The whole case, in one document. 全部论证，一份文档讲完。
Why Svalbard, why 20 to 30 metres deep, why stainless modules, why a DAO, and why the carbon dioxide budget, not the structure, is the binding constraint. Published under CC BY-SA 4.0. 为什么是斯瓦尔巴，为什么是地下 20 至 30 米，为什么用不锈钢模块，为什么采用 DAO 治理，以及为什么真正的约束是二氧化碳预算而不是建筑结构。以 CC BY-SA 4.0 发布。
ARK BUNKER is a 200 person underground ecological community at 79 degrees north, governed as a DAO and designed as a reusable open standard. It is not a luxury shelter for a few individuals. It is a minimum viable civilisation cell: a place where energy, water, food, air and decision making all close their loop inside a rock shield, and where the governance code is as carefully engineered as the life support.
ARK BUNKER 是位于北纬 79°、以 DAO 治理的 200 人地下生态社区，同时被设计为可复用的开放标准。它不是少数人的豪华避难所，而是一个"最小可行文明单元"：能源、水、食物、空气与决策，全部在岩层屏蔽内闭合循环，治理代码与生命维持系统同等精心地被工程化。
The feasibility study scores four candidate regions across eight weighted dimensions. Svalbard wins on legal access rather than on comfort: the 1920 Svalbard Treaty gives nationals of 46 states, China included, equal rights to residence and business, and the archipelago is demilitarised and seismically quiet. Permafrost 500 metres thick provides a stable excavation medium at a constant minus 4 degrees Celsius below 12 metres.
可行性研究在八个加权维度上对四个候选地区打分。斯瓦尔巴胜出的原因是法律准入而非舒适度：1920 年《斯瓦尔巴条约》赋予 46 个缔约国国民（含中国）平等的居住与经营权，群岛非军事化且地震静默。500 米厚的永久冻土提供了稳定的开挖介质，12 米以下常年保持零下 4 摄氏度。
The single most important finding is counter-intuitive and it drives the entire engineering budget. Plant-based carbon dioxide absorption covers about 3.4 percent of demand. In a sealed volume of roughly 15,000 cubic metres, 200 people reach 1,000 ppm in about 1.3 hours without active scrubbing. Mechanical carbon dioxide removal is therefore not a subsystem, it is the first priority item, and it is specified with full redundancy from day one.
最重要的发现与直觉相反，并决定了整个工程预算。植物对二氧化碳的吸收仅覆盖约 3.4% 的需求。在约 15,000 立方米的封闭空间内，若无主动捕集，200 人约 1.3 小时即达到 1000ppm。因此机械式二氧化碳脱除不是子系统，而是第一优先项，自第一天起就按全冗余配置。
Eight dimensions, four regions, one winner. 八个维度，四个地区，一个胜出。
Bore the rock, drop in stainless rooms. 先挖岩层，再落入不锈钢房间。
Excavation 开挖
- Method
- 工法
- 17.5 m shield TBM
- 17.5 米盾构 TBM
- Reference machine
- 参考机型
- Shanhe class, Yellow River crossing
- 山河号，黄河隧道
- Proven advance rate
- 实测进尺
- 718 m / month
- Reference depth
- 参考埋深
- 106 m · 10.6 bar
- Design depth
- 设计埋深
- 20 - 30 m
Habitat modules 居住模块
- System
- 体系
- Broad B-CORE stainless core panel
- 远大 B-CORE 不锈钢芯板
- Areal weight
- 面密度
- 34 kg / m2
- Design life
- 设计寿命
- 1,000 years
- Unit cost
- 单位造价
- $1,125 / m2
- Module count
- 模块数量
- 80
Shielding improves with depth but so does cost, water ingress risk and emergency egress time. Between 20 and 30 metres the rock gives a radiation protection factor above ten to the eleventh, full blast decoupling and a stable thermal sink, while stair egress stays under two minutes. Deeper buys very little and costs a great deal. 屏蔽效果随深度提升，但成本、涌水风险与应急疏散时间同步上升。20 至 30 米区间内，岩层可提供高于 10 的 11 次方的辐射防护因子、完全的爆炸解耦与稳定的热汇，同时楼梯疏散时间保持在两分钟以内。再深收益极小而代价极大。
Plants will not save you. Scrubbers will. 救你的不是植物，是洗涤装置。
The romantic version of a sealed bunker has a greenhouse doing the air recycling. The measured version does not. Across the planned growing area, photosynthesis removes about 3.4 percent of the carbon dioxide the community produces. The remaining 96.6 percent has to be removed mechanically.
浪漫版本的封闭堡垒里，温室负责空气循环；实测版本不是。在规划的种植面积上，光合作用仅能去除社区产生二氧化碳的约 3.4%，剩余 96.6% 必须依靠机械脱除。
Modelled on a sealed volume near 15,000 cubic metres, 200 residents drive the concentration from ambient to 1,000 ppm in roughly 1.3 hours with no active removal. At 1,000 ppm, cognitive performance already measurably degrades, which is the wrong direction for the people who have to fix the problem.
按约 15,000 立方米封闭容积建模，200 名居民在无主动脱除时约 1.3 小时内将浓度从本底推高至 1000ppm。而 1000ppm 时人的认知表现已出现可测量的下降，这对必须解决问题的人来说方向完全错误。
The design therefore specifies a molecular sieve carbon dioxide removal assembly as the first equipment item, sized for full demand plus margin, installed as two independent trains so that a single failure never removes more than half of capacity. The greenhouse stays, because it is valuable for food, water polishing and morale. It is simply not credited with air.
因此设计将分子筛二氧化碳脱除装置列为第一设备项，按满负荷加余量选型，并安装为两条独立通道，使单点故障永远不会损失超过一半能力。温室保留，因为它在食物、水深度处理与心理层面很有价值，但空气中的功劳不记在它头上。
Energy, water, food, air. 能源、水、食物、空气。
Photovoltaic array sized for the summer surplus, with the polar night carried by stored energy and generators. The thermal sink at minus 4 degrees removes most cooling demand.
光伏阵列按夏季盈余配置，极夜期间由储能与发电机承担。零下 4 度的热汇消除了大部分冷负荷。
Closed loop recovery target across grey water, condensate and humidity capture, with meltwater as a seasonal top up and a ninety day strategic reserve.
灰水、冷凝水与湿度捕集的闭环回收目标，以融雪水季节性补充，并保留 90 天战略储备。
Vegetable output under LED cultivation, covering fresh produce and a meaningful share of calories, with stored staples for a full year of autonomy.
LED 种植下的蔬菜产出，覆盖新鲜蔬果与可观的热量份额，并储备可供一整年自给的主粮。
HEPA H13 filtration with CBRN rating on all intake paths, positive pressure in the habitat, and two independent carbon dioxide removal trains.
全部进风通路配置 HEPA H13 并达 CBRN 等级，居住区维持正压，二氧化碳脱除双通道独立。
A 10 layer cube that replaces the field. 用十层立方取代耕地。
Roots are suspended in open air inside a sealed 6.0 x 6.0 x 4.8 m module and fed by a 5 to 20 micron nutrient mist at 0.5 to 0.8 MPa. Ten stacked layers convert 173 m³ of rock cavity into 240 m² of production surface, a tenfold land multiplier. Two modules run as N+1 and deliver about 108 kg of fresh produce per day, against a 200 person vegetable requirement of 80 kg per day. 根系悬空于密闭的 6.0 x 6.0 x 4.8 m 立方舱内，由 0.5 至 0.8 MPa 压力下的 5 至 20 微米营养雾滴供养。十层堆叠把 173 m³ 岩腔转化为 240 m² 生产面积，土地倍率 10 倍。双舱按 N+1 运行，日产约 108 kg 鲜产，对应 200 人每日 80 kg 蔬菜需求。
| Tier | 分层 | Layers | 层数 | Area | 面积 | Spacing | 层距 | DLI | 需光量 | Cycle | 周期 | Yield kg/m²/d | 产量 kg/㎡/天 |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| A · Leafy and herb | A · 叶菜与草本 | L01-05 | 120 m² | 0.45 m | 17 | 25-30 d | 25-30 天 | 0.28 | |||||
| B · Fruiting | B · 果菜 | L06-08 | 72 m² | 0.90 m | 25 | 60-90 d | 60-90 天 | 0.15 | |||||
| C · Tuber and dwarf tree | C · 根茎与矮化果树 | L09-10 | 48 m² | 1.20 m | 20 | 90-120 d | 90-120 天 | 0.20 | |||||
| Total per module | 单舱合计 | 10 | 240 m² | 4.8 m | 17-25 | 25-120 d | 25-120 天 | 0.225 |
About 95% less water than open field soil and roughly 40% less than DFT or NFT hydroponics, with nutrient demand down about 60%. No substrate, no discharge.
较露地土壤节水约 95%，较 DFT / NFT 水培再降约 40%，肥料需求下降约 60%。无基质、无排放。
Lighting energy is set by the daily light integral, not by the growing method. Two modules draw about 1,018 kWh/d, close to 30% of site generation. Aeroponics saves water, nutrient, space and labour. It does not save electricity.
照明能耗由需光量（DLI）决定，与栽培方式无关。双舱约 1,018 kWh/天，接近全站发电量的 30%。气雾省水、省肥、省空间、省人工，但不省电。
The concentrated stream released when the CDRA sieve regenerates is metered into the chamber at 1,000 to 1,200 ppm, a 20 to 30% yield gain for C3 crops. Plants still absorb only 3.4% of human output, so CDRA stays the primary control.
CDRA 分子筛再生产生的高浓度气流按 1,000 至 1,200 ppm 定量注入舱内，C3 作物增产 20% 至 30%。植物仍只吸收人体呼出量的 3.4%，CDRA 依旧是第一控制手段。
The polar night lasts about 84 days. The operating model assumes a six month rotation, circadian lighting at high intensity during the dark season, and a mandatory psychological screening and support programme. Engineering that ignores isolation fails at month four, not at year ten. 极夜约持续 84 天。运营模型设定六个月轮换制、暗季高强度昼夜节律照明，以及强制的心理筛查与支持计划。忽视隔离的工程会在第四个月失败，而不是第十年。
23.28 million dollars, and where it comes from. 2328 万美元，以及它的来源。
- Internal rate of return
- 内部收益率
- 5.2%
- Payback period
- 投资回收期
- 14.6 years
- Annual operating cost
- 年度运营成本
- $1.2M
- Breakeven occupancy
- 盈亏平衡入住率
- 60%
This is infrastructure, not a token. Returns come from residency fees, research contracts and energy services. Below 60 percent occupancy the site loses money, so membership recruitment is a financial variable, not a marketing one. 这是基础设施，不是代币。收益来自居住费、科研合同与能源服务。入住率低于 60% 即亏损，因此会员招募是财务变量，而不是市场动作。
What kills this project, and what we do about it. 什么会杀死这个项目，以及我们如何应对。
| Risk | 风险 | Severity | 等级 | Mitigation | 缓解措施 |
|---|---|---|---|---|---|
| Permit refusal | 许可被拒 | High | 高 | Eighteen month pre-engagement with the Norwegian Environment Agency, alternative sites pre-scored, capital released in tranches. | 与挪威环境局预沟通十八个月，备选场地预先打分，资金分批释放。 |
| Cost overrun on tunnelling | 隧道成本超支 | High | 高 | Fixed price contract with an experienced TBM supplier, 15 percent contingency, geotechnical survey completed before mobilisation. | 与成熟 TBM 供应商签固定总价合同，15% 不可预见费，进场前完成岩土详勘。 |
| Carbon dioxide system failure | 二氧化碳系统失效 | Critical | 极高 | Two independent removal trains, each sized above demand, plus chemical backup cartridges for seventy two hours. | 两条独立脱除通道，单条即超需求配置，另备 72 小时化学吸附备份。 |
| Governance capture | 治理被俘获 | Medium | 中 | Reputation cannot be bought, vote weight capped at 3 percent, signers rotate annually, court can freeze any transfer. | 声誉不可购买，票权上限 3%，签名人年度轮换，法院可冻结任何转账。 |
| Psychological attrition | 心理流失 | Medium | 中 | Six month rotation, 84 day polar night lighting protocol, screening before entry and continuous support on site. | 六个月轮换，84 天极夜照明方案，入场前筛查与在地在持续支持。 |
| Regulatory change on DAO entities | DAO 实体监管变化 | Low | 低 | Dual structure: Marshall Islands DAO LLC with a Swiss Association as a conventional fallback counterparty. | 双重结构：马绍尔群岛 DAO LLC，并以瑞士协会作为传统备用签约主体。 |
Everything above is open. Help us build it. 以上内容全部开放。请帮我们把它建成。
This document is released under CC BY-SA 4.0. Use it, challenge it, improve it. If you want to be inside rather than outside, the founding round is open. 本文档以 CC BY-SA 4.0 发布。请随意使用、质疑与改进。若你希望身处其中而非旁观，创始轮次正在开放。