Dr Saad Sayeef - CSIRO Energy - Bringing the Smart Grid to RAPS

Bringing the Smart Grid to RAPS

CSIRO ENERGY

Dr Saad Sayeef| Research Scientist RAPS 2016, Melbourne

March 2016

• Australia’s national science agency

• Established in 1926

• Over 5000 staff

• Annual budget ~A$1.2B

• 184 companies based on CSIRO IP

• 3500 patents granted or pending

• Currently working on projects in 68 countries

• Our Mission: • We deliver great science and innovative solutions

for industry, society and the environment

Introduction to CSIRO

Our business units – simplifying the model

9 Towards National Research Flagships

+ National Research Facilities

and Collections

FOOD, HEALTH & LIFE SCIENCE

INDUSTRIES

ENVIRONMENT

MANUFACTURING, MATERIALS &

MINERALS

ENERGY

INFORMATION & COMMUNICATIONS

Load

1

2

3

4

RAPS Systems Stronger Constraints, Effects and Motivations

Is it so far away?

Plug and Play Software

Local Controller

Local Controller

Local Controller

Local Controller

System Simulator

System Optimiser

System Planner

Manual Data Input

Original network and systems configuration

Performance Store

Solar Systems Explicit Storage

Systems Loads & Implicit

Storage

Spinning Reserve & Distributed Generation

System State

Recorder

System Controller

System State

Forecaster

Environmental Sensors

Plug and Play Solar

Plug and Play System Design

Plug and Play Software Integrated tools for operation, planning and review of a hybrid energy

deployment

– System Simulator

– Estimates the state of the underlying electricity network

– System Optimiser

– provides high-level control signals for on-site storage, discretionary loads, spinning reserve and other inverter-based systems

– System Planner

– Identifies optimal energy options according to planning objectives (capital expenditure, maintenance costs, environmental performance, energy export revenue and system reliability)

– Performance Store

– catalogues the software control signals and resultant system responses. This provides a mechanism for fault detection and diagnosis should a system error occur

Skycam solar forecasting

(Very) short-term solar forecasting

10-30 min forecast horizon, 10s update

Inexpensive whole-sky cameras (~$1000)

Low cost hardware means: Widespread deployment more practical, but

Image processing more challenging

Cloud Classification – Unprocessed Image

Red-Blue Ratio (RBR) Classifier

- Misclassifies near sun - Misses thin and near-horizon

clouds

+ Good performance for overcast & dark clouds

Random Forest Classifier

Misses dark & non-textured cloud areas

+ Sensitive to cloud edges, thin & distant clouds

+ Good near-sun performance

Combined RBR + Random Forest Classifiers • Bright red = models agree

• Darker red = only one model detects

cloud

+ The two models are complementary

New Model: Final Combined Result

Cloud Presence Detection Demo • Sudden cloud formation event

• Left chart: measured DNI, GHI, PV Power

• Red histogram bars show cloud % in concentric rings around sun

• Sharp increase, ~7 min before gives ample warning of shading event

Cloud Motion Vector Forecasting Demo • Clear morning, intermittent afternoon, approaching cloud front

• Detected 25 minutes in advance of shade event

• Left chart: measured DNI, GHI, Diffuse, PV Power, and forecast Cloud Pixel

Fraction

Different clouds = Different challenges

Diesel savings vs. solar forecast accuracy

De-risking, validation and development

CSIRO Minigrid Laboratory Large Commercial

Commercial

Residential

Thank You Dr Saad Sayeef Research Scientist Grids and Energy Efficiency Systems CSIRO Energy saad.sayeef@csiro.au

CSIRO ENERGY

Why? Solar Forecasting Timeframes & Applications Ground-based imagery (skycam) forecasting is useful for a range of applications

Fast update rate and high spatial resolution

Can forecast cloud movement accurately 1-30 min ahead

Short-term forecasting is

suitable for many

applications