Ian Wal is Associates Ltd

Auckland Transport

Auckland light rail peer review -

Part C: Cost estimates (Draft 2)

26 July 2015

N207/Rep/1552

Strictly
confidential

Ian Wallis Associates Ltd

Ian Wal is Associates Ltd
EXECUTIVE SUMMARY
1.  Scope
This is one of three papers prepared for Auckland Transport by consultants Ian Wallis Associates (IWA),
as the outputs of a (limited) peer review of AT’s proposals for a light rail network for Auckland.
This paper reviews the cost estimates prepared by the AT project team, and specifically:
“Whether the estimates of incremental public transport capital costs (infrastructure
and vehicles) and operating costs, relative to the bus-based base case, are sufficiently
robust for this stage of option assessment”.
This paper reviews the cost estimates prepared by the AT project team for the LRT option and the
alternative bus option, under three main cost groups: operating costs, vehicle capital costs, and
infrastructure capital costs. For each of these groupings, the review examined whether the project
team cost estimates are sufficiently robust at this stage of the project assessment, mainly by
comparing these estimates against cost data for LRT schemes in Australia.  Where the cost estimates
appear inconsistent with this evidence, it provides estimates, at an indicative level, that appear more
appropriate.
2.  Project team cost estimates
The project team best estimate costs for the LRT scheme were (NZ$2014):
•  Infrastructure capital costs: $2606 M
•  Annual operating costs (including vehicle capital charges): $36 Mpa, compared to equivalent
operating costs for a bus option of $67 Mpa.
3.  Peer review assessment
For infrastructure capital costs, we consider that the project team estimate ($2606 M) is broadly
consistent with infrastructure costs involved in recent LRT schemes (now operational or in the
planning/procurement stage) in Australia, after allowing for the characteristics of the proposed
Auckland scheme.
For operating (including vehicle capital) costs, we consider that:
•  The LRT operating costs are very considerably under-estimated.
•  The LRT vehicle capital costs are largely consistent with other evidence sources, but their
conversion into equivalent annual capital charges is not consistent with normal practice for
such schemes.
•  The bus operating costs and vehicle capital charges are broadly consistent with our own
estimates.
Based on these findings, our best estimates (aside from the infrastructure capital costs), are that:
•  The LRT option costs would be about $116 Mpa (including $16 Mpa for feeder buses in the
corridor) compared with the project team estimate of $36 Mpa (excluding any feeder buses).
•  The comparable bus option costs would be about $86 Mpa compared with the project team
estimates of $67 Mpa.
The table sets following out the main components of the project team’s costing analyses and our
assessment of these, that together result in the above summary findings.
IWA/N207/Rep/1554
26 July 2015
3

Ian Wal is Associates Ltd
TABLE: SUMMARY OF COSTING FINDINGS
Aspect
Project team assessment
Peer review comments and assessment
A.  VEHICLE TYPES AND CAPACITIES
1.  LRT vehicles
• Dimensions: length 45m (7 modules, 100% low floor).
• Dimensions: agree, reasonable assumption for planning purposes at this stage of
project.

• Capacity: planning standard 300 passengers/vehicle (based on full  • Capacity: standard of 4 standees/m2  not reasonable as peak hour average
seated load plus 4 standees/m2 of available floor area).
(although may be obtainable on individual trips). Based primarily on Melbourne
tram/LRV standards (and also having regard to AKL & WGN metro-rail standards),
propose planning standard of 2.9 standees /m2, resulting in total vehicle planning
capacity of 240 passengers /vehicle (ie 20% below project team figure).
2.  Bus vehicles
• For main analyses,  assumed ‘standard’ size buses in corridors  • Accepted project team assumptions on bus types and capacities as reasonable at
affected, with c. 37 seats and effective planning capacity of 50
this stage of project. However, have undertaken main cost assessments based on
passengers,  ie similar to majority of current AKL fleet. Also
both single-decker (50 passenger capacity) and double-decker (100 passenger
assumed larger single-deck buses (as now) on busway services. Also
capacity) vehicles.
considered double-decker buses, with up to 90 seats and effective
planning capacity of c. 100 passengers (did not use these in costing
analyses).
B.  VEHICLE CAPITAL COSTS
1.  LRT vehicles
• Two sets of estimates were made based on the above vehicle type,  • From review of recent information (mainly in Australia) on LRV costs, and al owing
in both cases relating to wire-free technology (ie battery-based,
for  wire-free technology, our best estimate of costs was about 10% below the
with recharging arrangements yet to be determined): WT  project team $6.80M (for convenience we used the $6.13M figure).
Partnership estimate $6.13M, Turner & Townsend $6.80M. This
latter figure was used in the costing assessment.

• The cost assessment also assumed a 30 year vehicle life, and vehicle  • Accepted the project team assumptions on vehicle life and mid-life renewal.
‘renewal’ after 15 years (at 20% of initial cost).  The annual capital  • Annual vehicle capital charge was estimated on an annuity basis over the vehicle
charge per vehicle was based on straight-line (historic cost)
life, using a 7.5%pa real interest rate (such an approach is much more consistent
depreciation over the vehicle life.
with a finance lease arrangement, or PPP arrangements, than the historic cost
depreciation approach). The result is for an annual capital charge per vehicle about
twice the project team figure ($530 kpa/vehicle as against $270 kpa).
2.  Bus vehicles  • For main analyses, assumed ‘standard’ bus cost of $500k, 12 year  • Estimated ‘standard’ bus cost of $440k (12% lower than project team) and for
life, and half-life renewal (at 20% of initial cost).  For double-decker
double-decker same cost as project team. In both cases, assumed 20 year life: this
buses, assumed costs would be 30% higher (although these figures
is consistent with general NZ practice, including prevailing practice in AKL and
were not used in the costing analyses).
PTOM contract proposals.
• As for LRV, annual capital charges based on historic cost  • As for LRV, annual capital charge calculated on an annuity basis. Due to off-setting
depreciation.
factors, the annual capital charge/ bus is very similar to the project team figures
(for both standard and double-decker buses).

IWA/N207/Rep/1552
26 July 2015
4

Ian Wal is Associates Ltd
C.  OPERATING RESOURCES
1.  LRT mode
• The project team estimated annual operating resources for the  • Based on the project team passenger capacity standard (300/vehicle), peer review
LRT operation in the planning year (2046) as:
estimated slightly (6%-7%) lower operating resources required. On proposed
o  Peak vehicles
49 (plus 10% spare vehicles)
adjusted standard (240/vehicle), peer review estimates were 60 PVR and similar
o  Service hours
181,000 pa (includes 10% for layover
service hours and service km to project team.


between trips)
o  Service km
2.94 mil ion pa
o  Route km
32.1km
2.  Bus mode
• Project team estimate of bus operating resources saved by  • Peer review estimates of (gross) bus operating resources saved by adoption of LRT
adoption of LRT scheme was:
were quite substantial y (c. 38%) higher than project team estimates.
o  Peak vehicles
254 (plus 10% spare vehicles)
o  Service hours
724,000 pa
o  Service km
10.92 mil ion pa

• No allowance appears to have been made for the residual  • Peer review  also estimated resources for residual (feeder) bus services required
(feeder) bus services that would be required at the southern end
(based on project team specification for these services).  After al owing for these
of the corridors served by LRT(1).
services, peer review estimates of net bus mode resource savings remain higher than
project team estimates, by c. 13%.
D.  UNIT OPERATING COSTS
1.  LRT mode
• The project team LRT operating cost model (based mainly on 1999  • The peer review team developed an average cost model based principal y on recent
UK data), excluding vehicle capital charges, was as fol ows:
information from Australian LRT systems, both those currently in operation

Opex = $3.41 * service km + $49.20 * service hr +
(Melbourne, Adelaide, Gold Coast) and those in the planning/procurement stage

$90,500pa * route km
(Sydney, Canberra). This work showed a range of operating costs of A$16-26 per
For the proposed operation, this gives an average cost of
service km. From this information, it was judged that the best estimate cost for the
$7.43/service km.
AKL scheme was about NZ$22 per service km. This figure is almost three times the
project team average cost estimate.
2.  Bus mode
• The project team bus operating cost model (derived from AT  • The peer review team reviewed the project team model, including using recent data
sources) was as fol ows:
on bus operator costs both in AKL and elsewhere in NZ. It was concluded that:

Opex = $2.10 * service km + $41.00 * service hour
o  The project team costs should be reduced by c. 5% to reflect recent AKL
Our understanding is that each of these cost items includes a
operator cost information. This adjustment was adopted for our main cost
component for bus operations overheads.
estimates.
o  A further cost reduction of up to 20% would be appropriate  to match NZ
efficient benchmark cost rates: this was used in sensitivity testing.
E.  TOTAL OPERATING (INCLUDING VEHICLE) COSTS

• The fol owing table presents the project team main estimates of  • The fol owing table presents the peer review main cost estimates, on a comparable
operating costs (including vehicle capital charges) for the  basis to the project team estimates for 2046.
representative year of 2046 – comparing the LRT option with the

bus option for the corridors concerned.
IWA/N207/Rep/1552
26 July 2015
5

Ian Wal is Associates Ltd

Cost category
Annual Costs $M
Bus Option
LRT option
Cost category
Annual Costs $M
Bus option
LRT option
Bus – op costs
52.6
--
Bus – op costs
68.7
12.7
Bus – cap charges
14.7
-
Bus –cap charges
17.7
3.2
LRT – op costs
--
21.8
LRT – op costs
--
65.5
LRT – cap charges
--
14.4
LRT –cap charges
--
34.9
Total costs pa
67.3
36.2
Total costs pa
86.4
116.4

• The principal sensitivity tests undertaken by the peer review on the above estimate
found:
o  Increasing LRV capacity from 240 to 300 (used by the project team)  would
reduce the (LRT-Bus) cost differences by c. $13 Mpa.
o  Using double-decker buses in place of standard buses in the bus option would
increase the (LRT-Bus) cost differential by c. $33 Mpa.
o  Reducing bus operating costs (excl capital charges) by 20% (as in efficient
benchmark cost estimates) would increase the (LRT-Bus) cost differential by c.
$11 Mpa.
F.  INFRASTRUCTURE CAPITAL COSTS
1.  LRT option
• Three sets of estimates were made by the project team of the  • The peer review of the project team infrastructure cost estimates considered the
infrastructure (‘design and construction’) costs for the LRT
recent Australian evidence on infrastructure costs for recent and planned LRT
schemes, as fol ows (NZ$2014):
schemes. It was concluded that the final/adjusted WT Partnership cost estimate of c.
o  WT Partnership (initial, corrected): $1631 mil ion
$2.6 Bil ion is broadly consistent with the review’s expectations based on the
o  Turner & Townsend (peer review): $3305 mil ion
Australian evidence and al owing for the AKL situation. Some further work is
o  WT Partnership (final, adjusted): $2606 mil ion
recommended to fil  a number of ‘gaps’ in this estimate and to provide a higher level
Our understanding is that the last of these figures has been
of confidence.
adopted as the ‘definitive’ estimate at this stage of project
development.
2.  Bus option
• For the alternative (bus) option, it appears that a literal  ‘Do  • The peer review understands the approach taken by the project team of assuming a
Minimum’ approach has been taken, which involves no significant
literal ‘Do Minimum’ option against which the LRT option is compared. However, we
infrastructure costs.
would note that, if it were to be decided to continue to operate buses as the main
mode in the corridors concerned over the next 30 years, then undoubtedly some
significant infrastructure costs would be appropriate to enhance conditions for bus
users.
Note: (1) It is possible that the project team has made an al owance, which has been netted off in deriving the above resource estimates, but this is not apparent from the documentation
provided.

IWA/N207/Rep/1552
26 July 2015
6

Ian Wal is Associates Ltd
1.  INTRODUCTION
This paper has been prepared for Auckland Transport by consultants Ian Wallis Associates (IWA), as
part of a peer review of AT’s proposals for a light rail network for Auckland.
The paper is one of three papers prepared by IWA for this peer review.
This paper (C) is concerned with cost estimates.  Specifically it addresses the following aspect of the
peer review task:
“Whether the estimates of incremental public transport capital costs (infrastructure
and vehicles) and operating costs, relative to the bus-based base case, are sufficiently
robust for this stage of option assessment”.
It reviews the cost estimates prepared by the AT project team for the LRT option and the alternative
bus option, under three main cost groups, ie:
•  Operating costs
•  Vehicle capital costs
•  Infrastructure capital costs.
For each of the cost groupings, the review examined whether the project team cost estimates are
sufficiently robust at this stage of the project assessment, by comparing the estimates against cost
data for operational LRT schemes in NZ and Australia.  Where the cost estimates appear inconsistent
with this evidence, we provide estimates, at an indicative level, that appear more appropriate.
The other two papers in this peer review set cover the fol owing aspects of the AT proposals:
Paper A:  Robustness of the deficiency analysis.  This paper focuses on demand and capacity analyses
regarding the maximum number of buses that can practically and reliably be accommodated on main
routes in the isthmus and city centre.
Paper B:  Options and assessment.  This paper examined the reliability of the options analysis results
in terms of the achievement of the study objectives, including whether all viable options have been
considered and the appropriateness of the multi-criteria analysis (MCA) assessment methodology.

IWA/N207/Rep/1552
26 July 20147
7

Ian Wal is Associates Ltd
2.  TASK STRUCTURE & METHODOLOGY
[This chapter to be reviewed.]
2.1.  Overview of LRT option
The new LRT system key parameters assumed included:
•  33.64km double track (32.06km used in OPEX model)
•  54 LRV (49 peak), each vehicle capacity 300 passengers
•  35 substations
•  43 stops
•  1 initial temporary depot in phase 1
•  1 permanent depot developed phase 2
The introduction of the LRT system wil  result in a restructure and reduction in urban bus services
resulting in the removal of 254 peak buses (279 total buses including spares) of 50 passengers capacity
each.
2.2.  Task Structure
This task is split into the following work streams:
•  OPEX (Operational)
•  CAPEX (Capital)
•  Vehicle CAPEX
OPEX for Bus and LRT option incorporates the fol owing costs:
•  Staff costs (all)
•  Depot rental, rates, taxes, cleaning, maintenance, repair, utility costs (both for Bus and LRT)
•  Route infrastructure (track, overhead, substations, points, structures, cleaning, maintenance,
repair, renewal, utilities etc. for LRT)
•  Control centre  (Operations and power control centre –  LRT only) maintenance, repair,
cleaning, renewal, utilities etc.
•  Light Rail Vehicle / Bus maintenance, renewal, repair, renewal, power
CAPEX for LRT incorporates the following costs:
•  Depot construction and commissioning costs
•  Control Centre construction and commissioning costs (possible to be incorporated within
Depot)
•  Route Infrastructure construction and commissioning
•  Structure such as tunnels, bridges, culverts etc.
Land procurement of all land required for Depot, substations and route alignment as well as any work
required to prepare for construction work such as addressing environmental issues (if any) Vehicle
CAPEX incorporates:
•  Light Rail Vehicle / Bus procurement, commissioning, depreciation, finance, lease costs
NB: Light Rail Vehicle Battery & “wire free” technology is assumed to apply to the LRT option
2.3.  Methodology
Methodology applied is based on:
•  Initial review of material provided by project team
IWA/N207/Rep/1552
26 July 2015
8

Ian Wal is Associates Ltd
•  Referencing the cost model and supporting material provided by the project team for the OPEX
and CAPEX (LRT) scope:
o  OPEX (Bus and LRT option): Key documents and model and in particular:
  Excel model “CCFAS2 costing sheet Option 4 Staging v4 (2-02-2015)”
  PDF “Costing Summary for Option 6, 8, 9A, 4 – Revision (1.2)”
  Q&A with project team as necessary
o  CAPEX (LRT option only): Key documents / model used including:
  Excel model “Draft Estimate – 22.12.2014”
  PDF “Report for Peer Review of Draft CAPEX Estimate” (Turner & Townsend –
23 December 2014)
  Various other reports provided
  Q&A with project team as necessary
o  Vehicle CAPEX:
  Extract holding and funding costs from OPEX model and shown separately
•  Based on experience of IWA team making comparisons to other systems (bus and light rail)
within Australia and New Zealand.
•  Using Auckland bus operator benchmark costs provide reworked indicative model for Bus
OPEX.
•  Note however that the proposed Auckland solution as per CAPEX model is a 100% “wire free”
system. Auckland are planning for a battery / recharging station solution. IWA are unaware of
other 100% similar (“wire free”) LRT systems in world. IWA in reviewing the costs within the
excel model developed relied upon advice from “supplier” experts. IWA also note that the
CAPEX peer review consultant (Turner & Townsend) identified the same issue (refer report for
“Peer Review of Draft Capex Estimate”, 23 December 2014).

IWA/N207/Rep/1552
26 July 2015
9

link to page 10 link to page 10 link to page 10 link to page 10 Ian Wal is Associates Ltd
3.  VEHICLE TYPES AND CAPACITIES
3.1.  Buses
3.1.1.  Project team assumptions
The project team considered two generic bus types, ie:
•  ‘Standard’ size buses, similar to the majority of current Auckland fleet, with c. 37 seats and an
effective planning capacity of 50 passengers.1
•  Double-decker buses, with c. 90 seats and an effective planning capacity of c. 100 passengers.2
The project team assumptions are set out on Table 3.1.
The standard size buses were assumed in al  the project team’s main analyses.
3.1.2.  IWA assessment
For the standard bus category, IWA reviewed planning capacity standards adopted for such bus sizes
in the main Australasian centres. Taking the Wellington fleet as an example, it was found that seating
capacity of ‘standard’ size buses varied between 37 and 51 seats, with standing capacity (@ 4
passengers/m2) varying between 15 and 28 passengers. On this basis, IWA considers that a planning
capacity figure of 50 passengers (average load/bus over AM peak 1-hour, peak direction, at point of
maximum loading), as adopted by the project team, is reasonable for current planning purposes for
this project.
For the double-decker bus category, again we consider that the project team’s assumption of a 110
(90 + 20) nominal capacity vehicle is reasonable.  On this basis, we assume a planning capacity of 100
passengers (ie including 10 standees, equivalent to 2 standees/m2 on the lower deck).
For the double-decker category, it appears that the project team did not consider this option in any
detail, and did not undertake any cost analyses (relative to the standard size buses).  Given the
passenger volumes anticipated in the relevant corridors, it would appear that the economics of
double-decker buses is likely to be very favourable relative to standard size buses, and this solution
would be more analogous to the LRT solution.  We have therefore analysed the operational and cost
implications of operating the main services in the corridors of interest with DD vehicles, at least in the
peak periods, and then compared the results with the ‘standard’ bus and the LRT options.
We have therefore used planning capacities of 50 passengers (standard buses) and 100 passengers
(DD) in our service analysis and cost modelling work.
3.2.  LRV
3.2.1.  Project team assumptions
The project team key assumptions on LRV, based on advice from Arup3, were as follows:
•  Vehicle length
45 metres (7 modules, 100% low floor)
•  Vehicle width
2.65 metres
•  Vehicle capacity
300 passengers, based on al  seats occupied plus standees @ 4/m2 of

available floor area4.

1 Project team specification (based on ADL Enviro 200 vehicles) is for buses with 37 seats and 18 standee spaces (@ 4/m2).
For service planning and costing purposes AT has taken average MLP capacity as 50 passengers.
2 Project team specification refers to a triple axle double-decker vehicle (12.8m) with capacity for 90 seats and 20 standees
@ 4/m2. We assume an effective planning capacity of 100 passengers for such a vehicle
3 Memo from Arup (Andy Wood) ‘LRV Assumptions for CCFAS2’, 3 Sept 2014.
4 Arup noted that an appropriate loading standard for structure design is 6 standees/m2.
IWA/N207/Rep/1552
26 July 2015
10

link to page 11 link to page 11 link to page 11 Ian Wal is Associates Ltd
All The project team assumptions are set out in Table 3.2.
3.2.2.  IWA assessment
We have no issues with the project team assumptions on LRV length and width.
The team assumption on vehicle capacity is based on 4 standees/m2, which is a standard often quoted
by LRV suppliers. It represents quite a crushed loading level (but less than the structural design load
of 6/m2 often used).  It is not realistic as a planning standard (average passengers/vehicle over peak 1
hour, peak direction) in Australasian conditions. Such a planning standard, which is relevant here, has
to allow for the variability in loading levels both through the peak (1 hour) period and within each
vehicle.
We reviewed the Australasian evidence on loading standards for planning purposes in current LRT
(and urban rail) systems, principally Melbourne (commonly quoted as the world’s largest tram/LRV
system). In summary, our findings were as follows:
•  The modern Melbourne trams are exemplified by the ‘E’ class5, which are 33 metres long, have
64 seats and have total planning capacity (peak 1 hour average at maximum load point) of 180
passengers. This represents 2.87 standees/m2  of available floor area  or 5.29 total
passengers/m length (with about 36% of total load being seated).
•  The ‘E’ class trams have higher planning load standards than the earlier (still operational)
Melbourne trams: the earlier trams (dominated by A, Z and B classes) average around 2.15
standees/m2 or 4.55 total passengers/m length (with about 62% of the maximum load being
seated).
•  In Wellington, the urban rail system adopts a nominal loading standard of 2.55 standing
passengers per m2 of standing area; but GW then suggests that the resulting total load figures
should in future be factored by around 0.85 to al ow for ‘loading diversity’6.  The outcome
would be for a planning load standard (peak 1 hour average) of wel  under 2.0 standees/ m2.7
•  For the Auckland urban rail system, advice received was that, for future service planning, a
ratio of 7 standees to 10 seated passengers (ie 59% of total passengers being seated) should
be adopted, and that this approximately equates to 2.5 standees/ m2.
•  We note that, as the typical passenger travel times on the AKL and WGN urban rail systems
are typically longer than is likely on the LRT system, this may justify a somewhat less generous
(higher) loading standard for the LRT.
Our conclusions on this aspect are that, for LRT service planning purposes at this stage:
(i)  A standard of 4 standees/ m2, as assumed by the AT project team, is not sufficiently generous
(and likely to be very unpopular with actual and potential users).
(ii)  Standards based on current operating practice for new trams in Melbourne would be more
appropriate. Based on the new E class trams, this would involve a standard of around 5.3
total passengers/metre length (and around 2.9 standees/m2 of available floor area).
(iii)  This would result in a capacity standard of about 240 passengers per 45m tram: this is a
reduction of some 20% from the 300 capacity assumed by the project team.
In the light of this conclusion, we adopt this revised standard in our subsequent cost assessment work.
Table 3.2 indicates (based on the Melbourne E trams) our approximate breakdown of the 240
passengers figure between seated (36%) and standing passengers.


5 About 17 E class trams are now in service, and PTV’s intention is that the entire fleet wil  comprise E class trams in future.
6 The current ‘loading diversity’ factor adopted in Wellington is, de facto, lower than 0.85.  It is expected that the 0.85 factor
wil  be appropriate when the more frequent peak services proposed are implemented.
7 GWRC ‘Wel ington Regional Rail Plan 2010 – 2035’. 2013 edition.
IWA/N207/Rep/1552
26 July 2015
11

Ian Wal is Associates Ltd
TABLE 3.1: BUS TYPES AND PLANNING CAPACITIES
Project team
IWA assessment
IWA comments
Item
Standard  Double-decker(1)
Standard
Double-decker(1)

Length (m)
11.0
12.8
11.0
12.8

Seats
37
90
37
90

Standing area (m2)
4.5
5.0
4.5
5.0
Lower deck only.
Standee rate/m2
4
4
4
4
Standees not allowed on DD top deck
Standee rated
18
20
18
20

capacity
Standee planning
13
na
13
10

capacity
Total planning
50
na
50
100

capacity
Seated %
74%
--
74%
90%

Note:
(1)  Project team work [AT51] identified two distinct types of DD buses: (i) 12.8m triple axle BRT bus - 90 seat + 20 standing capacity; and (i ) 11.0m twin axle FTN bus – 62 seats + 15
standing capacity.  Current Ritchie’s AKL DD buses are Volvo, Malaysian body, 12.1 metres length, 86 seats.

TABLE 3.2: LRV TYPES AND PLANNING CAPACITIES
Item
Project Team
IWA Assessment(1)
IWA comments
Length (m)
45.0
45.0
Noting comments by Arup, appears to be a reasonable assumption at this stage, giving a
sensible balance between vehicle size and service frequency.
Seats
?
87

Standing area (m2)
?
52.6

Standee rate/m2
4.0
2.9

Standee planning capacity
?
153

Total planning capacity
300
240

Seated %
?
36%

Notes:
(1)  IWA figures scaled up from the Melbourne E class trams (which are 33.0 m length).

IWA/N207/Rep/1552
26 July 2015
12

Ian Wal is Associates Ltd
4.  VEHICLE CAPITAL COSTS
4.1.  Bus Capital Costs
4.1.1.  Project team estimates
The project team estimates for buses considered a ‘standard’ size bus (c. 40 seats) and a double-decker
bus, and assumed:
•  An effective (economic) life of 12 years in both cases.
•  An initial purchase cost of $500,000 (standard) and $650,000 (double-decker).
•  A ‘half-life’ renewal cost of 20% of the new cost in each case.
•  The capital costs would be translated into annual capital charges, based solely on the historic
cost depreciation over the life of the vehicles.
These assumptions are set out in Table 4.1. The result is annual capital charges per bus of $50,000pa
for the standard bus size, and $65,000 for the double-deckers (ie 10% of the initial purchase price in
each case).
4.1.2  IWA appraisal
Our review of the project team bus unit capital cost estimates found that:
•  The effective life of buses should be taken as 20 years, consistent with the PTOM contract
specification and the NZTA Requirements  for Urban Buses  in New Zealand (RUB) and the
practices adopted in most other regions.  The project team 12 year assumption is inconsistent
with practices commonly adopted for urban route service buses in both NZ and Australia.
•  The new prices assumed by the project team are on the high side for the standard (c. 40 seat)
buses. Our best estimate is $440,000 for standard buses (based closely on bus prices paid by
AKL operators in recent years).  For double-decker buses, there is limited price data available
in NZ: at this stage, we adopt the project team estimate of $650,000 as appropriate.
•  As noted above, the project team derived an annual capital charge per bus based solely on
the historic cost depreciation from the new price over the assumed life.  We do not consider
this approach is appropriate as it makes no allowance for the economic cost of capital or
future inflation in bus prices.  Our preferred approach (adopted quite widely elsewhere) is to
estimate the capital charge on an annuity (mortgage) basis over the bus life, giving similar
annual payments to those resulting if the buses were financed through a typical finance lease
arrangement. The result is for a constant annual charge over the life of the bus of
approximately 11% of the new bus price.
•  As it happens, due to off-setting factors, our assessment of annual capital charges per bus is
not very different from that of the project team: the project team figures were 10% of their
bus purchase price (but not adjusted for inflation), ours equate to about 11%.
Our unit cost estimates per bus per year are set out in Table 4.1, for comparison with the project team
estimates.
4.2.  LRV Capital Costs
4.2.1  Project team estimates
As noted earlier (section 3.2), the project team assumed 45 metre length (2.65 metre width) vehicles
(and we accept this is a reasonable assumption at this planning stage).
The vehicle capital cost estimate used in the project team costing model was $6.8 million, with an
assumed life of 30 years.  We also note that the WT Partnership estimated a unit cost of $6.13 million,
IWA/N207/Rep/1552
26 July 2015
13

link to page 14 link to page 14 link to page 14 Ian Wal is Associates Ltd
based on a similar vehicle specification (this figure was not adjusted in the light of the Turner &
Townsend peer review of the infrastructure costs).
We note that, for their costing model, the project team assumed:
•  A cost allowance for vehicle ‘renewal’ after 15 years, of 20% of the initial vehicle costs (ie
$1.36 million)8
•  Straight line historic cost depreciation over the 30 year life, equating to $227,000pa per
vehicle (plus $41,000pa for the half-life renewal).
These figures are shown in Table 4.2.
4.2.2  IWA appraisal
There are significant difficulties in estimating LRV capital costs for situations such as this, for the
following reasons in particular:
•  Most of the more recent LRV procurements in Australia have covered both vehicle supply and
maintenance, with the result that the supply component of the costs is not separately
apparent.
•  The suggested AT proposals are for a ‘wire free’ LRT system.  However, to our knowledge,
there are no such systems of significant length operating anywhere in the world, and thus very
limited  information on both  relevant vehicle costs and operational experience. Turner &
Townsend also noted this difficulty in their Capex peer review. IWA has therefore had to rely
on advice from selected suppliers as to the vehicle capital costs applicable in this situation.9
•  Unit vehicle costs can be relatively sensitive to the size of the order, the maximum network
grade and whether the vehicles are single- or bi-directional.
We have discussed likely costs for 45m LRVs for Auckland with a number of potential suppliers, and
drawn the fol owing conclusions:
•  Typical prices for ‘standard’ 45m LRVs involving overhead wire operation from European/
American suppliers are in the range A$4.2 – A$5.0 million per unit.
•  These figures are based on LRVs of generic design that are fitted in to production runs of the
selected supplier; assume standard infrastructure configuration; are designed for bi-
directional operation; and involve a minimum order of 10-15 units.
•  For battery operations, costs would increase to about A$4.8 – A$5.6 million (two batteries
would be required @ c. $0.3 million each)10.  This translates (at a conversion rate of 1.10) to
about NZ$5.3 - $6.2 million.
We also note that:
•  These costs would reduce by c. 5% if bi-directional operation is not required.
•  LRVs from Chinese or East European (eg Poland) sources are likely to cost in the order of 25%
less than the above figures, without significant loss of quality.
•  At any particular time, discounted or second-hand LRVs may be available (eg Adelaide
purchased Alstom trams at a significant discount).

8 The project team model sheet [AT44.2] states that the renewal al owance is 20% of the original cost (which should be $1.36
mil ion). However, the figure shown in the cost spreadsheet is actual y $1.226 mil ion, which is about 18% of the original
cost.
9 We note suggestions that, while wire-free LRVs may be more expensive than comparable ‘catenary’ LRVs, the additional
vehicle costs may wel  be more than compensated for by the savings in establishing and maintaining an overhead wire
network.
10 It is not clear whether there would be significant off-setting cost reductions if pantographs are not required.
IWA/N207/Rep/1552
26 July 2015
14

link to page 15 Ian Wal is Associates Ltd
•  Currently batteries need to be replaced after c. 7 years. However, battery technology is
developing rapidly, with increased battery life and reduced costs.
Based on the above, we consider that the $6.8 million figure used in the project team’s costing is on
the conservative side, even assuming wire-free operation, and a reduction in the order of 10% would
appear more appropriate11.  For convenience, we adopt the WT Partnership figure of $6.13 million in
our further analyses.


11  An additional reduction in the order of 10% could apply in the case of wired operation, although increased costs would
be expected in other areas (eg maintenance of overhead wiring system).
IWA/N207/Rep/1552
26 July 2015
15

Ian Wal is Associates Ltd
TABLE 4.1: BUS UNIT CAPITAL COSTS
AT Team Assumptions(1)
IWA Estimates
IWA Comments
Item
Bus Type
Bus Type

Standard
Double-decker
Standard
Double-decker

Economic life (years)
12 years
12 years
20 years
20 years
Standard economic life of NZ urban buses is 20 years (as in PTOM contracts, NZTA
Regulations for Urban Buses – [check]
Capital costs ($000)

Initial purchase
500
650
440
650
‘Standard’ bus prices based on recent prices being paid (or estimated) by AKL bus

operators for 40-49 seater buses. Double-decker bus price estimates from other
information (limited data available).

Half-life renewal (20%)
100
130
66
98
‘Half-life’ renewal would not usual y apply to 12 year life; would generally apply to 20
year life, but less costly than AT assumption (have guestimated at 15% of purchase
price).
Annualised charge –

AT figures al ow for (historic cost) depreciation only. More appropriate estimates
depreciation only ($000)
would take account of the opportunity cost of funds, equivalent to the costs under a
finance lease arrangement or similar. On this basis, our estimate of the annual charge

Initial purchase
41.7
54.2

over a 20-year bus life is about 11% of the capital expenditure (ie $440k/$650k for the

Half-life renewal
10.8
14.0

two bus sizes).

Total annual
52.5
68.2
45.7
67.5
Note: (1) As in CCFAS2 Costing sheet, Assumptions (IWA 44.2). These figures derived from AT email chain 13 Nov – 11 Dec 2014 (IWA 44.3).

TABLE 4.2: LRV UNIT CAPITAL COSTS
Item
AT Team Assumptions(1)
IWA Estimates
IWA Comments
Economic life (years)
30 years
30 years
Arguably the effective life might be extended to 35 years (maybe more consistent with
the bus life assumption of 20 years).
Capital costs ($000)

Initial purchase
6800
6130
The project team $6.80 mil ion figure is as used in the AT costing analysis: an alternative

Half-life renewal (20%)
1226
1226
figure proposed (WT Partnership) was $6.13 mil ion.
Annualised charge –

The AT figures are based on straight line depreciation on the historic cost. IWA figures
depreciation only ($000)
are based on an annuity calculation, which would more closely reflect the annual costs
[To add]
under a finance lease (or PPP) arrangements.

Initial purchase
227

Half-life renewal
41

Total annual
268
528.5
Note: (1) As in CCFAS2 Costing sheet, Assumptions [IWA 44.2].

IWA/N207/Rep/1552
26 July 2015
16

link to page 17 Ian Wal is Associates Ltd
5.  OPERATING STATISTICS
Table 5.1 provides a summary of the net changes in annual operating statistics in the corridor of
interest as a result of introducing the full LRT scheme proposed, deleting the directly-competing bus
routes, and introducing a set of feeder bus routes to link with the southern ends of the LRT lines:
•  The project team estimates of the change in bus operating statistics appear to have been
made by just estimating the statistics for the ful  length of the present bus routes using the
corridor: no allowance appears to have been made for the provision of new feeder bus routes
to the LRT from the area currently served by bus routes south of the LRT termini.12  Clearly
this will overstate the net bus operations and cost savings resulting from the LRT proposals.
•  The IWA estimates have been specifically calculated as the difference between (i) the ‘base
case’ (no LRT) bus services in the corridor; and (ii) the residual feeder bus etc services if the
LRT is implemented. We thus consider that these estimations for ‘net bus services removed’
are more realistic than the project team figures.
Table 5.1 shows the IWA operating statistics estimates both for the ‘standard’ bus case (ie comparable
with the project team figures) and for the ‘double-decker’ case.  For the DD case, we have adjusted
the peak period frequencies on the main bus routes in response to the doubling of practical capacity
per bus (from 50 to 100 passengers), and hence adjusted the PVR, service hours and service km
statistics accordingly.
The table also shows the project team estimates for LRT operating statistics and two sets of IWA
estimates:
•  Based on the AT planning capacity of 300 passengers/vehicle, but with slight modifications to
optimise the service plan (including reducing the PVR).
•  Based on our suggested planning capacity of 240 passengers/bus (refer section 4.2.2). This
results in a c. 25% increase in peak service levels (trams/hour) and corresponding increases in
service hours and service km at peak periods.
The information given in Table 5.1 is used subsequently for the operating costs (including vehicle
capital charges) assessment in chapter 6.


12  For example, refer project team document CCFAS2 Costing sheet option 4 staging v4 (2-02-15) [IWA 44.1]
IWA/N207/Rep/1552
26 July 2015
17

Ian Wal is Associates Ltd
TABLE 5.1: ANNUAL OPERATING STATISTICS – Project team and IWA estimates
Change in Annual Operating Statistics (2046)

Comments
PVR
Service Hours(1)
Service Km(2)
Route Km
BUSES

AT project team:

Ex AT CCFAS2 costing sheet option 4 staging v4

Net bus services removed
254
723.9
10,921
Assumes 50 pax/bus (planning standard)
IWA review – single decker

Assume 50 pax/bus (planning standard), as AT

Services removed
352
993.3
14,991

Services added
63
185.4
2,733

Net services removed
289
807.9
12,258

IWA review – double decker

Assumes double-deckers (except Queen Street) - assumes 100

Services removed
184
554.4
8039

pax/bus

Services added
63
185.4
2733

Feeder services – assumes stil  use single decker buses.

Net services removed
121
369.0
5306

LRT

Ex AT CCFAS2 costing sheet option 4 staging v4
AT project team (300 cap)
49
181.0
2935
32.06
Assumes 300pax/vehicle (planning standard)
IWA review – double-decker

300 cap
46
168.7
2772
33.1
As AT

240 cap
60
182.7
2977
33.1
Assumes 240 pax/vehicle (consistent with MEL standards).
Notes:
(1)  Includes al owance of 10% for layover between trips.
(2)  No al owance included for dead (out-of-service) running.

IWA/N207/Rep/1552
26 July 2015
18

link to page 19 Ian Wal is Associates Ltd
6.
OPERATING COSTS
6.1  Bus operating costs
6.1.1  Unit cost estimates – project team
The unit bus operating cost function (standard size buses, excluding capital charges) adopted by the
AT project team is as follows (assumed in 2014/15 prices)13:

$2.10 * service km + $41 * service hour.
Further details and comments are provided in Table 6.1. We note in particular that:
•  The basis for the figures is not well documented: some documentation is provided in an AT
email chain, but the above cost rates (as applied) differ from those in the emails.
•  It appears that the email rates are taken more-or-less directly from variable cost rates in the
AT operator contracts.  It is not clear to what extent these rates reflect ‘marginal’ costs or
whether they include a full allocation of overhead costs.  In any event, no separate allowance
has been made for operating overhead costs, which for AKL contracted bus operators
comprise around 24% of total operating costs.
•  The project team rates are substantial y  higher than AKL operator direct costs for the
corresponding cost categories – around 70% higher per service km and c. 20% higher per
service hour.

TABLE 6.1: UNIT BUS OPERATING COSTS (STANDARD BUSES) – PROJECT TEAM ESTIMATES AND COMMENTS
Cost Category
Adopted Rate
Sources
IWA Comments
Bus R&M,
$2.10/service km  AT emails (AT44.3) give figure of
This rate is very much (c. 70%)
fuel, RUC
$2.00. Stated that this is based on  higher than recent rate
variable rates in AT bus operating  information on direct operating
contracts. Basis for adjustment
costs of AKL contracted bus
to $2.10 not clear.
services, but it may include an
allocation of (indirect)
overheads.
Drivers
$41/service hr
AT emails (AT44.3) give figure of
This rate is substantial y (c. 30%)
$30, which is said to include
higher than recent rate
driver (direct?) overheads. Stated  information on direct operating
that this is based on variable
costs of AKL contracted bus
rates in AT bus operating
services, but it may include an
contracts. Basis for adjustment
al ocation of (indirect)
from $30 to $41 is not clear.
overheads.
Other
--
--
General operator overheads
operating
typical y comprise around 24% of
costs
total costs (excl. cap charges) for
(overheads,
AKL bus operators.  While AT has
etc)
not included these separately, it
appears most likely they have
been allocated across the above
cost categories


13  Source: AT email chain Bus opex CCFAS2, 13 Nov – 11 Dec 2014 [AT44.3]
IWA/N207/Rep/1552
26 July 2015
19

Ian Wal is Associates Ltd
6.1.2
Unit cost estimates – IWA review
For a bus operation reflecting the operating conditions (with an average service speed of c. 16km/hr)
in the corridors of interest, we have compared the results of applying the project team cost
assumptions with:
(i)  IWA’s best estimates of unit costs for AKL bus operators (drawn from various confidential
information sources); and
(ii)  IWA’s estimates of ‘efficient’ NZ bus operator unit cost rates (drawn from IWA’s NZ bus
operating costs database).
Our main findings and conclusions for ‘standard’ size buses are:
•  The project team operating costs are c. 5% higher than IWA’s best estimates of actual AKL bus
operator costs over recent years.
•  There would appear to be potential for current AKL bus operator costs to reduce by up to c.
20%, in order to reach efficient NZ bus operator cost rates.  It is not clear at this stage to what
extent this reduction wil  be achieved in the forthcoming tender/negotiation rounds (under
the PTOM regime).
•  On this basis, we suggest that (for standard size buses):
o  For an improved costing assessment (based primarily on current AKL bus operator
costs), a 5% reduction in the project team unit operating costs should be made.
o  As a sensitivity test, a further reduction in unit operating costs of 20% (to meet
efficient NZ cost standards) should be applied.
6.1.3  Total bus operating cost comparisons (‘standard’ buses)
Based on the above, Table 6.2 presents a summary of bus costing estimates (for the full stage 5
scheme). It provides estimates of annual operating statistics for two operating cases:
•  Project team case (as documented).
•  IWA review case.  These are somewhat (c. 12%) higher than  the project team estimates,
despite adjusting for the feeder bus requirement.
Unit operating costs (including bus capital charges) are provided for three cases:
•  Project team case (as documented, discussed above).
•  IWA typical AKL operator case – taken as 5% less overal  than the project team case (refer
above).
•  IWA ‘efficient’ case – taken as 20% less overall than the typical AKL operator case (as discussed
above).
Key features of these results include the fol owing:
•  Our estimate of the net annual bus operations costs (including vehicle capital charges) based
on the project team operating statistics and unit cost rates is $67.28 Mpa, which is effectively
identical to the project team figure ($67.26M) –as would be expected.  (It should be noted
that these figures, as for the other results given in this section, represents the net cost for bus
services in the corridor of interest, ie it is the costs for a bus option in the corridor less the
costs that would be required for feeder bus operations for the LRT option.)
•  This figure reduces to $62.75  Mpa on the basis of IWA’s typical AKL unit bus cost rates
(involving a 5% reduction in unit operating costs from the AT figures, and 13% lower unit bus
capital charges).
•  But the total figure would be higher, at $70.45 Mpa, using the IWA AKL unit costs and IWA’s
higher estimate of operating statistics.
IWA/N207/Rep/1552
26 July 2015
20

Ian Wal is Associates Ltd
TABLE 6.2: BUS OPERATING COST SUMMARY – PROJECT TEAM & IWA REVIEW. (Al  figures relate to ful  (stage 5)
proposals, on annual basis)
Item
Unit Costs
PVR(4)
Service
Service Km
Totals
Hrs(1) (000)
(000)
$Mpa
A. ‘STANDARD’ BUSES

Operating statistics
AT project team:

Net services removed

254
723.9
10,921

IWA review:

Services removed

352
993.3
14,991

Services added

63
185.4
2,733

Net services removed

289
807.9
12,258

Unit cost rates
AT project team
52,500
41.00
2.10

IWA review - typical AKL
45,700
38.95
2.00

IWA review - efficient
45,700
31.16(7)
1.60(7)

Aggregate costs ($Mpa)

Project team
Project team
14.67
29.68
22.93
67.28
Project team
IWA review - typical AKL2
12.77
31.47
24.52
62.75
IWA review
IWA review - typical AKL
14.53
31.47
24.52
70.45
IWA review
IWA review - efficient AKL
14.53

59.27
B. ‘DOUBLE DECKER’ BUSES

Operating statistics

IWA review:

Services removed(5)

184
554.4
8039

Services added(6)

63
185.4
2733

Net services removed

121
369.0
5306

Unit cost rates
IWA review - typical AKL
67,500
40.90
2.59

(double-decker)
IWA review - efficient
67,500
32.72
2.07

Aggregate costs ($Mpa)

IWA
IWA - typical AKL

  Services removed
13.66
22.67
20.82
57.15

  Services added
4.68
7.58
7.08
19.34

  Net services removed
8.98
15.09
13.74
37.81
IWA
IWA – efficient rates

  Services removed
13.66
18.14
16.64
48.44

  Services added
4.68
6.07
5.66
16.41

  Net services removed
8.98
12.07
10.98
32.03
Notes:
(1)  Includes allowance for terminal time, at 10% of timetable time.
(2)  Taken as 5% off project team estimate (refer text), but excluding capital charges
(3)  Taken as 20% off IWA typical AKL figure (refer text), but excluding capital charges
(4)  Figures given relate to peak vehicle requirement (PVR). These are increased by 10% to al ow for spare buses.
(5)  These services are DD operation
(6)  These services are assumed SD operation.
(7)  Relative to the line above.

IWA/N207/Rep/1552
26 July 2015
21

link to page 22 Ian Wal is Associates Ltd
•  This total would then reduce, to $59.27  Mpa, based on IWA’s efficient cost rates (unit
operating costs 20% lower than the IWA AKL current unit costs).
In summary, we conclude that the project team estimate of net costs for the (standard) bus
option of about $67.3 Mpa, is reasonably robust at this stage of the project planning process14:
the range of estimates we have made are all within +5%/-12% of the project team’s net total.
6.1.4  Implications of larger (double-decker) buses
As an important sensitivity test, we have examined the implications in the bus option of deploying
larger buses on the main corridor (line haul) bus services, while retaining ‘standard’ size buses on the
bus feeder routes proposed under the LRT option.  Our assumption is that the ‘larger’ buses would
comprise double-decker buses, with about 85-90 seats and an effective capacity for planning purposes
of c. 100 passengers (refer section 3.1).
We have also estimated (but without the benefit of detailed information) that the capital costs for
double-decker (DD) buses would be 30% higher than for ‘standard’ buses (this is consistent with the
project team assumption), and unit costs/service km would be 30% higher and driver costs per service
hour would be 5% higher than for standard buses (the project team does not make estimates of any
operating cost differences).
The results of our sensitivity tests, based on these estimates, are given in the lower part of Table 6.2.
It is seen there that:
•  Based on IWA’s estimated operating statistics and the DD-specific passenger capacity and unit
cost factors (starting from IWA’s AKL unit cost figures), the net bus operating cost would
reduce from $70.5 Mpa for the standard buses to $37.8 Mpa with DD buses on the line haul
services.
•  Similarly, based on IWA efficient costs, the reduction would be from $59.3 Mpa to $32.0 Mpa.
•  The main factor ‘driving’ these cost reductions is the doubling of effective capacity/bus (with
the planning standard for a DD bus of 100 being twice that for a standard bus of 50).
•  This saving is only partially off-set by the increased unit costs for DD buses – which average
around 22% per bus service km.
•  The favourable result also reflects the net contribution of line haul services (which would be
deleted under the LRT option) and feeder services (which would be added).
These results would indicate a strong case for further work to be done to evaluate a DD-based solution
in more depth.
6.2  LRT operating costs
6.2.1  Unit cost estimates – project team
The AT project team formulated an LRT operating cost model (excluding LRV capital) based essentially
on three cost ‘drivers’, ie:
•  Service km
– which is the ‘driver’ of vehicle power and R&M costs.
•  Service hour   – the ‘driver’ of tram operations (essentially driving staff) costs.
•  Route km
– the ‘driver’ of all infrastructure maintenance and renewals.
The first two components are consistent with the approach taken in the project team’s bus costing
model; while the third component is essential y replaced in the bus costing model by RUC, which is
there treated as variable with service km.
Table 6.3 sets out further details of the components of the project team’s LRV cost model.

14 In our view, this apparent ‘robustness’ in qualitative terms is partly the outcome of lower operating statistics and higher
unit costs than our estimates.
IWA/N207/Rep/1552
26 July 2015
22

link to page 23 Ian Wal is Associates Ltd
We consider that the structure of the project team’s cost model is reasonable at this stage of the
planning process, but we consider the work undertaken to formulate the chosen rates is generally
inadequate and lacking robustness. The main basis for the rates appears to be a note by Arup, based
on some 1999 UK cost rates, which have then been inflated (by UK CPI statistics?) and converted to
NZ$ figures (using standard exchange rate information).15  There appears to have been no attempt to
make  any use of more recent Australian data sources, either for LRT systems already operating (eg
Melbourne) or for systems currently in the development stage (eg Sydney, Canberra): this is surprising.
IWA was unable to review the project team unit cost estimates on a component-by-component basis,
as minimal information was provided on the basis of their derivation. However, we have provided
some brief comments on each of the project team components in the RH column of Table 6.3.
TABLE 6.3: UNIT LRT OPERATING COSTS – PROJECT TEAM ESTIMATES AND COMMENTS
Cost Category
Adopted Rate
Sources
IWA Comments
Tram R&M
$1.10/service km  AT44.2. Based on Arup advice
Based on Australian
(GB 1999 data) adjusted for
benchmarking tram R&M ranged
inflation and currency
between A$2.00 and $2.70 per
conversion.
service km (2013/14 rates)
Tram power
$2.31/service km  AT44.2. Based on power
Benchmark power consumption
consumption 7 Kwh/km, price
rates from Australian evidence
30ȼ/kwh, plus al owance of 10%
ranged between 5.6 Kwh/km to
for station and depot power.
6.8 Kwh/km for Tram. Thus
consumption estimate used is
reasonable.
Tram staffing
$49.20/service hr  AT44.2. Assumed as for bus plus
Australian tram driver
(drivers)
20% uplift.
benchmark (2013/14) rates of
drivers ranged between $53 and
$64 per service hour.
Network
$66,000pa/double  AT44.2. Based on Arup advice
Benchmarked infrastructure
infrastructure
track route km
(GB 1999 data) adjusted for
maintenance cost per track km
maintenance
inflation and currency
varied between $100k pa and
conversion.
$200k pa (inclusive of renewal).
Most benchmarked costs are
Network
$24,500/double
AT44.2. Based on Arup advice
based on operating contracts
infrastructure
track route km
that c. 37% of network
below 20 years.
renewals
infrastructure maintenance.

6.2.2  Unit cost estimates – IWA review
Our approach to reviewing the project team’s LRT unit operating cost estimates was essentially to
derive independent estimates of operating cost rates, based largely on data from existing and planned
Australian LRT systems, and making some allowance for cost differences between Australian and New
Zealand cost environments where appropriate. We note that our costs for most of the Australian
systems considered were based on data from experienced international LRT operators. Rather than
trying to base our assessment on the project team’s various cost ‘driver’ categories, we took a more
aggregate approach, using operating costs/service km as our primary cost unit.  We found that the
range of average costs (per service km) for the Australian systems examined was quite wide, between
about A$16 and A$26 per service km (excluding LRV capital-related costs). On consideration of the
data, the systems covered and the environment of the AKL project, IWA’s ‘best estimate’ of likely costs
for the AKL project is around NZ$22 (2014/15) per service km.  We note that a fairly wide level of
confidence applies to this estimate at this stage: it should be possible to significantly refine this

15  The tram driver unit rate was an exception to this, being based directly on the project team unit rate for bus drivers (per
service hour), with a 20% assumed increase.
IWA/N207/Rep/1552
26 July 2015
23

link to page 24 link to page 24 Ian Wal is Associates Ltd
estimate, and its confidence interval, through a more detailed benchmarking exercise involving
selected Australian LRT systems.16
Our ‘best estimate’ figure of $22/service km may be compared with the project team figures, which
result in an average cost of $7.43/service km17 (ie only about one-third of our estimate).
6.2.3  Total LRT operating cost comparisons
Based on the above, Table 6.4 presents a summary of LRT costing estimates for the ful  (stage 5)
scheme – these estimates cover the LRT service costs only, and exclude the associated feeder bus
services (which were covered in the earlier section).
The table provides estimates of annual operating statistics for three cases:
•  Project team estimates (as documented) - based on practical vehicle (45m) planning capacity
of 300 passengers.
•  IWA adjusted estimates, based on re-working of the proposed service plan (slightly lower
operating statistics), with 300 passenger planning capacity.
•  IWA ‘improved standards’ estimate, with passenger planning capacity reduced to 240 per
vehicle (45m) resulting in about 30% increase in PVR and about 8% increase in service hours
and service km.
Two sets of unit costs have also been applied:
•  Project team unit costs, as applied in their cost assessments.
•  IWA average cost estimate, with operating cost of $22/service km as noted above plus a
vehicle capital charge estimated on an annuity basis.
The lower section of the table sets out annual LRT operating costs (including vehicle capital charges)
based on various combinations of the three sets of operating statistics and two sets of unit costs.  Key
findings are summarised as follows:
•  Based on the project team LRV capacity assumptions, operating plan and unit cost rates, the
annual cost is $36.24 Mpa, which is virtually identical to the team’s own estimate ($36.25
Mpa), as would be expected.
•  However, when the IWA unit cost rates are applied, major increases in this estimate result,
reflecting IWA’s much higher estimates for both unit operating costs and  vehicle capital
charges.
•  The IWA best estimate, based on the vehicle capacity standard of 240 passengers and IWA’s
best estimate unit cost figures, is for an annual LRT cost of $100 M. This would reduce to $88M
if the 300 passenger capacity standard (as adopted by the project team) were to be applied.


16  The indicative cost estimate range of between $16 and $26 per km can be for a variety of reasons including the difference
between industrial awards (which are state based in Australia), dead running differences, fleet size and age, infrastructure
design and age, average speed and whether the operator is responsible for ticket sales and revenue protection activities
amongst others.
17  $21.804 Mpa for 2.935M service km: refer AT CCFAS2 costing sheet option 4 staging v.4 (02.02.2015) [Ref AT 44.2]
IWA/N207/Rep/1552
26 July 2015
24

Ian Wal is Associates Ltd
TABLE 6.4: LRT OPERATING COST SUMMARY – PROJECT TEAM & IWA REVIEW. (Al  figures relate to ful
(stage 5) proposals, LRT services only on annual basis)
Change in Annual Operating Statistics

Totals
PVR
Service Hours(1)
Service Km(2)
Route Km
(000)
(000)
AT project team:

  300 cap
49
181.0
2935
32.06

IWA review

300 cap
46
168.7
2772
33.1

240 cap
60
182.7
2977
33.1

Unit cost rates

AT project team
267.5
49.20
3.41
90,500

IWA review
528.5
--
22.00
--

Aggregate costs ($Mpa)

AT ops (300), AT UC
14.42
8.91
10.01
2.90
36.24
AT ops (300), IWA UC
25.90
--
64.57
--
90.47
IWA ops (300), IWA UC
26.74
--
60.98
--
87.72
IWA ops (240), IWA UC
34.88
--
65.49
--
100.37
Notes:
(1)  Includes al owance of 10% for layover between trips.
(2)  No al owance included for dead (out-of-service) running.

IWA/N207/Rep/1552
26 July 2015
25

link to page 26 link to page 26 Ian Wal is Associates Ltd
7
INFRASTRUCTURE CAPITAL COSTS
The draft cost estimate provided was developed and supported through a useful and detailed excel
model however no supporting detail was provided by WT Partnership that could be used to evaluate
the LRT CAPEX model basis and assumptions. A LRT system scope and specification paper was not
provided for peer review to support the model. The parameters sighted were high level that IWA
considered was not ideal for a subsequent peer review process.
A peer review was previously carried out by Turner & Townsend on the initial CAPEX model prepared
by WT Partnership. It appeared from reviewing this work that Turner & Townsend had direct access
to WT Partnership whereas IWA due to scope and time constraints of task relied upon materials
provided.
The Turner & Townsend review provided a materially different view on an appropriate CAPEX cost
estimate for the proposed LRT system.
IWA notes that the WT Partnership CAPEX model assumed a ‘wire free’ system.
Subsequently, after reviewing the peer review report IWA notes that WT Partnership provided an
adjusted cost estimate having taken into account the peer review feedback provided by Turner &
Townsend. This resulted in a significantly higher CAPEX than the original estimate but was still less
than the Turner & Townsend (peer review) estimate.  IWA is of the view that this adjusted estimate
($2,626m / $78.07m per route km) is a reasonable cost estimate at this early stage of the project.
7.1  Project team estimates
The draft CAPEX estimate model was prepared by WT Partnership and subsequently reviewed by
Turner & Townsend.  Table 7.1 provides a high level comparison outcome between the ‘initial’ WT
Partnership model and the Turner & Townsend estimate.
The version peer reviewed by Turner & Townsend was the initial version dated 2/12/2014. The
subsequent version (22/12/2014), adjusted after reflecting on the peer review paper, was materially
different at $2,626m (compared to the initial version $1,614m).18
The WT Partnership initial cost estimate (2/12/2014) for Option 4 was $1,614m whilst the peer review
(Turner & Townsend) provided a cost estimate of $3,305m. The initial cost of $1,614m was increased
by Turner & Townsend to $1,631m being the correction of a mathematical error within the initial cost
estimate model: Table 7.1 shows the initial cost estimate and the Turner & Townsend estimate.
Table 7.1 CAPEX comparisons: WT Partnership and Turner & Townsend review
Turner &
(NZ$000’s)
WT Partnership cost
estimate 2/12/14
Townsend peer
review
Direct construction costs

Track (33.64 km alignment)
352.1
211.5
Power (inc. 35 Nr substations)
76.2
285.6
Systems
19.8
90.8
Utilities
63.7
326.8
Roadworks
13.9
47.5
Stops (43 Nr)
54.9
59.4
Depot & Stabling Facility
74.4
105.8
Direct construction costs
655.0
1,127.4
Preliminaries
156.4
414.9
Traffic management
29.5
50.7

18  Both these figures have been adjusted to remove the LRV costs from the model.
IWA/N207/Rep/1552
26 July 2015
26

link to page 27 Ian Wal is Associates Ltd
Contractor’s design
26.2
202.9
Indirect construction costs
212.1
668.5
Total construction costs
867.1
1,795.9
Contractor’s OH&P
130.1
269.4
Total contractor’s costs
997.2
2065.3
Planning & Development
29.9
41.4
Client Design
29.9
60.8
Client Costs
20.0
92.5
Owners management costs total
79.8
194.7
Subtotal
1,076.9
2,260.0
Property acquisitions
282.3
282.3
Subtotal
1359.3
2,542.3
Risk and contingency
271.9
762.7
Subtotal
**1,631.2
3,305.0
Track km
33.64
33.64
D&C cost ($m) per km
48.49
98.25

Within the Turner & Townsend peer review report explanations were provided justifying the
differences in assessed costs. As well, within the Turner & Townsend report the presentation was in a
different format than provided by WT Partnership. For changes made subsequently by WT
Partnership, refer Table 7.2 in which they retained in the same format as initially presented.
Table 7.2 CAPEX comparisons: WT Partnership initial and updated costs As per draft estimate model
Al costs $m
WT Partnership initial cost
provided (WT Partnership
estimate (2-12-2014)
22-12-2014 updated cost
estimate)
Land & Property
282.3
281.97
Investigation & Reporting
29.9
39.39
Developed Design & D&C Monitoring
49.8
118.18
Design & Construct contract
362.0
439.54
Track (34 km alignment)
352.1
238.03
Power (inc. 35 Nr substations)
76.2
279.17
Systems
20.1
50.77
Statutory Undertakers Equipment
63.7
102.77
Highway Costs
13.9
105.84
Stops (43 Nr)
54.9
76.50
Support Facilities
74.4
79.85
Design
26.2
121.28
Programme & Project Management
185.9
316.03
Subtotal
867.5
1,370.24
Off site overhead and Profit
130.1
205.54
Total D&C
997.6
1,575.77
Contingency P50
254
610.84
Subtotal @4Q14 prices
1,613.6
2,626.16
Track km
33.64
33.64
D&C cost ($m) per km
47.97
78.07
**NB: Difference in Turner & Townsend initial cost estimate of $1,613.6 and $1,631.2 in above table (8.1) was due to the
original cost estimate having a mathematical error subsequently corrected.

IWA/N207/Rep/1552
26 July 2015
27

Ian Wal is Associates Ltd
Within the peer review report and the subsequent adjustments of the WT Partnership model (within
the Excel model) justification for differences assessed and subsequently made can be traced. IWA
makes no judgement on the peer review assessment and changes subsequently made other than to
compare to Australian indicative CAPEX costs and methodology used (refer section 7.2 following).
Note that if the WT Partnership adjusted estimated cost dated 22/12/2014 was used then the variance
between the peer review and their CAPEX would be a much closer outcome to the Turner & Townsend
peer review estimate.
Turner & Townsend advises that, based on their research and experience of planning for light rail
systems in Australia, the initial “draft cost estimate” WT Partnership provided is overly optimistic.  It
was recommended by Turner & Townsend that the draft estimate that they reviewed be examined in
more detail and supported by more evidence-based data.
The methodology used by Turner & Townsend in carrying out the peer review was advised to be based
on experience and benchmarking to other light rail systems they have judged to be similar. They also
established the following key principles:
•  Escalation – assumed to be 4th quarter 2014 and no provision for escalation from 1st quarter
2015 to commencement of revenue services.
•  Property acquisition – Includes land acquisitions, relocations, alterations / refurbishments /
demolitions and legal and other fees excluded from scope of their review thus used values used
within the draft estimate.
•  Existing utilities – Used an overall rate per km using based on ‘similar’ projects as well as
experience in other light rail projects (utility costs presented as a % of light rail infrastructure
and system total cost).
•  Trackform – Overall rate per metre for trackform from other projects used and compared to
draft estimate.
•  Rail alignment (structures) – Rate per metre for elevated structure compared to other similar
structures.
•  Stops – Cost based on similar stops in Sydney light rail project and others globally.
•  Precincts – Road works and footpath works between stops were assessed against typical %
these works ‘usual y form as part of overall scope’.
•  Bulk and traction power – Unit rates for supply and installation were applied to the element
quantities of substations, overhead wiring, small power and lighting from their ‘internal cost
database’.
•  Rail systems – Benchmarked rates from “internal cost database” for supply and installation of
communication and security systems, tram and traffic control signalling, combined services
route, passenger information systems and ticketing systems (applied to alignment length).
•  Depot and stabling – Rates per facility/building, including fit out, and an operations control
centre were compared to other similar projects on a total cost and cost per LRV basis.
•  Contractor’s indirect costs - The specific project requirement for preliminaries including the
core management and site based teams, temporary construction facilities such as construction
sites,  office  buildings  and  lay  down  areas,  insurances,  traffic  management and  main
contractor’s design costs were analysed as percentages of total direct construction cost and
benchmarked against comparable projects.
•  Project / owner’s costs - Requirements for external appointments and internal costs calculated
as a percentage of total construction costs and benchmarked against comparable projects
•  Risk and contingency - Based on the information available upon which the estimate has been
prepared,  together  with  the  assumptions  made,  direct  costs,  indirect  costs,  engineering,
procurement and management, the adequacy of the total risk allowance as a percentage of the
total cost was assessed and benchmarked against comparable projects.
IWA/N207/Rep/1552
26 July 2015
28

link to page 29 Ian Wal is Associates Ltd
The above methodology is a useful approach.  However in IWA view, since all light rail systems are
unique and have individual issues to be addressed, many resulting in material differences in cost, a
more appropriate approach would be to carry out a detailed analysis of the alignment and thus have
a more robust understanding and thus cost estimate. The WT Partnership appears to have taken this
approach although the materiality of changes made to the version (2/12/2014) reviewed by Turner &
Townsend and a subsequent version (22/12/2014) does raise questions.
The adjusted (22/12/14) WT Partnership cost estimate of $2,626m ($78.07m per route km) was much
closer to the Turner & Townsend cost estimate of $3,305m ($98.25m per route km) and because the
methodology fol owed was based on a detailed alignment assessment, in IWA view this is a more
credible basis to establish a cost estimate to use at this stage of the project.  Table 7.3 summarises the
cost estimates per route km for both WT Partnership versions and the peer review cost estimate.
Table 7.3 CAPEX summary of estimates
WT Partnership

(2/12/14) model
Turner & Townsend
WT Partnership

corrected
peer review
(22/12/14)
Project total CAPEX
$1,614m
$3,305m
$2,626m
Cost per Km (33.64k)
$47.97m
$98.25m
$78.07m
7.2  LRT design and construction cost comparisons (Australia)
IWA  undertook high-level comparisons of the project team capex estimates with corresponding
estimates for various recent and current Australian LRT systems.  CAPEX costs were sourced relating
to the Adelaide extension, Sydney extension, Sydney, Gold Coast Light Rail and ACT Light Rail systems.
These indicate a CAPEX build cost per route (double track) km in the range A$66m to $100m (excluding
vehicle CAPEX).  Extensions were materially less and thus are excluded.
IWA notes that the ACT Light Rail is currently in the process of being tendered and whilst the Gold
Coast Light Rail build has been competed it is understood within the industry that the D&C component
was wel  over budget Recent Sydney Light Rail tender cost (accepted) also significantly exceeded the
cost estimate developed by the project team. The above indicative costs incorporate estimated
overrun costs.
These comparatives tend to suggest that the Turner & Townsend estimated cost is high whilst the WT
Partnership ‘initial’ estimate was unrealistically low, but has subsequently been updated.
The cost estimate subsequently estimated by WT Partnership appears however to be a more credible
benchmark ($78.07m per route km) and broadly comparable to the Australian benchmarks (inflated
to reflect updated outcomes) although the impact on costs of a 100% “wire free” LRT system should
be reviewed further in order to provide a more robust estimate.
IWA nonetheless considers that based on comparison to LRT systems within Australia that the WT
Partnership adjusted cost estimate is broadly within IWA expectations.
7.3  Summary infrastructure capital findings
The methodology of WT Partnership and Turner & Townsend was different.  The WT Partnership
approach appeared to be based on a detailed assessment of the selected alignment whilst the Turner
& Townsend methodology was based largely on benchmarking and drawing on information from
within their “data base”. For a peer review process the peer review is a reasonable approach as was
the WT Partnership approach for their purpose.
The level of contingency used in the CAPEX model initially by WT Partnership was 20%. Turner &
Townsend used 30% (to be reduced as the design progresses). WT Partnership subsequently used 30%
in their reworked cost estimate.  IWA is of the view that contingency level at this stage should be larger
IWA/N207/Rep/1552
26 July 2015
29

Ian Wal is Associates Ltd
and points out the recent Australian light rail projects where the cost material y exceeded the cost
estimates and/or actual build costs. At this stage IWA are of the view that the 30% used by Turner &
Townsend may be too low. IWA notes that other similar Light Rail projects at this stage of development
provide a contingency of up to 40%.
The WT Partnership model is detailed with unit rates being applied for all works. IWA review of the
detail did not identify any gap, but questions why overhead wire is required for depot operations
whilst the total network otherwise is wire free.
Lack of a system scoping document and detailed assumptions by WT Partnership to support their
CAPEX model is also a gap that would be very useful to reference. Adding a “tab” with detailed outline
of the methodology they applied and assumptions would significantly improve this model.
The initial WT Partnership draft estimate was equivalent to $47.97m per route km, whilst the Turner
& Townsend estimate was equivalent to $98.25m per route km. WT Partnership subsequently
reviewed and adjusted their cost estimate to $78.07m per route km. This is a cost that, based on
Australian examples, IWA considers to be a reasonable initial cost estimate. However, we caution that
the lack of a detailed scope means that there is no assurance that WT Partnership accounted for all
material items and issues within their model.
IWA also note that advice provided by Ken Davis (Title “Light Rail Power System”, dated 220/10/2014)
advised that “in a hilly city such as Auckland, overhead line is likely to be the best or only solution in
some places”. This limitation has not been addressed within material sighted.
IWA note a report by URS titled “Structural Review Light Rail Transit (LRT) Loadings/Specific Bridge
Review” and note with the Grafton Bridge that “it is improbable that the bridge will be capable of
being strengthened further to carry two LRT tracks”. Subsequently IWA is advised the solution
proposed was that the central 120m of Grafton Bridge would be designed with a single track section
(or interlaced gauntlet tracks) with traffic signal control. Therefore it would only have to support the
dead weight of a single track slab, and the centre span of the bridge could only be ever occupied by
one LRV at a time. Subsequent advice is that the intent is to use a single track only. The view expressed
was that this should not create any difficulties operationally. IWA does not agree with this view in the
absence of a more detailed assessment.

IWA/N207/Rep/1552
26 July 2015
30

Document Outline