Distribution Network Planning A Real Options approach to support - - PowerPoint PPT Presentation

1

Distribution Network Planning

A ‘Real Options’ approach to support decisions on reinforcement versus post-fault demand-side-response (DSR)

Dr Rita Shaw – Tuesday 7th June 2016 Edinburgh University - International Centre for the Mathematical Sciences (ICMS) Conference on ‘Energy Management: Flexibility, Risk and Optimisation’

2

Agenda

Summary and questions Introducing Electricity North West A real-options decision support tool Reinforce, or capacity from customers?

3

Introducing Electricity North West

4.9 million 25 terawatt hours 2.4 million 56 000 km of network  £12.3 billion assets

19 grid supply points  66 bulk supply substations 363 primary substations  33 000 transformers

4

The GB electricity structure

Free Market

Generation Trading

Regulated

Transmission Distribution

Free Market

Retail

All participants regulated by Ofgem

5

Total to be spent on ENWL network 2015 - 2023

RIIO regulatory framework

£1.8

BILLION

£24.6

BILLION

£10 8%

30%

Total to be spent on the GB distribution network 2015 - 2023 Resulting annual average savings in consumer bills in RIIO-ED1 The power distribution part of a dual fuel bill Network reliability increase since 2002

ED1 = Electricity Distribution 14 DNO areas Eight years

RIIO = Revenue = Incentives + Innovation + Outputs

6

The lights are on – what's the problem?

 Smart solutions are the key to unlocking this puzzle 

Smart Solutions Reliability Affordability Sustainability

 The network operator ‘Trilemma’ 

7

Our smart grid development Deliver value from existing assets Leading work on developing smart solutions Five flagship products (second tier/NIC) £42 million Customer choice

8

Capacity to Customers

Capacity to Customers

Utilised capacity

New commercial contracts

Combines proven technology and new commercial contracts Innovative demand side response contracts Remote control equipment on HV circuit and close the NOP

Technical innovation

Latent capacity

Current demand

Effectively doubles the available capacity of the circuit Enhanced network management software

Capacity to Customers unlocks latent capacity on the electricity network

Facilitates connection of new demand and generation without reinforcement Allow us to control a customer’s consumption on a circuit at the time of fault

9

Trial complete ... now when do we use?

When is C2C cost effective ...? ... or when should we reinforce?

Spend £ every year for capacity from customer OR spend £££ now to build capacity in new asset?

Answer depends

• n costs,

capacity and views of future demand

10

Outlook for future demand

Why could demand go up? Why could demand fall?

11

Long-term electricity demand scenarios

52 54 56 58 60 62 64 66 68 70 72 2005/06 2007/08 2009/10 2011/12 2013/14 2015/16 2017/18 2019/20 2021/22 2023/24 2025/26 2027/28 2029/30 2031/32 2033/34 2035/36 GW Historic Gone Green Slow Progression No Progression Consumer Power

Average cold spell peak demand e.g. National Grid Future Energy Scenarios – July 2015 Set of peak demand scenarios, tailored to a specific substation Methodologies for annual update of long-term DNO load scenarios

12

The problem

Uncertain scale and timing of future load So uncertain scale and timing

• f capacity

requirements Objective – cost-effectively provide just the capacity required

(C2C) DSR provides a new source of capacity

BUT capacity delivered in location-specific lumps. Sometimes large reinforcements, sometimes marginal release

13

DSR then reinforce if required

Years

Is this new strategy cost-effective, and risk-appropriate?

RReinforce? Short-term peak scenario Now Could start investment Have DSR or reinforce Higher demand scenario Lower demand scenario Demand level Capacity with DSR Initial capacity Capacity after reinf.

14

Network Innovation Allowance project

Demand Scenarios with Electric Heat and Commercial Capacity Options Create improved demand forecasts and implement in a DNO-appropriate Real Options approach Due to complete by end of 2016 Reports will be at www.enwl.co.uk/thefuture

15

A real-options approach (1)

• Traditional CBA / NPV approach assumes 1 view of future.

‘Real options’ works with the uncertainty.

• RO values flexibility of decision-making under uncertainty

– Branch of mathematical finance, relevant to engineering – Ofgem expressed an interest (initially in relation to GDNs) – Useful as traditional reinforcement is financially material and irreversible

• Flexibility in when and how we invest for network capacity

– eg traditional large reinforcements, or marginal capacity release by DSR or incremental reinforcements

• Based on uncertainty in long-term peak demand scenarios

– And sensitivity to volatilities in demand and in other inputs – Information is delayed

16

A real options approach (2)

Uncertainty is financially material Investment is at least partly irreversible Invest  Abandon  Defer  Expand Flexibility exists Decision to invest based

• n uncertain

information ‘Real options’ are useful for investments when... Worked with University of Manchester on initial development of methodology and tool (Dr John Moriarty and Dr Pierluigi Mancarella)

17

A real-options approach (3)

Key findings Phase 1 report December 2013 “Flexible investment strategies in distribution networks with DSR: Real Options modelling and tool architecture” Phase 1 report can be shared A DNO-suitable approach can be implemented in Excel. Can be based on annually-updated set of probability-weighted demand scenarios, plus demand volatility around those scenarios. Many options exist for decision-metrics on cost and risk

18

Moriarty report – stages in RO model

Probability weights (or more generally, a probability measure) are constructed for the possible states of the world at those times, reflecting how likely the respective states are. For each of these states, the information that would be available to management as a basis for their decision making is identified. The decisions that would be taken by rational management in each of these states are identified, for example using a binomial tree. Identify significant future decision points in the investment project (one potential metric) A probability weighted average is taken

• ver these

possible futures to arrive at the present project value.

A B C D E

19

A real-options approach (4)

When would we use RO?

Applied to every project with DSR potential, or to derive policy –TBC Scoping stage – find useful DSR scale and maximum price before approaching DSR customers Before committing to investment – Justifying efficiency of load-related expenditure before commitment to DSR or reinforcement Options models provide the cost and risk metrics to support decisions about efficient investment Should we do DSR, reinforce, or DSR then maybe reinforce? How much DSR? When? At what price? DSR while wait for demand increase? Large or small reinforcement? Like-for-like or oversized asset replacement?

20

‘Real options’ methodology

Number of occurrences Cost (£)

Strategy A Total NPC Weighted over All Scenarios

Number of occurrences Cost (£)

Strategy B Total NPC Weighted over All Scenarios

Working with University of Manchester to develop cost and risk metrics in a decision-support tool

• with business and regulatory perspectives

0.00 1.00 2.00 3.00 4.00 Excess load (MVA) Strategy B - Boxplot of Excess Load per Year by quartile 0.00 0.50 1.00 1.50 2.00 2.50 3.00 3.50 4.00 Excess load (MVA) Strategy A - Boxplot of Excess Load per Year by quartile

£0 £2,000 £4,000 £6,000 £8,000 £10,000 Network Losses Total Network Losses Total

Total Net Present Cost (£k)

Scenario 1 Scenario 2 Scenario 3 Scenario 4 Scenario 5 Weighted mean

Analysis of general reinforcement paid for by DNO and customer in general, not just connections reinforcement.

21

Objectives of our work

Develop an

• ptions

assessment method for strategic planning Utilise the ‘options’ expertise at UoM to validate Develop an informed position with Ofgem Support C2C for dissemination and BAU transition Simple to apply v Reflects actual project With appropriate recognition by Regulation and Finance colleagues

22

Real options – where we are now

Currently structured into one 34Mb Excel model Derive policy? Streamline for BAU stage after prototype complete? Creating prototype model harder than we thought, but now in use 2 strategies, each with up to 3 interventions Up to 5 demand scenarios, each with 2 x 100 Monte Carlo variations Electricity North West developed UoM’s early prototype

23

RO model structure

Site demand forecasts Framework inputs Strategy A inputs Strategy B inputs Strategy A Strategy B (repeated structure) Cost and risk distributions Least regret cost and risk analysis Capacity output per macro-scenario Cash flow output per macro-scenario

Calculations Summary metrics Inputs

24

Comparing two strategies

Do nothing until Intervention 1 Intervention 2 Intervention 3

Demand Response Invest in Asset Do nothing

Intervention 1

Trigger level as % of capacity 100%

When do I need to commit to a strategy? How much DSR do I need? What does it cost? Is the network risk acceptable? A – DSR then reinforce if required B – Reinforce Engineers define interventions and strategies, compare via model. Model does not define the intervention strategies, or find optimal.

25

Representing risk of excess load

0.00 0.50 1.00 1.50 2.00 2.50 3.00 3.50 2015 2017 2019 2021 2023 2025 2027 2029 2031 2033 2035 2037 2039 2041 2043 2045 2047 2049 2051 2053 2055 2057 2059 2061 Excess load (MVA) gy p p y q 0.00 0.50 1.00 1.50 2.00 2.50 3.00 3.50 2015 2017 2019 2021 2023 2025 2027 2029 2031 2033 2035 2037 2039 2041 2043 2045 2047 2049 2051 2053 2055 2057 2059 2061 Excess load (MVA) gy p p y q

Strategy A – Boxplot of Excess Load per year by quartile Strategy B – Boxplot of Excess Load per year by quartile

100% - excess load not above this value 75% - upper quartile likelihood 50% - median excess load 25% - lower quartile likelihood

0% - excess load not below this value

26

Example probability distributions

DSR is always cheaper, but with greater uncertainty in total cost (width of distribution).

Number of occurrences Cost (£)

Strategy A Total NPC Weighted over All Scenarios

Number of occurrences Cost (£)

Strategy B Total NPC Weighted over All Scenarios

0.00 0.20 0.40 0.60 0.80 1.00 1.20 1.40 1.60 1.80 2.00 2014 2015 2016 2017 2018 2019 2020 2021 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 Excess load (MVA)

Strategy A - Boxplot of Excess Load per Year, Weighted over All Scenarios

Med-Q1 Q3-Med 0.00 0.50 1.00 1.50 2.00 2.50 3.00 2014 2015 2016 2017 2018 2019 2020 2021 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 Excess load (MVA)

Strategy B - Boxplot of Excess Load per Year, Weighted over All Scenarios

Med-Q1 Q3-Med

DSR Reinf. DSR Reinf.

Reinforcement strategy carries network risk during implementation lead-time. Network risk is smaller with DSR, as DSR more quickly adjusted to network load, but occurs over longer timeframe. Net present cost Network risk over time (probability and scale of load exceeding capacity)

27

At least two NPV financial views The model rapidly presents two vital perspectives on cost DNO commercial view Regulatory customer-view based on the Ofgem Cost Benefit Analysis (CBA) framework for setting RIIO-ED1

28

Comparing strategies across views

Total cost saving to customers inc. losses in Regulatory View

3.5% discount rate for 45 years, inc. RAB financing and losses

Yes No Cost saving in DNO commercial view

Higher discount rate, Fewer years, Different incentives

Yes Proceed, good for DNO business and customers Do not proceed, forego cost saving to DNO since high losses costs to customers No Proceed as losses reduced, but increased cost to DNO so needs regulatory support? Do not proceed, no benefit for DNO or customers Total cost saving to customers inc. losses in Regulatory View

3.5% discount rate for 45 years, inc. RAB financing and losses

Yes No Cost saving in DNO commercial view

Higher discount rate, Fewer years, Different incentives

Yes Proceed, good for DNO business and customers No Do not proceed, no benefit for DNO or customers

Cannot imagine DSR which reduces losses, so irrelevant for DSR, but this is the case for low-loss transformers

29

Use of the model in practice

Set up inputs, sense-check capacity chart v. scenario

Step 1

Check residual excess load acceptable within the planning horizon for both strategies

Step 2

Compare the strategies based on commercial cost perspective

Step 3

Compare the strategies based on customer cost perspective

Step 4

Make business-decision on multiple criteria, may include outputs outside of the model

Step 5

Hurdle

30

Real options – next phase

Explore / validate those different financial perspectives Explore new case studies Complete the prototype Develop the decision methodology Transition to business as usual

31

The future approach

Target of £10m against investment plan First customer signed Implement through new managed connection agreement and retrofit automation equipment

• n agreed switch

Decision making process includes viability of DR option Electricity North West personnel to approach customers to purchase post fault DR

Ongoing relationship management required

32

Benefits

Stimulates the market for future commercial solutions to manage the network Contribute to

• utperformance
• f RIIO-ED1

Facilitates a reduction in carbon costs of network reinforcement Customer satisfaction Avoid costs and risks due to uncertainty of demand and connection of low carbon technologies

33

Want to know more? Thank you for your time and attention

futurenetworks@enwl.co.uk www.enwl.co.uk/thefuture 0800 195 4141 @ElecNW_News linkedin.com/company/electricity-north-west facebook.com/ElectricityNorthWest youtube.com/ElectricityNorthWest

e