Nepal Tunnel and Hydropower Worker Safety
A working plan.
नेपालीमा सारांश
२६ अगस्त २०२६ मा नेपाल–तिब्बत सिमानानजिक करिब ५,२०० मिटरमा हिमनदी खस्दा लेन्दे हुँदै भोटेकोशी र त्रिशूलीमा आएको बाढीले त्रिशूलीको सतह तीस मिनेटमै नौ मिटर बढाइदियो। २ सेप्टेम्बरसम्म १,११४ शव भेटिएका र ३,९१६ जना सम्पर्कविहीन रहेको प्रधानमन्त्रीले संसद्मा बताए। सम्पर्कविहीनमध्ये करिब ९३३ जलविद्युत् आयोजनाका कामदार थिए — धेरैजसो हेडरेस सुरुङभित्र।
पाँच कुरा असफल भए: सुरुङभित्र चेतावनी पुगेन, भित्र बाँच्न मिल्ने ठाउँ थिएन, भित्र को थियो भन्ने कसैलाई थाहा थिएन, उद्धारको क्षमता नै थिएन, र सहयोग स्वीकार गर्ने संरचना थिएन। यी सबैको समाधान नयाँ प्रविधि होइन — संसारभर चलिसकेको साधारण इन्जिनियरिङ हो, जुन नेपालमा कहिल्यै अनिवार्य गरिएन।
राष्ट्रिय स्तरमा लागू गर्न करिब ४.५ देखि ७ करोड अमेरिकी डलर लाग्छ — यही एउटा विपद्मा भएको ४ अर्ब डलरभन्दा बढी क्षतिको डेढ प्रतिशत। पूरा प्राविधिक दस्तावेज तल अङ्ग्रेजीमा छ। सुझाव पृष्ठको अन्त्यमा नेपाली वा अङ्ग्रेजी — जुनसुकै भाषामा दिन सकिन्छ।
0.The one-page version
On 26 August 2026 a glacier detachment at roughly 5,200m near the Nepal–Tibet border sent a debris flood down the Lende into the Bhote Koshi and Trishuli. The Trishuli rose nine metres in thirty minutes. By 2 September the Prime Minister told Parliament that 1,114 bodies had been recovered and 3,916 people were still out of contact. Around 933 of the missing were hydropower workers, most of them inside headrace tunnels.
Nothing about this was technically unforeseeable. Tapovan, Uttarakhand, 2021 was the same event with the same outcome. The July 2025 Lende flood that destroyed the Miteri bridge was the warning shot on the exact same river.
Five things failed, in order:
- No warning reached the tunnel face. A surge that took roughly an hour to travel from the glacier to the projects should have triggered an in-tunnel evacuation. It did not, because nothing connects upstream sensing to a hooter inside a headrace.
- No survivable space inside. Workers hung off ceiling rungs for six hours. There were no refuge chambers, no air, no supplies, no second adit.
- Nobody knew who was inside. Trapped counts at Trishuli 3A were reconstructed from village hearsay. IPPAN listed a dozen projects with staff simply "out of contact."
- No rescue capability existed to deploy. Six days passed at 3A before a working access route existed. Parliament was told an excavator arrived at 144 hours when five was possible. When rescuers finally blasted the crown they found water standing to the roof of a 7m tunnel, and had no dewatering capacity to do anything about it.
- No command structure that could accept help. Foreign specialist teams arrived late and were held at arm's length. Power producers said publicly that time was lost.
Fixing this is not a moonshot. It is five layers of cheap, boring, well-established engineering that Nepal has never mandated. Total capital cost for national coverage is in the range of USD 45 to 70 million, roughly 1.5 percent of the USD 4 billion+ the Finance Minister has already put on this single disaster.
The plan below is organised as five defence layers, then a delivery roadmap, then the political and financing mechanics that determine whether any of it actually happens.
1.Failure analysis in detail
1.1 The hazard is now structural, not exceptional
ICIMOD's March 2026 assessment found Hindu Kush Himalaya glacier ice loss has doubled since 2000. Nepal has roughly 1,466 glacial lakes. Nepal's hydropower build-out is concentrated in exactly the high-relief, glacier-fed, seismically active valleys where glacial detachments, rock-ice avalanches, permafrost-driven slope failures and moraine-dam breaches are all becoming more frequent.
The design assumption that a headrace tunnel is a safe enclosed space is wrong in this terrain. It is a sump with one exit.
1.2 The specific failure chain at Upper Trishuli 3A
- Glacier detaches ~5,200m, falls ~1,200m, entrains rock and debris, hits the Lende.
- Surge travels down Bhote Koshi to Trishuli. Government received information at 08:45, warned settlements at 09:00, sent mobile alerts at 09:35. None of that reaches a man 800m inside a tunnel with no radio.
- Slurry fills the tunnel. Water stands to the crown of a 7m bore.
- Portal and intake are buried. Rescuers spend days simply locating the intake, because there is no surveyed, surface-referenced portal marker and no digital as-built to consult.
- Crown is finally located and drilled on day 5. Air pumped, food dropped, rescuers rappel in, call out, no response.
- Estimated 180m from portal to the suspected worksite, with 35 to 40 people believed inside.
Each of those five steps has a standard, off-the-shelf countermeasure. None was in place.
2.Layer 1: Upstream detection and warning
Goal: a surge detected upstream automatically sounds a hooter at the tunnel face within 60 seconds, with no human in the loop.
2.1 What Nepal already has
DHM operates hydro-met stations nationally. GLOF-specific early warning exists at Tsho Rolpa (sensors plus 19 sirens across 18 villages and 3 relay stations, installed from 1998) and at Imja (hydro-met stations, GLOF sensors and automated sirens in six settlements, installed 2016 with UNDP and GEF). Both were built as community warning systems, for villages.
Two gaps: coverage is a handful of named lakes out of ~1,466, and no hydropower project is a subscriber. The Tsho Rolpa experience also produced a documented lesson that over-sophisticated systems fail on maintenance, which should shape procurement.
2.2 What to build
Tier A: basin-scale trigger network
Radar water-level gauges plus vibration/seismic sensors at 3 to 5 points per high-risk basin, satellite-telemetered (Iridium SBD, not GSM, because towers go down first). Nepal lost 198 telecom towers in this event. Target the 20 highest-risk hydropower basins first: Bhote Koshi/Trishuli, Tama Koshi, Dudh Koshi, Arun, Marsyangdi, Budhi Gandaki, Seti, Kali Gandaki, Langtang, Modi.
Approximate cost: USD 40k to 80k per basin fully installed. USD 1.5M for twenty basins. This is not a large number.
Tier B: satellite and remote-sensing screening
Sentinel-1/2 and Planet imagery on a standing pipeline to detect new lake formation, moraine dam changes and slope deformation. ICIMOD and Kathmandu University already have the capability.
The owner is not ambiguous, contrary to what is often assumed. DHM's published mandate explicitly covers snow hydrology, glaciology and limnology, it holds a Government of Nepal mandate to monitor all hydrological and meteorological activity in the country, and its stated position is that no other agency may carry out such activity without liaison with DHM. The Hydrology and Meteorology Policy 2024 governs the sector. DHM already runs 24/7 flood forecasting and early warning through the monsoon.
So this is not a mandate gap. It is a resourcing and operational-linkage gap, and that reframing matters for what you ask for. DHM's annual budget for FY 2081/82 was around NPR 504 million, roughly USD 3.6 million, for a department responsible for all national hydromet monitoring including glaciology. Asking for a new institution is the wrong ask and will be resisted. The right asks are: fund DHM's cryosphere function properly, and create a statutory duty running from DHM's alert to the licensee's siren, which is the link that does not currently exist in any form.
Tier C: cross-border data sharing with China
After the July 2025 Gyirong flood, Nepal and China verbally agreed to share river-level and glacial-hazard data. As of 26 August 2026 that agreement had never been formalised. This is the single highest-leverage, lowest-cost item on the entire list and it is a signature. It should be the first ask in Parliament.
Tier D: last-mile delivery into the tunnel
This is the piece nobody has built. The chain is: sensor trigger, satellite uplink, project control room, hardwired hooter and strobe circuit running the full length of the headrace, plus leaky-feeder or through-the-earth radio. Sirens must be on a dedicated circuit with battery backup, independent of project power.
Rule to legislate: any alert triggers automatic evacuation. No supervisor discretion, no confirmation step. False alarms are acceptable and should be budgeted for. The cost of a false evacuation is a lost shift. The cost of a missed one is Rasuwa.
3.Layer 2: In-tunnel survivability
Goal: anyone inside can reach a survivable, air-supplied space within 15 minutes, and stay alive for 72 hours.
This is standard practice in underground mining worldwide and standard in modern road and rail tunnels. It has simply never been applied to Nepali hydropower tunnels.
3.1 Refuge chambers
Mandate for any tunnel where any working face is more than 500m from a portal:
- Refuge chambers at maximum 1,000m spacing along the drive, plus one within 300m of the advancing face (mobile units are fine for the face).
- Capacity sized to peak shift headcount plus 25 percent.
- 72-hour independent air: compressed air cylinders plus CO2 scrubbing, not reliant on a surface compressor line, because the compressor line is what floods.
- Positive pressure and a water-rated door. This is the Nepal-specific modification. Standard mining refuge chambers are designed against gas and fire, not against 7m of slurry. Chambers here need a pressure-rated, sealable door, an elevated sill, and mounting at the tunnel crown or in a dedicated raise above invert level.
- Water, rations, first aid, chemical toilet, lighting.
- A pre-drilled borehole to surface for each fixed chamber, cased, capped, GPS-logged, with the coordinates lodged with DoED and the Army. This is the item that would have changed the outcome at 3A. Rescuers spent days finding the tunnel. With a logged borehole they drop a camera and a microphone on day one, and they know within hours whether anyone is alive and where.
Reference standards to adapt rather than write from scratch: US MSHA 30 CFR Part 7 Subpart L (refuge alternatives), India DGMS circulars on refuge chambers in underground mines, Australian MDG 3608 / Queensland refuge chamber guidance, and NFPA 520 for subterranean spaces. The ITA (International Tunnelling Association) WG5 material on health and safety in works is the right umbrella. Nepal should adopt by reference with a water-ingress amendment, not attempt an original standard. Adopting by reference can be done in months. Writing an original standard takes years and Nepal does not have the specialist drafting capacity.
Cost: an imported 20-person mining refuge chamber runs roughly USD 80k to 200k installed. A locally fabricated unit built to the same functional spec should land at USD 30k to 60k. There is a genuine domestic manufacturing opportunity here and it should be deliberately seeded.
3.2 Second egress
Every drive over 1,500m gets an intermediate adit or a vertical escape shaft. The Silkyara investigation in India flagged absence of escape routes as a root cause in 2023. Nepal's tunnels have the identical defect. New projects: condition of licence. Existing projects: retrofit programme, prioritised by drive length times peak headcount.
3.3 Communications and lifelines
- Leaky feeder cable or through-the-earth (TTE) radio for the full drive length. TTE is preferable in Nepal because it survives cable severing, which a flood guarantees.
- Continuous handrail or lifeline along one wall with distance-to-portal markers every 50m, tactile, readable in zero visibility. Ram Chandra's group took six hours to find their way out by feel. Markers are paint and steel. They cost nothing.
- Self-rescuers (SCSR units) issued per person, not stored centrally.
4.Layer 3: Knowing who is underground
Goal: within 5 minutes of an incident, an authoritative list of names inside, and their last known chainage.
This is the cheapest layer and the one whose absence caused the most anguish in Rasuwa. Families learned nothing for a week. Trapped counts came from "information gathered from people in nearby villages."
- Mandatory tag-in/tag-out at every portal. NFC or UHF RFID cards, an offline-capable Android reader, a local board and a cloud mirror. Off-the-shelf, roughly USD 5k to 20k per portal.
- Zone readers at intervals down the drive to give last-known-position, not just in/out. Optional for small projects, mandatory above 2,000m drive length.
- A national worker register. Every underground worker in Nepal, with employer, project, next of kin, blood group, and a DNA reference sample held under consent. Nepal is currently flying in foreign DNA experts and 2,000 body bags with no reference database to match against. That is a solvable administrative failure.
- Sub-contracted and daily-wage labour must be in the same system. In practice this is where the gap will be, and it is where the deaths concentrate. Workers at these projects came from Jumla and other far-western districts on informal terms.
This layer alone, at national scale, costs under USD 3 million and could be operational in six months. It should be started immediately and not made to wait for the rest of the programme.
5.Layer 4: A real national tunnel rescue capability
Goal: heavy tunnel rescue equipment on site anywhere in Nepal within 12 hours, operated by a standing trained unit.
5.1 The core insight
The bottleneck at Rasuwa was not exotic technology. It was water, mud and access. A flooded, silt-choked headrace is a pumping problem before it is a rescue problem, and Nepal owns no pumping capacity at that scale. Rescuers drilled to the crown and found water to the roof, then had nothing to do about it.
Silkyara 2023 is the optimistic precedent (41 workers, all recovered after 16 days). But Silkyara was dry rock with a 60m blockage. Tapovan 2021 is the honest precedent: glacier flood, hydropower tunnel, slurry-packed, most people inside did not come out. Mud-filled is a categorically different problem from rock-collapsed, and Nepal must plan for the mud case.
5.2 The cache
A pre-positioned National Tunnel Rescue Cache, staged in Kathmandu on heli-liftable pallets, with a duplicate forward cache in Pokhara or Damauli:
Dewatering (highest priority, the current total gap)
- High-head electric submersible pumps, 150 to 500 kW class, capable of pumping heavy silt-laden water. Godwin HL / Flygt BIBO class or equivalent.
- Large-bore layflat discharge hose in bulk, 200mm+.
- Diesel gensets sized to the pumps, plus fuel bladders.
- Slurry and sludge pumps for the mud fraction specifically.
Access and boring
- A truck- or heli-portable drill rig capable of 300mm to 900mm vertical boreholes to 200m. The Chile 2010 San José rescue turned on exactly this class of machine. Nepal cannot justify one alone; this is the right candidate for a regional shared asset with India and Bhutan.
- Auger boring set for horizontal pipe insertion through debris (the Silkyara method).
- Rock breakers, hydraulic cutters, and enough excavators that a Pranaya Magar never again has to post on Facebook that only one machine was running.
Search and life detection
- Borehole cameras with 200m+ cable, and pan-tilt heads.
- Acoustic/seismic life detectors (Delsar LD3 class).
- Thermal imagers.
- Small tethered ROVs for flooded sections. This is the piece that would have let teams see inside 3A on day two instead of guessing.
- Portable ground-penetrating and seismic-refraction kit for void location, with realistic expectations: GPR is close to useless at depth in wet rock, so budget for seismic and electromagnetic methods instead.
Life support
- High-volume air compressors and ducting for tunnel ventilation.
- Supply capsules for borehole delivery of food, water, comms.
Approximate cache cost: USD 8 to 15 million. For scale, this single event cost Nepal more than USD 4 billion.
5.3 The unit
Equipment without a standing team is a warehouse. Create a National Tunnel and Mine Rescue Unit, roughly 120 people, under the Nepali Army with civilian technical secondment from NEA, NTA and the Department of Mines. The Army already has personnel trained in disaster rescue in India, China and the US. This unit is the vessel to concentrate them in.
- Quarterly live exercises at real hydropower projects, unannounced.
- Standing MoUs with India's NDRF, China's rescue teams and Korea's 119 unit, pre-signed, with pre-cleared customs and airspace protocols. The delay in accepting foreign help was widely criticised. That criticism is fair but the deeper problem is that there was no pre-agreed mechanism, so every offer required a political decision under maximum stress. Pre-signing removes the decision.
- Named on-call incident commander with statutory authority over site, so that the technical lead is not competing with visiting ministers. The surveyor at 3A asked lawmakers to let technical experts work or step back. That should be law, not a Facebook post.
5.4 Data prerequisites (near zero cost, very high value)
Make these conditions of licence, effective immediately:
- Digital as-built drawings of every tunnel, deposited with DoED and NDRRMA, updated monthly during construction, in a machine-readable format with surveyed coordinates.
- Surface-referenced chainage markers: permanent surveyed monuments on the ground surface above the tunnel alignment every 250m, GPS-logged, so rescuers can locate a point inside the tunnel from above without excavation.
- Portal and adit coordinates in a national registry.
- Emergency response plan per project, filed and exercised annually.
Days were lost at 3A locating an intake and a crown. That should be a database lookup.
6.Layer 5: Governance, and the enforcement problem
This is where plans like this usually die. Nepal does not primarily have a standards problem, it has an enforcement problem. DoED does not have inspectors who can meaningfully audit a tunnel drive, and it will not have them within five years.
So do not route enforcement through inspection. Route it through capital.
Three chokepoints, in order of effectiveness:
6.1 The lending chokepoint (fastest, most effective)
The numbers make this stronger than expected. An IPPAN-published study puts bank lending to hydropower at around NPR 447 billion out of a total loan book of about NPR 5,563 billion, up from NPR 2.76 billion in July 2009. An NRB spokesperson stated in July 2026 that roughly NPR 500 billion is outstanding to the energy sector across A, B and C class institutions plus infrastructure development banks, with headroom for around NPR 900 billion more. IPPAN puts total financial commitments near NPR 870 billion.
Critically, this exposure exists because NRB directed it. The central bank mandates that banks place a defined share of their total loans (reported as 10 percent, within a prescribed 10 to 15 percent band) in the energy sector. Bankers have said publicly that they lend to hydropower largely because of that requirement, despite the geographic risk. NRB has also given hydropower loans favourable classification treatment, where only the unpaid instalment is downgraded rather than the whole facility.
That changes the argument. The lever is not something that has to be invented; it is an existing directed-lending regime with an existing prudential carve-out, and both can carry conditions. The proposal is narrow: make the tunnel safety schedule a condition of qualifying for prescribed-sector treatment and for the favourable classification carve-out. A project that does not comply stops counting toward a bank's mandatory energy quota. That single change makes every lender in Nepal an enforcement agent overnight, at zero cost to the state.
The timing is also right. The Nepal Bankers' Association put around NPR 6.5 billion of hydropower loans at risk from this flood across four damaged projects, with total banking exposure to the disaster around NPR 10 billion once non-hydro borrowers in Rasuwa, Nuwakot and Dhading are included. Nabil alone has NPR 3.3 billion in the 37 MW Trishuli project and NPR 498 million in Rasuwagadhi. Bankers are already saying lending to the sector may contract. Banks are looking for a way to price and manage this risk right now, which is exactly when a compliance standard is welcome rather than resented.
IFC, ADB and other DFIs already carry E&S performance standards that require most of this. They simply have not been enforced on domestically financed projects. One NRB circular does more than a decade of DoED inspection.
6.2 The insurance chokepoint
Nepal Insurance Authority plus the reinsurers who now have to price Himalayan hydropower after a USD 4 billion loss year. Reinsurance capacity for Nepali hydro is about to get expensive and conditional regardless of what anyone in Kathmandu wants. Get ahead of it: define the safety package as the qualifying standard for insurability, so that compliance buys premium relief and non-compliance becomes uninsurable, which makes it unfinanceable.
6.3 The PPA and licence chokepoint
NEA's power purchase agreement and DoED's generation licence are the state's direct levers. Attach the package as licence conditions on new licences immediately, and on renewals and capacity revisions for existing plants. Include a stop-work power for the incident commander and for a certified site safety officer, protected from dismissal.
6.4 Statutory work, in parallel
- Amend the Electricity Regulation to incorporate the tunnel safety schedule.
- Bring tunnelling explicitly under the Labour Act 2074 OSH provisions with an underground-specific schedule.
- Create a Chief Inspector of Tunnels post with real technical staff, seconded initially from NTA and IOE Pulchowk.
- Mandatory published incident reporting, with a no-blame technical investigation function modelled on aviation accident investigation, separate from criminal liability. Without this, nobody will ever tell the truth about a near miss.
7.Costing summary
Correction on scope. An earlier draft scaled this to roughly 150 projects. The actual denominator is larger: around 189 hydropower projects above 1 MW are operational, and around 260 projects above 1 MW hold construction licences from DoED, for roughly 450 in total. Not all have significant tunnels, and Nepal's fleet is overwhelmingly run-of-river with widely varying headrace lengths, so the number of projects in scope needs to be established from DoED records before these figures mean anything.
Treat the table below as a floor, not an estimate. If the tunnelled subset is 250 rather than 150 projects, the refuge chamber and egress lines roughly double and total capital lands nearer USD 90 to 100 million. That is still small against a USD 4 billion+ single-event loss, but the plan should be presented with the honest number, because being caught understating cost is how credibility is lost in a committee room.
| Layer | Item | Capital | Annual O&M |
|---|---|---|---|
| 1 | Basin sensor networks, 20 basins | USD 1.5M | USD 0.3M |
| 1 | Satellite screening cell, staffed | USD 0.3M | USD 0.4M |
| 1 | In-tunnel alarm circuits, retrofit | USD 4M | USD 0.5M |
| 2 | Refuge chambers, ~400 units | USD 20M | USD 2M |
| 2 | Second egress retrofit, priority projects | USD 12M | minimal |
| 2 | Comms, lifelines, self-rescuers | USD 5M | USD 1M |
| 3 | Tag-in/tag-out, national register, DNA baseline | USD 3M | USD 0.5M |
| 4 | National rescue cache | USD 12M | USD 1.5M |
| 4 | Standing rescue unit, 120 pax | USD 2M setup | USD 2.5M |
| 5 | Regulator capacity, registry, digital as-builts | USD 1M | USD 1M |
| Total | ~USD 61M | ~USD 9.7M/yr |
Against USD 4 billion+ in losses from one event, and 431 MW of generation knocked offline from 12 damaged projects.
Funding sources: Green Climate Fund and Adaptation Fund (this is textbook climate adaptation and Nepal has a strong case), World Bank and ADB post-disaster reconstruction lending, a levy of 0.5 to 1 percent on hydropower project capital cost paid into a ring-fenced safety fund, JICA (which has just completed Nagdhunga and has live Nepali tunnel-sector relationships and expertise), and bilateral technical assistance from India, China and Korea whose teams are already on the ground.
8.Roadmap
Days 0 to 30: the window
The political window is open right now and it is short. Parliament is already debating the 144-hour excavator. The Gen-Z protest anniversary falls on 8 to 9 September. Attention moves on in six to ten weeks.
- Nepal Tunnelling Association plus Nepal Engineers Association plus IPPAN issue a joint technical demand document. One document, three signatures. Not three separate press conferences.
- Demand a statutory independent technical inquiry into the tunnel response, with published terms of reference and a fixed reporting date. Not a ministerial committee.
- Sign the Nepal–China hydrological data sharing agreement. It is already verbally agreed. It costs nothing.
- Emergency directive from DoED requiring every licensee to file, within 30 days: portal coordinates, as-built alignment, peak underground headcount, current emergency plan.
- Begin the DNA reference and worker register for all currently employed underground workers.
- Order the dewatering package of the rescue cache on emergency procurement. It is the longest lead time and the largest gap.
Days 30 to 90
- Publish the Tunnel Safety Schedule as a licence condition, adopted by reference from MSHA/DGMS/ITA with the water-ingress amendment.
- NRB circular and Insurance Authority directive attaching the schedule to lending and insurability.
- Tag-in/tag-out pilot at 5 projects, then national mandate.
- Stand up the National Tunnel and Mine Rescue Unit, first cohort into training with NDRF and Chinese/Korean partners.
- Tender locally fabricated refuge chambers, with a design competition run through IOE Pulchowk and Kathmandu University.
Days 90 to 365
- Basin sensor networks live in the top 10 basins, wired to project alarm circuits.
- Refuge chambers installed at all under-construction drives over 500m.
- Full rescue cache delivered and first national exercise conducted.
- Chief Inspector of Tunnels office operational.
- Second-egress retrofit programme underway at the highest-exposure existing projects.
Years 1 to 3
- All 20 priority basins instrumented, cross-border feed live.
- Refuge chamber retrofit complete across operating fleet.
- Domestic manufacturing of chambers and monitoring hardware established.
- Regional shared large-bore drilling rig agreement with India and Bhutan.
- Statutory framework fully in force, first cycle of independent incident investigations published.
9.Institutional map: who owns what
| Function | Owner | Support |
|---|---|---|
| Policy and licence conditions | MoEWRI, DoED | NTA, NEA (Engineers Assoc.), IPPAN |
| Standards drafting | Nepal Tunnelling Association | ITA, IOE Pulchowk, DGMS/NDRF liaison |
| Upstream sensing | DHM | ICIMOD, Kathmandu University, UNDP |
| Alarm delivery to site | Project licensees | DoED audit |
| Worker register and DNA | Ministry of Labour, NDRRMA | Nepal Police forensics |
| Rescue unit and cache | Nepali Army | APF, NEA, Dept. of Mines |
| Financing lever | Nepal Rastra Bank, Nepal Insurance Authority | Bankers' association, reinsurers |
| Independent investigation | Statutory commission, reporting to Parliament | International tunnelling experts |
The single biggest institutional risk is that this gets handed entirely to one ministry and dies there. It needs a named PMO-level owner with cross-ministry authority, because it spans energy, labour, home, finance and defence.
10.Where a technical builder actually adds value
Honest assessment: nobody outside the tunnelling profession should be writing engineering standards or specifying pumps. But there are three pieces of this plan that are pure software and data problems, that nobody currently owns, and that a small technical team can ship in weeks rather than years.
1. The National Tunnel Registry (build this first)
An open, public dataset of every tunnel project in Nepal: operator, drive length, portal coordinates, peak headcount, as-built availability, refuge chambers yes/no, emergency plan filed yes/no. This does not exist. Its absence is why rescuers spent days looking for an intake. It is scrapeable and assemblable from DoED licence data, NEA records, IPPAN membership and EIA filings, and it is a two to four week build. Once it exists, the compliance gap becomes visible and unignorable, which is what forces the rest of the plan. Publish it. Make non-compliance legible.
2. Worker accountability system
Offline-first Android app plus NFC cards plus a cheap portal reader. Must work with no network, sync when it can, and be usable by a supervisor with limited literacy in Nepali. Pilot with two willing IPPAN members, open-source it, then let the licence condition create the market. Total build: one engineer, six to eight weeks.
3. Alert routing middleware
DHM sensor feed into a rules engine into SMS, satellite messaging and hardwired siren relays at project control rooms, with acknowledgement tracking so an unacknowledged alert escalates. The sensors partly exist. The sirens can be bought. The connective layer between them is what is missing and it is a modest piece of software.
The fourth contribution, which is not software: be the connective tissue. NTA has the technical credibility and no product capability. IPPAN has the money and a defensive posture. NEA has 36,000+ members and mobilisation capacity. NDRRMA has the mandate and no tools. Families of the missing have moral authority and no organisation. Nobody is currently holding those five together into a single coherent demand. That role is open and it matters more than any individual artefact.
11.How this plan fails
Naming the failure modes so they can be watched for:
- Attention decay. Six to ten weeks and the news cycle moves. Everything not locked into a circular, a licence condition or a signed procurement by mid-November will not happen.
- A committee instead of a commission. A ministerial committee with no statutory power, no deadline and no publication requirement is the standard method of burying this. Insist on statutory footing and a fixed public reporting date.
- Gold-plating. The Tsho Rolpa system reportedly failed partly because it was too technologically advanced to maintain. Every specification in this plan should be judged on whether a project electrician in Rasuwa can keep it working with no vendor support for five years.
- Exempting existing projects. New-build-only rules protect nobody currently underground. The retrofit programme is the harder political fight and the more important one.
- Skipping the sub-contracted workforce. If the register and the refuge chamber capacity cover only direct employees, the people who die will be the same people who died this time.
- Confusing rescue with prevention. The rescue cache is the visible, fundable, photogenic item. Layers 1 to 3 save far more lives for a fraction of the money and will get a fraction of the attention. Budget accordingly and protect their funding.
Appendix: standards to adopt by reference
Do not draft from scratch. Adopt, amend for water ingress and Himalayan hazard, and publish.
- MSHA 30 CFR Part 7, Subpart L and 30 CFR 75.1506: refuge alternatives in underground coal mines. The most detailed prescriptive refuge chamber standard available.
- DGMS (India) circulars on refuge chambers and underground emergency preparedness. Regionally proximate, similar labour context, and Nepal already has personnel trained in India.
- ITA-AITES Working Group 5 guidance on health and safety in tunnelling works, and ITA guidance on tunnel emergency planning. Establishes international credibility for the framework.
- NFPA 520, Standard on Subterranean Spaces.
- EU Directive 2004/54/EC and Austrian RVS 09.02 series: escape route spacing and emergency ventilation logic. Road-tunnel focused but the egress geometry principles transfer.
- Australian MDG 3608 / Queensland Recognised Standard 08 on refuge chambers, for a functional rather than prescriptive drafting model.
- ISO 22320 on emergency management and incident command, for the command-structure gap.
Every one of these is public. The drafting work for Nepal is a schedule of amendments, not an original document, and a competent working group can complete it in eight to twelve weeks.