1. Cost reflectivity of investment proxies
CAP131 Working Group (4)
17 November 2006
Slides updated to reflect:
feedback and comments from WG2
include available data from Scotland
alternative proxies raised
effect of different substation costs (the collar)
comparison to current FSL and entry CAPEX
“s-curve” comparison – 25/50/75/100

2. Action
Action
To explore the cost reflectivity of various proxies for infrastructure investments, which need to be supported by a user commitment
Aim
To establish whether there are more cost-reflective investment proxies than TNUoS tariffs and consider whether these are “better” when assessed against the Applicable CUSC Objectives

3. Content Analysis framework and methodology
Locational investment proxies
zonal proxies
nodal proxies
sensitivity analysis
Non-locational investment proxies
Wider assessment of proxies
Conclusions

4. Cost reflectivity of investment proxies
Analysis framework and methodology

5. Process Map
Identify groups of projects that require similar system reinforcements
Establish unit capital investment costs (£/kW) for groups
Compare annual investment proxy to capital investment costs
Calculate average & spread between project groups as measure of cost reflectivity
Express as a number of years of the investment proxy

6. Investment costs for groups
Several groups of projects considered, where each group requires similar transmission investments
represents a sample of ~31GW of generation (between )
Investments consider a range of scenarios within group
One investment group is illustrated below
Gradient gives average capital investment cost in £/kW
[Note, all investment costs expressed in unit terms, where kW refers to generation capacity]
P1+P2+P3 P3 P2 P1

7. Number of project scenarios
Project group data
Project Group
Gen Capacity (MW)
Number of project scenarios
Group A (EW)
2400 12
Group B (EW)
4300 10
Group C (EW)
5100
Group D (EW)
Group E (EW)
3100 6
Group F (EW)
2200 3
Group G (EW)
2500 24
Group H (Scotland)
3400
1 (contracted)
Group I (Scotland)
23 E&W projects considered 67 Scotland projects considered

8. Establishing number of years of proxy
For each group, actual investment costs (£/kW) are compared with applicable investment proxy (£/kW):
where a zonal proxy has been used, the zone that best represents projects within a group has been used
Range of annual proxies
For each Group
Actual Capital Cost (£/kW)
Number of years of proxy
Annual
Proxy Cost (£/kW) = ×
across all investment Groups
Average and St. Dev.

9. Cost reflectivity of investment proxies
Locational Investment Proxies
Zonal proxies
Nodal proxies
Sensitivity analysis

10. Investment proxies considered Locational proxies
All proxies been derived from the TNUoS model
Retained £3/kW p.a. de-minimis for local substation works
Nodal investment proxies
full tariffs
locational tariffs
Zonal investment proxies
full TNUoS tariffs
locational TNUoS tariffs
re-zeroed TNUoS differential tariffs (no residual)
Removes the revenue recovery adjustments i.e. the residual
To remove effects of an arbitrary slack node (see sensitivity analysis)

11. Zonal investment proxies
Slide illustrates shape of the proxies compared against TNUoS.
[Easiest to explain shapes starting with TNUoS, then add the collar, then remove residual, then re-zero. In practise, the order does not matter]
£3 /kW p.a. collar

12. Analysis results Zonal investment proxies
Zonal Investment Proxy
Years of Proxy
Average SD
Zonal TNUoS tariff
12.8
3.7 (29%)
Zonal TNUoS, locational only
16.4
5.8 (35%)
Zonal TNUoS, re-zeroed at London
8.0
1.8 (27%)

13. Nodal investment proxies
Where possible, the nodal marginal km at the node where generation is expected to connect has been converted to a tariff by considering:
re-referencing quantity
security factor
expansion constant
Has not been applied to all projects, as nodes do not presently exist within transport model
analysis covers ~19GW of sample pool (~31GW)
England & Wales only
same parameters used to calculate zonal tariffs

14. Analysis results Nodal investment proxies (c.f. Zonal)
Nodal Investment Proxy
Years of Proxy
Average SD
Nodal “TNUoS tariff”
15.1
5.3 (35%)
Nodal “TNUoS”, locational only
17.7
6.0 (34%)
Zonal Investment Proxy
Years of Proxy
Average SD
Zonal TNUoS tariff
12.3
3.6 (29%)
Zonal TNUoS, locational only
15.8
6.6 (36%)
Zonal TNUoS, re-zeroed at London
7.6
1.9 (24%)

15. Observations Locational proxies
Nodal approach does not add extra cost-reflectivity
no smoothing for lumpiness of a nodal cost
adds complexity to the methodology for little return
Trade-off between number of years of the investment proxy and the magnitude of that proxy
low number of years, high proxy cost
high number of years, low proxy cost

16. Sensitivity analysis Two key sensitivities tested:
effect of different zero points
effect of different substation costs

17. Why consider re-zeroing?
Additional generation triggers reinforcement to load centres
DCLF model establishes a system centre (the “slack node”)
The “slack node” is the most interconnected node and is arbitrarily assigned an incremental cost of zero (£0MWkm)
The ‘real’ incremental cost at the “slack node” is not zero
investment required to transfer the marginal MW to the load

18. Is Thorpe Marsh an appropriate centre?
Costs incurred “south” of the system centre to transport energy to location of load
CAP131 is about the commitment made prior to costs being incurred
Thorpe Marsh is the slack node (SN)
“B7” SN
“B8”

19. Effect of different zero points Alternative zero-points considered
Investigated the effect of considering different zero points
Have considered:
Peninsula (Zone 21)
Central London (Zone 16)
S.Wales & Gloucester (Zone 19)
South Yorkshire & North Wales (Zone 14)
Peterhead (Zone 1)
The aim is to identify which zero-point provides the best fit to actual investment costs incurred
Contains the slack node, Thorpe Marsh
[Zone 14 was used effectively as the “baseline” against which the other options were considered. Zone 16, Central London, is an obvious choice given the flows typically observed on the system. Zone 21, Peninsula is a bounding extreme.]
[The residual has been removed to exclude revenue collection effects]
[There might be other alternative ways by which an alternative zero-point could be established, e.g. some sort of sensitivity analysis, but don’t expect this to add much.]

20. Effect of different zero points Investment proxies with different zero points

21. Effect of different centres of load Identifying the most appropriate zero point
Note, a lower number represents a better fit
Best Fit

22. Local substation costs
Used to determine de-minimis level of investment proxy
Calculated the cost of a generic substation for a 1GW plant
designs will clearly vary, some more expensive others less but this is a generic methodology
Expected capital cost ~£30m
For interim arrangements, a 10 year commitment was sought and therefore an annual £3/kW commitment needed
Needed to check whether still appropriate
Assessed sensitivity of different local costs

23. Local substation costs Sensitivity results
Note, a lower number represents a better fit
Best Fit

24. Cost reflectivity of investment proxies
Non-Locational Investment Proxies

25. Investment proxies considered Non-Locational, nodal proxy
An alternative investment cost proxy suggested within Working Group is based on:
capacity requested, TEC (MW)
length of local line required to connect, L (km)
minimum of 15km for substation assets
TO and voltage / circuit specific expansion factor, EF
expansion constant, EC (£/MWkm)
Generic cost proxy = TEC × L × EF × EC
… shouldn’t the security factor be in here too?

26. Illustration Investment Proxy = TEC (MW) × L (km) × EF × EC (£/MWkm)
Assumes 400kV OHL
EF = 1
EC ≈ 10 £/MWkm
500MW, 15km
Liability = £75k
650 marginal km
Proxy does not consider the costs associated with wider reinforcements shown by the red arrows SN
500MW, 15km
Liability = £75k
-500 marginal km

27. Observations Non-Locational, nodal proxy
Does not consider any wider reinforcement costs associated with a given project at a given location
Very difficult for users to apply prior to application
length of connection unknown
circuit type unknown (OHL / cable, voltage)
Potentially subjective use of expansion factors
what is a representative value for project / system / TO region

28. Cost reflectivity of investment proxies
Wider assessment & Conclusions

29. Wider assessment of investment proxies
Cost reflectivity is not an applicable CUSC objective, it is part of facilitating competition
Important, therefore, to consider other criteria against which the investment proxies should be assessed
transparency of methodology to derive proxy
ease for both Users and National Grid to determine commitment
ease of application to new entry points i.e. nodes that don’t exist
others?

30. Wider assessment of options
Investment Proxy
Transparency of proxy
Ease of use by Users
Copes with change
Cost Reflectivity
Locational proxies
Zonal TNUoS tariff
  
 
Zonal TNUoS, locational only 
Zonal TNUoS, re-centred
Nodal “TNUoS tariff” 
 
  
Nodal “TNUoS”, locational only
Non-locational proxies
Expanded connecting line length

31. Cost Reflectivity Comparison

32. Timing of spend – actual v generic (Sample size 24 projects)

33. Shape Comparison - 25/50/75/100 v 40/75/85/100

34. Cost Reflectivity Conclusions
A pure non-locational approach is a poor investment proxy
TNUoS tariffs provide a reasonable investment proxy
Locational TNUoS tariffs re-centred on Central London provide a more cost reflective locational investment proxy
But, this would require arrangements to overcome transparency issues and could confuse new entrants
A de-minimis level of £3 /kW reasonably reflects local costs
X=12.8 is a good approximation of total TO planned entry CAPEX
25/50/75/100 provides a better overall fit