Section 4 - BG/Keyspan Proposal

Section 4: BG/KeySpan Proposal

4.1 Introduction

The BG/KeySpan document examines the costs of supplying the island of Ireland’s natural gas needs through a North-South pipeline, and through twinning the existing Scotland-Ireland pipeline¹⁶

Unlike the BGE/Sofregaz Gas 2025 report, it contends that supplying Ireland’s future needs through Northern Ireland is a better option than twinning the existing Scotland-Ireland pipeline.

Throughout this section, for convenience, we refer to this Northern Ireland route as the BG/Keyspan option. It is the equivalent of UK2 in the Sofregaz report, while the Scotland-Ireland route is the equivalent of UK1.

BG/Keyspan believe that Bord Gáis Éireann’s favouring of UK1 is based on a scenario where

there is very high gas demand in Ireland over the period, and

there is no indigenous gas discovery post-Kinsale.

They believe that this scenario has a low probability of actually happening. They also state that their preferred option is "likely to attract EU Grant (unlike the Scotland-Dublin duplicate)", a situation which they contend further favours their preferred option.

Their proposal would involve the construction of a North-South pipeline, and reinforcement (through duplication and compression) of the existing Scotland-Northern Ireland pipeline (SNIP) as needed.

BG/Keyspan calculate the NPV of their proposal and the twinned Scotland-Ireland pipeline, using three demand scenarios high, central case, and low. The high and central cases are the same as Gas World B and Conventional Wisdom in the original BGE/Sofregaz Gas 2025 report. The low case is made up of the ESB’s low demand forecast for gas for electricity generation, combined with 90 per cent of the central demand case for other consumers.

Their calculations indicate that their proposal has a lower NPV in all but the high demand case.

Furthermore, because investment is less front-loaded in their proposal, it is more flexible if actual demand turns out to be less than forecast, or indigenous gas comes on stream.

They also claim a greater security of supply benefit from their option. BG/Keyspan quote a 1997 study by Ove Arup on a possible N-S pipeline, which they state estimated that the improved security of supply for the island as a result of such a pipeline could be worth the equivalent of IR£40 million, in 1996 money. The question of security of supply is returned to in more detail later in this report.

Their costings are as follows, using the central demand case and an 8% discount rate:

Despite being more expensive in absolute terms, both for capital and operating costs, BG/Keyspan claim that their option is cheaper in NPV terms, by Stg£16 million. This could be due to expenditure being more back-end loaded for BG/Keyspan’s preferred options although BG/Keyspans aggregation of capital expenditure in five-year tranches affects the NPV calculations. The following chart demonstrates this:

For the high and low demand cases, the equivalent values are as follows:

BG/Keyspan estimate that the Scotland-Ireland route is better by £39 million (NPV) in the high demand case, while in the low demand case the Northern Ireland route is better by £24 million.

The Ireland and Northern Ireland demand scenarios used by BG/Keyspan are given in the following

Tables. Only the high and central demand scenarios are specified.

A number of factors are relevant in the Northern Ireland scenarios:

The central case does not foresee increase in the powergen usage of gas, beyond the amount already being used. However, there are proposals to replace the existing gas-fired capacity at Ballylumford with new CCGT plant, which will be more efficient. This will allow for the expansion of electricity generation from gas.

The scenarios foresee very strong growth in the non-powergen market, up to 2015. Thereafter, there is no growth.

There is almost no growth in any sector after 2015.

The low demand scenario is not specified. From discussions and information provided by BG personnel, this scenario relates to lower than expected demand in the Republic. For the electricity market, it is based on the ESB low demand forecast, and involves having one less power station on stream up to the year 2015, after which it is the same as the central demand case. For other markets, the scenario is 10 per cent less than the central demand case.

The BG/Keyspan report gives summary capex (split between pipelines and compressors) and opex costs, lumped together on a 5-yearly basis, for the central demand case. For the high and low demand cases, only the figures for the entire period of 25 years are given. We requested annual figures from BG, along with details of the elements of capital investment, but they declined to provide them, citing confidentiality reasons.

This limits our ability to comment on the report or its findings.

The differences between the NPVs under the various options are relatively small. In the central case the difference is only £16 million, while in the low demand case it is only £24 million. Indeed, the largest difference is in the high demand case, which favours the Scotland-Ireland option by £39 million.

Since the results of BG/Keyspan’s cost estimations are not robust across the various demand scenarios, it would be useful to attach probabilities to the three scenarios, and work out the resultant weighted average NPVs. If we were to attach equal probabilities (i.e. 1/3) to each scenario, then the weighted average NPV difference would be

24 x 1/3 + 16 x 1/3 + (-39) x 1/3 = £0.3m.

In other words, there is no difference between the NPVs of the BG/Keyspan option and the Scotland-Ireland option, in terms of expected values, where the three probabilities are equal and the assumption that aggregation of capital expenditure in five-year tranches has no impact on BG/Keyspans NPV calculations.

Equal probabilities may not be appropriate BG/Keyspan state that "in our view (the high demand) scenario has a relatively low probability" - but so long as there is a positive probability to the high demand scenario, it weakens the argument for the BG/Keyspan option.

In any event, whatever number is arrived at, it must be judged modest given the margins of error involved in the exercise.

It would appear that despite being more expensive in absolute terms, the BG/Keyspan option is cheaper in NPV terms, at a discount rate of 8 per cent and the central demand case, because it does not involve as much up-front investment.

The 8 per cent rate is higher than that used by Sofregaz (.), which in turn is higher than the usual rate of 5% used for Irish public investment in non-commercial projects (commercial projects in the public sector, attract a higher discount rate to reflect the risk profile of the project) or the 7% used in both the evaluations of the first interconnector and the Ove Arup report referred to above. Rates as high as 7 or 8% are no longer being permitted to regulated utilities in the United Kingdom.

Given the foregoing, it is appropriate to test the BG/Keyspan proposal for sensitivity to different discount rates. This cannot be done with accuracy, because the report only gives 5-yearly capex and Opex figures, and these only for the central demand case. As already mentioned, BG would not provide us with yearly figures, citing confidentiality reasons.

Without yearly figures it is impossible to carry out a proper sensitivity analysis. However, we did carry out an exercise based on the 5-yearly costings in the report, for the central demand scenario.

We first tried to re-estimate BG/KeySpans NPV calculations, from the capex and opex figures, using a discount rate of 8 per cent. This suggested that BG/KeySpan back-loaded expenditure within each five-yearly period. Their rationale for doing this is not clear.

Subjecting this imputed expenditure schedule to a range of discount rates indicates that the BG/KeySpan and Scotland-Ireland options have an equal NPV at a discount rate just below 6 per cent. Any rate lower than this will favour the latter route.

A different expenditure schedule, for example where expenditure happened in the middle of each five-year period, or was spread evenly throughout each period, would further favour the Scotland-Ireland route.

The question of sensitivity to discount rates is returned to later in Section 9 of this report.

The Capex figures in the BG/Keyspan report are for the most part lower than the corresponding BGE/Sofregaz capex figures. The costs compare as follows:

These differences are in fact smaller than the differences between the BG/Keyspan figures and those in the original BGE/Sofregaz Gas 2025 Report.

In order to reconcile these differences, the Department of Public Enterprise and the Department of Economic Development in Northern Ireland commissioned a study, referred to as the Shannon-McManus report. This is the subject of the next section.

The BG/Keyspan document seeks to demonstrate that bringing gas to Ireland via Northern Ireland is a better option than bringing it through a second interconnector from Scotland.

Their report is succinct, and lacks much of the detail contained in the BGE/Sofregaz report. This limits the degree of analysis that can be carried out.

However, a number of observations can be made:

The capital cost of the BG/Keyspan’s preferred option is higher than that of the Scotland-Ireland interconnector. It is cheaper in NPV terms because expenditure is postponed, assuming aggregation of capital expenditure in five-year tranches has no effect on relative NPVs, but this is dependent on the discount rate used. BG/Keyspan use a rate of 8 per cent, which is higher than that normally used in Ireland for projects of this type. At a discount rate of 6 per cent, the cost advantage of the BG/Keyspan option disappears, on the timing assumptions we have made.

The advantage of the BG/Keyspan option is not robust with respect to demand scenarios. In the high demand scenario, the Scotland-Ireland interconnector is cheaper. If we attach equal probabilities to the three scenarios examined, and calculate a weighted average NPV for the two routing options, then both options cost the same, using BG/Keyspan assumptions.

The differences between the NPVs of the various options, on BG/Keyspan assumptions, are in any event small for projects of this size, and are within the margin of error for an exercise such as this.

There are sources of difference between the BG/Keyspan and Sofregaz studies other than the discount rate used. Unit cost assumptions are slightly different but more important differences arise due to differing technical assumptions about gas flow rates and related matters. These are considered in the next section.

To Section 5