Meridian Energy of Christchurch, New Zealand operates the Manapouri hydro power station, an underground facility that is located in Fiordland National Park at the west end of Lake Manapouri on the south island of New Zealand. The 178m difference in elevation between Lake Manapouri and Deep Cove, an inlet from the Tasman Sea, was recognised as a potential source for hydro power generation in the late 1800s. Several studies were undertaken and by the latter half of the 1960s, construction methods and the demand for electricity made development of the site viable.

From Lake Manapouri, water is drawn down through seven penstocks to underground turbines and then discharged through a 9.8km long tunnel into Deep Cove, Doubtful Sound and eventually into the Tasman Sea. This tunnel was constructed by conventional drill and blast techniques and is known as the first Manapouri tailrace tunnel (1MTT). The tunnel is concrete-lined throughout and has an inverted horseshoe cross-section with a nominal diameter of 9.15m. In plan view, the tunnel runs straight from the draft tube manifold area at the power plant to the exit portal at Deep Cove. The elevation shows the tunnel was constructed as an inverted syphon, with the maximum depth being approximately 30m below the outlet into Deep Cove. The main portion of the tunnel has a very gradual downward slope from east to west of 0.1%.

The full potential of the power station has never been realised due to higher than anticipated friction in the tailrace system; it has never been able to safely produce more than 585MW of a potential 700MW of power. A number of attempts to detect and/or significantly reduce this head loss were unsuccessful. There was concern that a failure may have occurred during the original commissioning and even rumours that some heavy construction equipment had been left behind and is blocking the tunnel. Since the 1MTT cannot be de-watered for inspection (there is no means of closing off the exit portal), the operators were forced to conclude the existing facility would never be able to operate to its design capacity.

Lost power

With the increasing demand for domestic power, several engineering studies and environmental assessments were initiated to resolve the ‘lost power’ issue. This process resulted in the design and commissioning of a second tunnel called 2MTT. Numerous logistical challenges had to be overcome since the site is inaccessible by road and located in an environmentally and culturally important area (Fiordland National Park is in a work heritage area).

After extensive site preparation, the tunnel boring machine (TBM) entered the tunnel portal at Deep Cove on 20 May 1998. As part of the 2MTT construction process, plant outages were scheduled to accommodate activities in preparation for connecting 2MTT to the generating station. Meridian contracted Aquatic Sciences Inc (ASI) of Canada. It was decided to use the ASI Mantaro remotely operated vehicle (ROV) during one of these outages to conduct a full length underwater investigation of 1MTT to help establish the cause of the 30m head loss.

ASI has proven the viability of ROV long range tunnel inspections as a safe and cost-effective technology to determine the internal condition of tunnels. ASI has completed continuous surveys of 10km from a single access point using this robotic system, including inspections of a 120km water supply tunnel in Finland and an 8km headrace tunnel in Chile.

The ASI Mantaro ROV was specifically developed for long flooded tunnel and pipe inspections. The vehicle has 12 electric thrusters that can be used to position the vehicle virtually anywhere in the flooded tunnel. The 10km long control cable consists of power conductors, fibre optic threads, Kevlar strength member and an outer protective jacket for abrasion resistance, all combined into a small 13mm diameter umbilical. This configuration produces a cable with minimal drag, enabling excursions to the full cable length.

The ROV pilot controls all of the thrusters, camera and sonar functions from surface control consoles that are generally installed in a mobile site trailer. All sensor data is displayed in real time at the surface control station.

Three monitors are used to display camera video, navigation and profile sonar images. The ROV depth, heading, distance of travel, and vehicle status information are displayed as an overlay on the camera video screen. The ROV pilot records all video data in VHS format with audio commentary. All sonar video information is recorded in S-VHS format and the tape audio channels are used to record the original signals from the sonar head and processor. Working beside the ROV pilot a data technician co-ordinates the logging of all of the videotapes and data records, using a computerised database to ensure that records can be easily retrieved and correlated at a future date.


For the 1MTT inspection, the ASI Mantaro was configured with four cameras, three scanning sonars, an acoustic doppler current profiler (ADCP), and a five-function manipulator. Of the four cameras, one was a high resolution colour camera with 12x zoom capability; one was a standard commercial grade, colour CCD camera suitable for clear water use; and two were low light, high resolution, black and white cameras for turbid conditions. Variable intensity halogen lamps provided lighting for the cameras.

One of the sonars, a compact 675kHz unit, was mounted on the front of the vehicle and used for navigation and collision avoidance. Two other sonars were mounted at the rear of the vehicle to gather profile data of the tunnel cross-section. One of these units, operating at 2.25MHz, has a very high resolution and can generate very detailed images. Because of the higher frequency, the useful measuring range of this unit is limited to approximately 12m. The other profiling sonar operates at a lower frequency of 675MHz and has a range capability of 100m, but at a lower resolution.

The ADCP is an acoustic sensor that can measure water velocities in several discrete layers, usually in oceans, rivers and streams. Since one of ASI’s tasks was to identify the locations of possible water leakage in or out of the tunnel and to measure these flows, the ADCP was mounted horizontally on the front of the ROV. The system was configured to detect flows as low as 2.5cm/sec when operated according to a specific protocol.


The 1MTT inspection was originally scheduled to occur during a planned shutdown in February 2000. Due to the logistics associated with alternative power supply and transmission, the shutdown was re-scheduled to May 2000. The ASI Mantaro system, already en route to the site, was diverted and stored at a local warehouse facility.

Meridian and ASI arranged the delivery of the system to the site for the beginning of the second week of May 2000. ASI personnel arrived later in the week to complete the site mobilisation of the ROV system. All equipment was tested and operationally ready by Wednesday 17 May. During this time, Meridian and ASI held several meetings to discuss the operating procedures and possible schedules for the inspection in detail. Although a full 48-hour period had been allocated for the plant shutdown, the economics of local power generation required flexible schedules that provided for plant start-up prior to the end of this time period.

Several inspection schedules were developed, based on the estimated time for recovery of the ROV and the amount of notification that could be provided should Meridian decide to re-start the plant. It was decided to fly the ROV as quickly as possible up to the generating station and note any gross anomalies for closer inspection during a follow-up period.

ASI typically classifies observations on a scale of 1 to 5 based on the level of importance. During the pre-inspection meetings, engineers from Meridian established the classifications as they would pertain to 1MTT. They also detailed areas of special interest and assigned them with a priority of interest scale. These areas were coded using a system that indicated the type of abnormality that was to be expected.

The plant shutdown started at 19:00 hours on Friday 19 May. Water flows were sufficiently reduced to allow deployment of the ROV into the outlet channel by 23:30. Generally, the ROV was flown upstream towards the generating station along the tunnel invert. This ensured that any gross anomalies were identified for a more thorough inspection during the vehicle recovery. The maximum penetration was completed by 06:20 hours the following day at a cable pay-out of 9632m. For the outward (downstream) excursion, the ROV was positioned near the crown of the tunnel to readily identify and inspect anomalies along the crown.

Due to dramatic changes in the pricing of local power delivery (almost ten times the starting price in less than 24 hours), Meridian decided to terminate the shutdown prior to the 48-hour mark. This limited the ROV inspection to only one pass in and out of the tunnel, which had to be completed as quickly as possible. Since there were very few significant anomalies that warranted detailed inspection, the crew was able to complete the inspection and recover the ROV by 20:15 hours on Saturday, marking the end of a continuous 21-hour inspection mission. This represented the condition at the extreme limit of the planning scenarios.

Throughout the inspection, all camera and sonar images were recorded onto videotape and all pertinent observations were logged in the ASI computerised database. Meridian engineers and other officials were present at all times.

Excellent condition

The data records made during the inspection were printed out upon completion of the inspection to form the bulk of a preliminary report. This document also included several images from the high frequency profiling sonar that illustrated typical and atypical conditions. A highlights tape, made during the inspection using selected camera and sonar video images, formed the video portion of the preliminary report. This was submitted within 24 hours to Meridian for its review.

The inspection results generally confirmed that the tunnel and lining were in excellent condition. It also eliminated the speculation that the tunnel was blocked by debris and dispelled the ‘forgotten heavy equipment’ myth. However, at the time of writing, some specific findings are under review that may help to determine the cause of the head loss.