- Reason for not smart:
|
|
- Frequency matches generation and demand
|
|
- Voltage is generator and transformer
|
|
- Current is the upper limit of devices and have to provide spare capacity
|
|
|
|
## Smart grid
|
|
|
|
- Features:
|
|
- Cost efficiency
|
|
- Economically efficient, sustainable
|
|
- Quality
|
|
- Security
|
|
- Safety
|
|
- How to
|
|
- Integrate the power infrastructure with Information and Communication
|
|
Infrastructure
|
|
- Problems present:
|
|
- Energy Security
|
|
- Energy Sustainability
|
|
- Energy equity
|
|
|
|
## Reason
|
|
|
|
- The UK Government pledge
|
|
- Reduce carbon emission by 68% by 2030
|
|
- more , 2035
|
|
- net zero by 2050
|
|
- Renewable with the traditional grid: power flow is less predictable
|
|
- Renewable with changing landscape: user demand is more challenging to predict
|
|
- Just In Time Delivery
|
|
- Dealing with excessive energy
|
|
- Store
|
|
- Sell
|
|
- Run power station
|
|
- Problem: Excess power may damage device or parts of grid
|
|
- Not enough energy
|
|
- Use stored energy
|
|
- Hydro power
|
|
- Buy
|
|
- Problem: risk of black out
|
|
- This means we should use smart grid:
|
|
- 
|
|
- 
|
|
|
|
## Implementation
|
|
|
|
### Optimizing the electric grid
|
|
|
|
- Continuous: Accurate forecast of demand and production to increase cost
|
|
efficiency, and resilience of electricity provision
|
|
- Predictive: Informed planning of renewable energy and storage to reach net
|
|
zero
|
|
- Integration and collaboration: integrate with other infrastructures for
|
|
efficiency and collaborative intelligence
|
|
|
|
### Big data and smart grid:
|
|
|
|
- Sources:
|
|
- 
|
|
- Characteristics of big data:
|
|
- Volume: Amount of data:
|
|
- Likely terabytes in size
|
|
- Stores transactions
|
|
- In form of stream or batch
|
|
- Variety: Lots of data types:
|
|
- Structured or unstructured
|
|
- Multi-factor, and probabilistic
|
|
- Velocity: Speed of data flow:
|
|
- Real time
|
|
- Near real time
|
|
- In stream or batch
|
|
- Value: Extracting insights
|
|
- For business
|
|
- For revenue streams
|
|
- Has operational value
|
|
- Uncertain: Data uncertainty
|
|
- Authenticity on data
|
|
- Origin of data
|
|
- Availability of data
|
|
- Accountability
|
|
- Analytics: the more sophistication, the more value.
|
|
- Hindsight: descriptive of what happened
|
|
- Insight: diagnostic of current situation
|
|
- Foresight: Prediction, and perspective
|
|
- 
|
|
- Example:
|
|
- 
|
|
|
|
### Load shifting / Peak shaving
|
|
|
|
- To flatten the instantaneous energy demand, by load shifting and peak shaving
|
|
- Using the peak and off-peak electricity tariff (tax)
|
|
|
|
### Other considerations for smart grid
|
|
|
|
#### Privacy concern
|
|
- The more fine grained data is, the more risk in privacy
|
|
|
|
#### Flexibility, and comfort of saving electricity
|
|
|
|
- Gadgets owned:
|
|
- TV: least beneficial
|
|
- EV charger: essential to EV owners
|
|
- Work from home vs. Work from office
|
|
|
|
### Use cases : Household power usage
|
|
- Usage differs per number of occupants
|
|
- Try to apply load shifting for each house hold
|
|
- Predcit:
|
|
- Timing for high / low energy usage per day
|
|
- Same in a week and a year
|
|
- Most energy consumption during peak
|
|
- Can you estimate when they wake up or go to sleep
|
|
- predict if work from home or out
|
|
- Design dual tariff to shift load
|
|
- Time of day to trigger
|
|
- If it's the same every day of week
|
|
- Every month of year
|
|
|
|
### Example solutions
|
|
|
|
- Digital twin: has privacy concern
|
|
- Adaptive dual tariff: low rate before, and high rate after threshold
|
|
- Reinforcement learning: to learn about the environment
|
|
- These results in smoothed electric curve
|