Optimal design of a subsea power grid

1
Optimal design of a subsea power grid

• T. Trötscher, M. Korpås, J. O. Tande
• SINTEF Energy Research

2
Setting and drivers

• More exchange capacity between countries
• Large scale integration of off-shore wind
• Shore power for oil and gas rigs
• Scandinavian hydro power for balancing
• New HVDC technology

Take a holistic approach Do not build parallel
infrastructure
3

• Hydro power
• Reservoirs
• Regulating power

Wind farm
Ekofisk Oil rig
?
NORD.LINK / NorGer
4
The problem to be addressed
Problem Connect off-shore wind farms and oil-rigs to the on-shore grid
Objective Maximize social economic benefit
Exogenous variables Location and power rating of wind farms, possible land connection points, statistical description of wind and power prices. Consumption at off-shore loads
Unknowns Where to build cables and with what power rating
and the challenge to overcome
the number of possible cable configurations
grows exponentially with the number of connection
points. It is impossible to evaluate all
combinations.
5
Solution

• Its possible to formulate the problem as a
  standard linear or quadratic optimization program
  by making some assumptions
• Constant or linear marginal costs
• Linear representation of the underlying physical
  power grid
• Stochastic nature of wind power generation and
  power prices are represented by a random sample
  of historical data

6
Example model
Node Comment
Møre Wind farm 1000MW
Troll Wind farm 1000MW and load 200MW
Ekofisk Wind farm 1000MW and load 200MW
Bergen Generation capacity 3000MW and load 2000MW
Stavanger Generation capacity 3000MW and load 2000MW
Wilhelmshafen Generation capacity 4800MW and load 2400MW
7
(No Transcript)
8
Results
788 mill. EUR
7531 mill. EUR

• Optimized grid

Optimized grid, locked to radial connections
Optimized grid without wind power
9
How large does the random sample need to be?

• To take into account the stochastic nature of
  wind power generation and power prices, the
  optimization is conducted with regard to a random
  sample that represent various wind conditions and
  prices levels. How large must the random sample
  be?

• Convergence for a sample size of 150. Model
  dependent!

10
Conclusion

• The model provides plausible results
• Optimized grid is better than radial solution
• Value of wind power is given implicitly
• Useful as a decision support tool
• Systematic approach
• Stepwise grid expansion
• Future work
• Cost function for cables
• Input ? Result
• Sensitivity analysis