从实验室闪爆到加氢站爆炸：氢的四道防线是怎么断的From laboratory flash explosions to hydrogen station explosions: How the four lines of defense for hydrogen were breached

On March 20, 2026, a flash explosion occurred in a laboratory at the School of Materials Science, Huxi Campus, Chongqing University. Two students were conducting an annealing experiment at the time. After the tube furnace heated up, the experimenters suspected a hydrogen leak and immediately closed the main valve of the hydrogen cylinder, but did not cut off the power to the tube furnace. Shortly thereafter, a flash explosion occurred on site, and the shock wave flipped over the ethanol cleaning tank, causing a secondary explosion. The accident resulted in one death and three injuries. The reason this accident is repeatedly cited is not that it happened in a university laboratory, but that it fully demonstrates the typical failure structure of hydrogen accidents: over-aged hoses, hydrogen accumulated at the top of the furnace, exposed heating wires, and the “turned off the gas but did not cut off the power” mistake, which has been written incorrectly in textbooks for decades.

Looking ahead, the deaths and injuries on April 6, 2026, at a hydrogenation section of a chemical company in Gansu, the late-night hydrogen transport trailer explosion on February 24, 2026, in Colton, California, USA, and a report that mistakenly cited the 2019 Kjørbo, Norway incident as a new 2026 accident, have brought hydrogen safety issues from the laboratory to the industrial and public opinion level.

1. The Four Key Factors of the Chongqing University Accident: All Are Indispensable

First, let's fully break down the technical chain of this accident. There are four direct causes of the accident, listed in chronological order: hydrogen leakage; the PTFE hose in the tubular furnace intake pipeline was overused and aged with cracks, causing hydrogen to escape from the damaged area and accumulate at the top of the furnace to explosive concentration; the presence of an ignition source, as the tubular furnace heating wire was exposed, creating a persistent hotspot that ignited the accumulated hydrogen; secondary explosion, where the shockwave from the first explosion overturned the ethanol cleaning tank, and the 99% ethanol vapor mixed with hydrogen to form a secondary fireball, exacerbating casualties; operational mistakes, as after detecting the leak, only the gas source was shut off, without simultaneously cutting off power to eliminate the ignition source.

The accident report also listed management-level issues: the laboratory had no hydrogen alarm, no explosion-proof ventilation facilities, a lack of risk identification, and safety training was merely formal. These five management problems, together with the four technical causes, form a complete failure diagram. Of particular note is the item 'no hydrogen alarm'—if there had been a hydrogen concentration alarm at the top, personnel could have evacuated before the concentration reached the lower explosive limit; if there had been explosion-proof ventilation, the hydrogen accumulating at the top of the furnace would have been diluted to a safe concentration. Both of these are one-time investments that are effective in the long term; their absence indicates that the problem was not that the technology was unavailable, but that management did not treat the infrastructure as a necessity.

There is another easily overlooked factor: the overuse of PTFE hoses. PTFE itself has excellent resistance to hydrogen, high temperatures, and chemical media, making it a common material choice for hydrogen pipelines. However, its failure mode is aging and cracking rather than corrosion—cracks may not be visible to the naked eye, but the strength is decreasing. This is the flip side of the same issue as hydrogen embrittlement in carbon steel pipelines: carbon steel is opened from the inside by hydrogen, whereas PTFE is opened from the outside over time.

2. Gansu Chemical Accident: The Triple Overlap of Overpressure, Static Electricity, and Post-Resumption Period

Around 10:40 on April 6, 2026, an accident occurred in a three-story reinforced concrete workshop at a chemical plant in Gansu, resulting in 3 deaths and 2 minor injuries. There was first an explosion sound at the scene, followed by widespread disaster, and the open flames were extinguished at 12:14. The company's main products are anthraquinone and fine chemicals, involving hydrogenation reactions. It should be noted that as of April 7, 2026, the authorities had not released a full accident investigation report. The following technical analysis is based on industry briefings, process characteristics, and publicly available information from multiple sources, and should be considered a hypothesis awaiting verification rather than a conclusion.

The risk level of the hydrogenation process is the highest in chemical engineering for three overlapping reasons: the reaction conditions involve high pressure (2 to 10 MPa), high temperature, and intense exothermic reactions; the explosion limits of hydrogen are 4.1% to 74.1%, which means there is practically no 'safe concentration,' and it is colorless and odorless, making leaks hard to detect, easy to diffuse, and ignitable by static electricity; the equipment chain involves reactors, pipelines, valves, flanges, and safety accessories, and failure at any link can lead to loss of pressure control.

According to process deduction, the most likely direct triggering chain is: overly fast hydrogen flow, failed stirring, or insufficient cooling causing reaction runaway and overpressure; failure, lack of inspection, or incorrect selection of safety valves or rupture disks preventing overpressure relief; flange gasket welds tearing under overpressure leading to high-speed hydrogen release; static electricity acting as an ignition source, causing a spatial flash explosion; the shockwave damaging buildings and equipment and igniting materials, insulation, and cables, eventually evolving into a major fire. At the deeper root cause level, resuming work after the Qingming Festival is a typical high-risk period—equipment status is unstable, personnel are lax, and operations are unfamiliar, making it easy to encounter rushed work, trial production, and operations exceeding temperature, pressure, or load limits; safety devices fail and special operation management is chaotic; alarms are not activated or fail; early response errors prevent emergency shutdown and evacuation.

This accident is completely different in physical mechanism from the one at Chongqing University: in the former, the ignition source is self-sustained (hot spots on a heating wire), whereas in the latter, uncontrolled pressure leads to high-speed release and static electricity. The former aims to prevent the "ignition source," while the latter aims to control "pressure and flow," and the two cannot replace each other.

3. The four physical properties of hydrogen determine the four forms of the accident.

Looking at several accidents, the dangers of hydrogen can be summarized into four physical properties, each corresponding to a type of accident pattern. This is the basic framework for designing hydrogen safety.

The first property is that the flammability concentration range is extremely wide. Hydrogen’s explosive limits are 4.1% to 74.1% by volume, with the lower limit being only about one-tenth of that in air, meaning that even a small leak can form a combustible mixture. This property dictates that the first line of defense in an accident must be rapid detection and rapid shutoff; relying on humans to smell it does not work—because hydrogen is colorless and odorless.

The second point is that the volumetric energy density is low. The volumetric calorific value of hydrogen is about one-third that of natural gas, which means that hydrogen storage and transportation must rely on high pressure or low temperature, making the container itself a source of risk. In the late night of February 24, 2026, at a vehicle storage area in Colton, California, USA, a hydrogen transport trailer exploded during operations and caused a fire. A passenger car at the scene also caught fire. The fire department controlled the fire within about 20 minutes and initiated a hazardous material response procedure. The accident resulted in 1 death and 1 serious burn injury. The trailer involved was reported to belong to Pilot Company, which responded that it has suspended operations related to compressed hydrogen gas and is cooperating with the investigation. The specific cause has not yet been announced, and the investigation is ongoing.

Article 3 is that it penetrates and diffuses quickly. Hydrogen molecules are small and highly permeable, making them prone to leakage and accumulation in parts such as pipelines, valve chambers, and connectors. Pipelines, valve chambers, and trenches are often unmanned areas, and periodic inspections have monitoring blind spots. By the time anomalies are detected during inspections, concentrations are often close to dangerous levels. This property determines that the second line of defense must be online continuous monitoring, rather than periodic testing.

Article 4 is that it has an extremely low ignition energy and a fast combustion speed. Hydrogen’s minimum ignition energy is far lower than that of natural gas, and even a tiny static spark, electrical switch spark, or cigarette butt can ignite it; once ignited, the flame spreads quickly, is colorless (difficult to see under visible light), and can easily form a deflagration in open spaces, fleeting and hard to control. This property determines that the third line of defense is to eliminate ignition sources and strictly control hot work operations.

Matching these four physical properties with the four lines of defense, the framework emerges: the first line is source control — aging management of materials, regular replacement of seals, interlocks for pressure and flow; the second line is detection and shutoff — continuous online monitoring plus automatic interlock shutdown; the third line is ignition source control — explosion-proof electrical systems, no exposed hot spots, hot work management; the fourth line is consequence mitigation — ventilation, explosion relief, accident isolation, and emergency preparedness plans. The Chongqing University accident report shows that all four lines of defense were completely absent at that time.

4. The 2019 Kjørbo accident was misreported: the break in the information chain itself is a risk

Since September 2026, a Korean media report circulated within the hydrogen energy circle, claiming that on September 3, 2026, a hydrogen refueling station on the outskirts of Oslo, Norway, exploded. The preliminary cause of the accident was identified as a malfunctioning pressure relief valve, and the Norwegian government requested emergency inspections of all hydrogen stations nationwide. After Platinum Way cross-checked the local Norwegian media, the Norwegian Directorate for Civil Protection (DSB), and relevant companies, it found that this so-called accident could nowhere be verified.

The details described in the report point to another real incident: on June 10, 2019, a violent explosion occurred at the Kjørbo Uno-X hydrogen refueling station near Sandvika, Norway. The shockwave was so strong that the airbags in nearby vehicles were deployed, and the explosion tremors were recorded by seismic monitoring equipment tens of kilometers away. Fortunately, no one died. The final investigation concluded that the cause was not a pressure regulator valve failure but an installation problem with an end plug component of the high-pressure hydrogen storage system — of the four related bolts, two were not tightened to the specified torque, causing the seal to gradually fail. High-pressure hydrogen was continuously released, forming a flammable hydrogen cloud around the equipment, which burned and generated a pressure shock when it reached a suitable concentration and encountered an ignition source. The investigation also specifically emphasized that no pressure vessel ruptured; what actually happened was the ignition of a leaked and open-space hydrogen-air mixture.

There are two lessons for the industry from this accident. The first comes from the conclusions of the DSB investigation: the chain of the accident was not 'a worker failed to tighten a bolt, so the hydrogen station exploded,' but closer to an assembly error occurring, quality control not detecting it, pre-commissioning leak and pressure tests failing to catch the problem, the equipment being put into formal operation, and eventually a leak occurring. In other words, it was not just one part that failed, but all inspection stages simultaneously failing to detect it. There are deficiencies in the company's internal safety management system, and there were flaws in the leak and pressure tests for the hydrogen refueling facility before it was officially put into use; these two points represent institutional gaps.

The second layer comes from misinformation itself. Hydrogen energy is an industry that heavily depends on public trust. When an accident occurs at a hydrogen station, the public often sees not a particular bolt, but merely the words 'hydrogen.' A news report about an accident with incorrect date or location can have an actual impact far greater than that of an ordinary industrial accident. For the industry, a mature safety culture is not about promoting 'hydrogen is absolutely safe,' but about analyzing each accident in detail: which component failed, why the detection did not catch it, why the system did not prevent the accident from escalating, and finally, how to ensure it does not happen again.

It should be noted that the incidents cited here are existing accidents from 2019 used as technical analysis samples. No credible report of a hydrogen station explosion in Norway on September 3, 2026, was found in public information, and the authenticity of the report from South Korean media is subject to obvious doubts in public channels. The facts of the incident should be based on the official investigation conclusion.

5. Hydrogen balloons on the residential side: cognitive deficiencies beyond technical flaws.

At 2:47 a.m. on March 14, 2026, a fire broke out in a residential home in Hongkou District, Shanghai. The fire originated in the hallway area of the home, with a burn area of approximately 10 square meters. On-site inspections and evidence analysis by the fire department determined that the cause of the fire was the household making hydrogen balloons at home. During the process, the hydrogen balloons and leaking hydrogen came into contact with an ignition source, instantly triggering combustion and fire. The accident resulted in two injuries, severe damage to interior decorations, furniture, appliances, and other items, and affected surrounding residences and common areas of the building corridors due to high-temperature smoke and firefighting water.

The technical complexity of such accidents is not high, but they reveal another issue: awareness. Hydrogen is colorless, odorless, extremely light, and highly diffusive, and it is a dangerous chemical explicitly regulated by the state. Making, storing, and using it at home puts one in a completely unprotected hazardous operation state—without explosion-proof, ventilation, pressure relief, or flame-retardant protection measures. However, many people mistakenly believe that 'making and playing with a small amount of hydrogen balloons at home is risk-free,' ignoring the highly flammable and explosive nature of hydrogen.

From an industry perspective, the significance of this is: when hydrogen energy spreads to civilian end-use, the distribution of accident risks will overflow from professional sites to non-professional populations. These types of risks cannot be solved by equipment upgrades alone; they can only be addressed simultaneously through standards, access regulations, and public education. This is also why the application conditions for comprehensive hydrogen energy pilot projects explicitly require that projects comply with national safety, energy-saving, environmental protection, and quality regulations and standards. The cities applying must ensure that the project implementation does not trigger safety risks, and it is clearly stated that 'use hydrogen where appropriate, and avoid all scenarios rushing in with low-level repetitive construction.'

6. Turn lessons into clauses: three actions that can be immediately implemented

The value of accident reports lies in their ability to be converted into clauses. The four warnings given in the accident report by Chongqing University happen to be management actions that can be implemented directly. Here, they are matched with technical requirements.

The first rule is that equipment maintenance must not exceed its service life. Pipelines, valves, and other wear-prone parts must be replaced strictly according to the maintenance schedule, with clear individual responsibility. This rule has significance for hydrogen systems beyond the laboratory—the replacement cycle of hoses, seals, and filter elements in industrial installations is a major control measure against hydrogen embrittlement and leaks, and the failure of these components is precisely gradual and invisible.

The second rule is that when a leak is detected, you must 'shut off the source and cut the power' simultaneously. Closing the valve first but not cutting the power, leaving the hot spots intact, is equivalent to only doing half the job. In industrial settings, this rule should be expanded into a set of interlock logic: when a concentration rise is detected, shut off the gas source valve, cut off potential ignition sources (heating circuits, heat tracing, lighting according to zoning strategies), start forced ventilation, and if necessary, interlock the shutdown. This is where the value of automation systems lies—human first responses are unreliable, while interlocks do not hesitate.

Article 3 states that risk identification must cover multi-factor coupling scenarios. Combination risks such as hydrogen plus high temperature plus ethanol should be identified in advance and special control plans formulated. This article is the most universally applicable: situations where hydrogen energy coexists with high-temperature surfaces, organic solvents, dust, and oxidizers are not uncommon in chemical plants and laboratories, whereas single-factor risk lists often fail to cover combinations.

Article 4 states that emergency facilities must be fully equipped and effective. Combustible gas detectors, explosion-proof ventilation systems, and firefighting equipment are all indispensable. 'Fully equipped' addresses whether they are available, while 'effective' addresses whether they remain continuously functional—regular calibration and avoiding false alarms being silenced are the most common ways detectors fail.

Looking at these four points together with the four lines of defense outlined earlier, the hydrogen safety system is actually quite simple: the source requires selecting the right materials and replacing parts on schedule, detection must be continuous and able to automatically interlock, ignition sources should be eliminated through engineering means rather than relying on human attention, and accidents should be mitigated through ventilation and isolation. The difficulty is not in the technology—technical solutions for hydrogen refueling stations, hydrogen storage containers, and pipeline monitoring are already quite mature—the difficulty lies in ensuring that each measure remains effective in a real-world environment where some people are lazy, some find it troublesome, and some think, 'It was fine last time.'

【In-depth Analysis of Failures and Defects · Causes】The common structure of the four accidents is: hydrogen itself does not have inherently dangerous characteristics. Its trouble lies in four physical properties combined: a wide flammable concentration range from 4.1% to 74.1%, volumetric energy density only one-third that of natural gas, rapid permeability and diffusion, and extremely low ignition energy. These four properties together mean that any management lapse can be amplified into a flash explosion. The Chongqing University accident, which resulted in 1 death and 3 injuries, involved the simultaneous failure of four lines of defense — hose overdue for service, accumulation on the furnace top, heating wire hotspots, valves closed without cutting power, plus no alarm system and no explosion-proof ventilation. The chain in the Gansu accident, which caused 3 deaths and 2 injuries, involved overpressure, leakage, static ignition, and secondary combustion, occurring during a high-risk period of resuming work. The cause of the hydrogen trailer accident in California, which resulted in 1 death and 1 injury, has not yet been officially released. Judgement: The core issue of hydrogen safety is not 'whether hydrogen is safe,' but 'whether the next line of defense can still stop a failure when any one line fails.' In the Kjørbo accident, this layer was ignored — tightened bolts are not a line of defense; quality control of assembly, airtight and pressure testing before commissioning, and online monitoring during operation are.

The references in this article are sourced from China National Petroleum Corporation, China Southern Power Grid (enterprises); the National Development and Reform Commission, National Energy Administration, National Standard Information Public Service Platform, China National Institute of Standardization, Sensor Experts Network (institutions); China Petroleum Network, China Energy Network, Platow, Energy Magazine, Anqing Fire Department, Xinhua Net (media), and other public information.