Energy never
disappears.
It leaks.
能源从不消失。
它只是漏掉了。
Your plant buys 100 units of energy every day. Roughly 40 end up in the product. The rest walks out through stacks, leaks, throttled valves and machines left running on Sunday. This station is a walk down that flow — click anything, open it, go deeper, and see what each leak is worth.
工厂每天买进 100 份能量，真正进入产品的大约只有 40 份。其余的从烟囱、漏点、被憋住的阀门，还有周日仍在空转的设备走掉了。这一站，我们沿着这条流走一遍——任何地方都可以点开、再点开、一层层往下挖，看清每一处泄漏值多少钱。
The cheapest megawatt-hour is the one you never had to buy. The IEA calls efficiency the first fuel — it is the only fuel you already own.
最便宜的一度电，是你根本不需要买的那一度。国际能源署把能效称为"第一能源"——它是你早已拥有的唯一燃料。
An EEP is not a project list. It is a loop the plant runs on itself.
EEP 不是一张项目清单，而是工厂对自己持续运行的一个闭环。
A project ends. A program keeps going: it measures, finds the hot spots, ranks them by money and effort, fixes them, then locks the gain in so it cannot quietly drift back. Five moves, repeated forever. Open each one.
项目会结束，项目化的"程"不会：计量、找热点、按钱和难度排序、动手改、再把成果锁住，不让它悄悄退回去。五个动作，循环不止。逐个点开看。
01Meter — you cannot manage a number you don't have计量 — 没有数字，就没有管理
Meter — you cannot manage a number you don't have计量 — 没有数字，就没有管理
One utility bill for the whole site tells you nothing. Sub-meter the four big consumers (compressed air, boiler/gas, chillers, the main production line) and 80% of the mystery disappears.
全厂一张电费单说明不了任何事。给四大用能户（压缩空气、锅炉/燃气、冷冻站、主生产线）装上分表，80% 的谜团就消失了。
Look at a night or a weekend. Whatever the plant still draws when nothing is being made is base load — usually 15–30% of peak, and usually the fastest money in the building.
看一眼夜里或周末。什么都不生产时还在耗的功率就是基础负荷——通常是峰值的 15–30%，也通常是全厂最快见效的钱。
Normalise energy by output: kWh per tonne, per unit, per shift. Absolute kWh lies when production changes; specific energy consumption (SEC) does not.
用产量归一化：每吨、每件、每班的 kWh。产量一变，绝对电量就会骗人；单位产品能耗（SEC）不会。
02Map — walk the flow, mark the hot spots画流 — 走一遍现场，标出热点
Map — walk the flow, mark the hot spots画流 — 走一遍现场，标出热点
Energy is a flow, so draw it as one: fuel and power in on the left, product and losses out on the right. The fat arrows going nowhere are your program.
能源是一条流，就把它画成流：左边是进来的燃料和电，右边是产品和损失。那些又粗又流向"无处"的箭头，就是你的项目。
Run an energy treasure hunt: a mixed team of operators, maintenance and finance walks the plant on a weekend and at 3 a.m. Teams typically surface 5–15% of site energy in low/no-cost finds.
组织一次能源寻宝：由操作工、维修和财务混编的小队，在周末和凌晨三点各走一遍现场。这类走查通常能挖出全厂 5–15% 的低成本/零成本节能点。
Take ears, an ultrasonic leak detector and a thermal camera. Leaks you can hear, insulation you can feel, and traps that blow through — that is a full day of findings.
带上耳朵、超声波检漏仪和热像仪。听得见的漏气、摸得到的烫管、吹穿的疏水阀——一天就能记满一本。
03Rank — money, effort, risk, in that order排序 — 按钱、难度、风险，就这个顺序
Rank — money, effort, risk, in that order排序 — 按钱、难度、风险，就这个顺序
Do the free things first — not because they are big, but because they buy you credibility and the budget for the big ones.
先做不花钱的——不是因为它大，而是因为它替你换来信任和后面大项目的预算。
Score every idea: annual saving ÷ install cost = payback. Under 1 year, just do it. 1–3 years, build the case. Over 5 years, park it until the equipment dies anyway — then buy the efficient one.
给每个点子打分：年节省 ÷ 投资 = 回收期。小于 1 年，直接干；1–3 年，做商业论证；超过 5 年，先放着——等设备到寿更换时，直接买高效型号。
Always ask "what is the risk to production?" An efficiency measure that trips the line once costs more than it will ever save.
永远追问一句"对生产的风险是什么？"一次停线，就能吃掉这项措施一辈子的节省。
04Act — fix, tune, then replace动手 — 先修、再调、最后换
Act — fix, tune, then replace动手 — 先修、再调、最后换
Never buy a bigger machine to feed a leaking system. Seal it, set it right, and only then size the new equipment — the new one is often a size smaller and cheaper.
绝不要为一个漏气的系统去买更大的机器。先堵漏、调对参数，再选型——往往能小一档，还更便宜。
Order of attack: turn it off → turn it down → recover the heat → replace the machine. Cost rises at each step; so does the paperwork.
进攻顺序：先关掉 → 再调低 → 回收余热 → 最后换设备。每往后一步，成本和审批都在涨。
Give every action one owner and one date. "Engineering will look at it" is how a program dies.
每项措施只写一个责任人和一个日期。"工程部会看一下"——项目就是这样死掉的。
05Lock it in — savings drift back without a system锁住 — 没有体系，节省会自己溜回去
Lock it in — savings drift back without a system锁住 — 没有体系，节省会自己溜回去
A setpoint someone changed back, a VSD switched to bypass, a schedule wiped by a PLC update — most "lost" savings are not technical failures, they are memory failures.
被人改回去的设定值、被切到工频的变频器、被 PLC 升级抹掉的时间表——大多数"消失的节能量"不是技术问题，是记忆问题。
Put SEC on the daily production board next to output and scrap. What gets reported at the morning meeting stays fixed.
把单位产品能耗放到日报看板上，紧挨产量和废品率。早会上要汇报的东西，才不会退回去。
ISO 50001 is the ready-made skeleton for this loop, and it is what most customer and CDP audits will ask you for anyway.
ISO 50001 就是这套闭环的现成骨架，而且客户审核和 CDP 问卷通常正好要它。
Follow the energy down the plant. Every dot is a hot spot — open it.
跟着能量一路往下走。每个圆点都是一个热点——点开它。
This is the path energy takes from the meter to the product. The red badge is the typical loss at that node — how much of what enters never leaves as useful work. Open a node to see where it leaks and exactly what to do about it.
这是能量从电表走到产品的路径。红色徽标是该节点的典型损失——进来的能量里有多少没有变成有用功。点开任一节点，看它漏在哪里，以及具体该做什么。
Where the bill is written账单在这里写成
3–10% of bill 电费
Where the bill is written账单在这里写成
Demand peaks you pay for all year · poor power factor penalties · base load humming through nights and holidays · tariff bands nobody has re-read since installation.
全年都在付费的需量尖峰 · 功率因数罚款 · 夜间和假期仍在空耗的基础负荷 · 装机以来再没人看过的电价时段。
Base-load audit at 3 a.m. 2–8% of site · power-factor correction to >0.95 1–3% of bill · shift one batch process off the peak window 3–8% of bill.
凌晨三点做一次基荷排查 全厂 2–8% · 功率因数补偿到 >0.95 电费 1–3% · 把一个批次工序挪出尖峰时段 电费 3–8%。
The biggest single fuel user全厂最大的单一燃料用户
15–25% up the stack 随烟气走
The biggest single fuel user全厂最大的单一燃料用户
Excess air (stack O₂ above 4%) · flue gas leaving at 200 °C+ with no economizer · continuous blowdown dumped to drain · boilers cycling on and off at low load.
过量空气（烟气含氧超过 4%）· 200 °C 以上的烟气直接排空、没有省煤器 · 连续排污直接进地沟 · 低负荷下频繁启停。
O₂ trim / combustion tune-up 1–3% fuel · stack economizer on feedwater 4–6% fuel · blowdown heat recovery 1–2% fuel · shut down a spare boiler kept hot for nothing 1–3% fuel.
氧量自动调节 / 燃烧调整 燃料 1–3% · 给给水加省煤器 燃料 4–6% · 排污热回收 燃料 1–2% · 停掉那台"空热着"的备用炉 燃料 1–3%。
Every 22 °C drop in flue gas temperature ≈ 1% fuel saved. Every 10 °C rise in feedwater temperature ≈ 1% fuel saved.
烟温每降 22 °C ≈ 省 1% 燃料；给水温度每升 10 °C ≈ 省 1% 燃料。
The quiet 20% between boiler and use从锅炉到用汽点，悄悄丢掉的 20%
10–20% of steam 蒸汽量
The quiet 20% between boiler and use从锅炉到用汽点，悄悄丢掉的 20%
Failed-open traps blowing live steam · bare valves, flanges and elbows nobody re-insulated after the last repair · condensate running to drain instead of back to the deaerator · dead legs feeding equipment removed years ago.
失效敞开、直接吹汽的疏水阀 · 上次检修后再没包回保温的阀门、法兰和弯头 · 冷凝水直排地沟而不是回到除氧器 · 通向早已拆除设备的死管段。
Annual ultrasonic trap survey + repair 5–10% of steam · insulate bare valves & lines (removable jackets) 3–5% fuel · return condensate 10–15% fuel · isolate dead legs 1–3%.
每年一次超声波疏水阀普查并修复 蒸汽量 5–10% · 裸阀门管道加可拆保温套 燃料 3–5% · 冷凝水回收 燃料 10–15% · 切除死管段 1–3%。
A trap population with more than 10% failures is out of control. Best-run plants sit under 3%.
疏水阀失效率超过 10%，说明已经失控；管得好的工厂在 3% 以下。
Heat that leaves before it works还没干活就跑掉的热
30–50% in exhaust 随排风
Heat that leaves before it works还没干活就跑掉的热
Over-drying past the spec moisture · exhaust fans pulling far more air than the process needs · doors and openings left open between batches · damaged refractory and hot shells you can feel from a metre away.
干过头，含水率远低于标准 · 排风量远超工艺需要 · 批次之间炉门/开口敞着 · 破损的耐火层和一米外就能感到烫的外壳。
Control to the moisture endpoint, not the timer 5–15% · recover exhaust heat to preheat make-up air 10–25% of oven fuel · trim exhaust flow / add O₂ or humidity control 5–10% · seal and re-insulate 2–5%.
按含水率终点控制，而不是按定时器 5–15% · 用排风余热预热补风 烘箱燃料 10–25% · 削减排风量／加装氧量或湿度控制 5–10% · 密封与重做保温 2–5%。
The most expensive utility in the building全厂最贵的公用工程
80–90% to heat 变成热
The most expensive utility in the building全厂最贵的公用工程
Only 10–15% of the electricity fed to a compressor reaches the tool as useful work. On top of that: leaks (typically 20–30% of output), pressure set 1–2 bar higher than any machine needs, open blow-guns used for cleaning, and compressors idling unloaded at 30–40% of full power.
喂给空压机的电，只有 10–15% 变成气动工具端的有用功。此外还有：泄漏（通常占产气量 20–30%）、比任何设备都高出 1–2 bar 的压力设定、拿吹枪当扫把用，以及卸载空转仍吃满载 30–40% 功率的机组。
Quarterly ultrasonic leak hunt + tagged repair 20–30% of air energy · drop system pressure by 1 bar ≈7% power · VSD on the trim compressor + sequencing control 15–35% · recover compressor heat to hot water or space heat up to 90% of input · replace open blowing with engineered nozzles 30–50% of that use.
每季度超声波查漏并挂牌整改 空压电耗 20–30% · 系统压力降 1 bar ≈7% 功率 · 调峰机上变频＋群控 15–35% · 压缩热回收做热水或采暖 可达输入的 90% · 把开口吹扫改为专用喷嘴 该用途 30–50%。
If you can hear it in a running factory, it is big. Walk the plant on a silent Sunday and every hiss is money.
在轰鸣的车间里还能听见的漏点，一定不小。挑一个安静的周日走一圈，每一声"嘶——"都是钱。
Two-thirds of every industrial kWh工业用电的三分之二
20–50% throttled away 被节流吃掉
Two-thirds of every industrial kWh工业用电的三分之二
Motor-driven systems take about 65–70% of industrial electricity — and most of them run flat out into a half-closed valve or damper. Oversized pumps, untrimmed impellers, and "we always ran it at 100%" are the pattern.
电机驱动系统占工业用电约 65–70%，而其中大多数都在满速运行、再用一个半关的阀门或风门把流量憋回去。选型过大、叶轮没切削、"我们一直就是开 100%"，是最常见的画像。
VSD on any centrifugal fan or pump with variable demand 20–50% · open the valve, slow the motor · trim or replace the impeller to the real duty 10–25% · buy IE4 at failure, never rewind twice 2–6% per motor · fix belt slip / switch to synchronous belts 2–4%.
需求波动的离心风机水泵一律上变频 20–50% · 把阀开大，把电机调慢 · 按真实工况切削或更换叶轮 10–25% · 坏了就换 IE4，别反复重绕 单台 2–6% · 治皮带打滑／改同步带 2–4%。
Affinity law: power goes with the cube of speed. Run a fan at 80% speed and you use about half the power. See the step-by-step maths at Stop 05.
相似定律：功率与转速的三次方成正比。风机降到 80% 转速，功率约剩一半。第 05 站有逐步算给你看。
Comfort that runs when nobody is there没人的时候还在开的舒适
10–30% avoidable 可避免
Comfort that runs when nobody is there没人的时候还在开的舒适
Chilled water made colder than the process needs · condensers fouled so the compressor works against a higher lift · all chillers part-loaded instead of one running near its sweet spot · cooling and reheating the same air · schedules that expired years ago.
冷冻水做得比工艺需要还冷 · 冷凝器结垢、压缩机被迫顶着更大压差 · 所有冷机都在低负荷分摊，而不是让一台跑在最佳点 · 同一股空气先冷却再加热 · 几年前就该失效的时间表。
Raise chilled-water supply temperature 2–3% per °C · clean condensers, hold approach <2 °C ≈2.5% per °C · free cooling / air-side economizer 10–30% of cooling · sequence chillers to best load point 5–15% · fix schedules & setpoints 3–5% per °C. Full checklist at Stop 04.
提高冷冻水供水温度 每 1 °C 省 2–3% · 清洗冷凝器，逼近温差控制在 2 °C 以内 每 1 °C ≈2.5% · 自然冷却／新风经济器 冷量 10–30% · 按最佳负荷点排序运行 5–15% · 修正时间表与设定值 每 1 °C 3–5%。完整清单见第 04 站。
Small share, fastest win占比不大，见效最快
50–70% recoverable 可回收
Small share, fastest win占比不大，见效最快
Old fluorescent or metal-halide high bays, one switch for a whole hall, warehouse aisles lit at production levels, and daylight coming through roof lights straight past lamps nobody dims.
老旧荧光灯或金属卤化物高天棚灯、整个厂房一个开关、仓库通道按生产区照度点亮、天窗自然光旁边的灯却从不调暗。
LED retrofit 40–60% · add occupancy & daylight control on top a further 20–30% · zone the switching to how people actually work · re-set lux to the task, not the whole hall 10–20%.
LED 改造 40–60% · 再叠加人体感应与自然光调光 再省 20–30% · 按人真实的作业方式分区控制 · 按作业面而非整个厂房设定照度 10–20%。
Where the other half of your fuel goes你另一半燃料的去向
20–50% of fuel in 进厂燃料
Where the other half of your fuel goes你另一半燃料的去向
Flue gas, oven exhaust, compressor cooling air, hot effluent, cooling-tower plume. All of it was paid for at full price and is being given to the sky.
烟气、烘箱排风、空压机冷却风、高温废水、冷却塔白雾。这些都是按全价买来的能量，正整份送给天空。
Match a hot stream to a cold demand before buying any recovery kit — the pairing is the project. Compressor heat → wash water up to 90% of compressor input · oven exhaust → make-up air 10–25% · flue gas → feedwater 4–6% fuel · high-grade streams → ORC power site-specific.
先把一股热源和一个冷需求配上对，再谈买设备——配对本身就是项目。压缩热 → 清洗热水 可达空压机输入 90% · 烘箱排风 → 补风预热 10–25% · 烟气 → 锅炉给水 燃料 4–6% · 高品位余热 → ORC 发电 因厂而异。
Fourteen measures that work in almost any factory.
十四招，几乎在任何工厂都成立。
Filter by system, then open a card: why it is good, how to think about it, and what to check first. Percentages are typical published ranges — treat them as a starting hypothesis for your own meter, not a promise.
先按系统筛选，再点开卡片：好在哪、该怎么想、先查什么。百分比是公开资料的典型区间——把它当作对你自家电表的初始假设，而不是承诺。
Compressed air压缩空气01 · Hunt the leaks, tag them, fix them01 · 查漏、挂牌、修好
20–30%payback < 6 months回收期 <6 个月
01 · Hunt the leaks, tag them, fix them01 · 查漏、挂牌、修好
An untreated plant leaks 20–30% of everything it compresses, 8,760 hours a year, whether or not anyone is working. It is the purest waste in the building.
没治理过的工厂，压出来的气有 20–30% 在漏，一年 8,760 小时不停，不管有没有人上班。这是全厂最"纯粹"的浪费。
A leak is not an air problem, it is a maintenance-priority problem. Tag every leak with a number and a cost per year and the work order stops being optional.
漏气不是气的问题，是维修优先级的问题。给每个漏点挂上编号和"年损失金额"，工单就不再是可有可无的了。
Shut all production, keep the compressor on, and time the loaded cycles. Load time ÷ total time = your leak rate. Good plants are under 10%.
停掉全部生产、只开空压机，记录加载时间。加载时间 ÷ 总时间 = 泄漏率。管得好的工厂低于 10%。
Compressed air压缩空气02 · Drop system pressure by 1 bar02 · 系统压力降 1 bar
≈7%free, today零成本，今天就能做
02 · Drop system pressure by 1 bar02 · 系统压力降 1 bar
Compressor power falls about 7% for every bar of discharge pressure removed — and lower pressure also shrinks every leak in the system, so the real saving is bigger.
排气压力每降 1 bar，空压机功率约降 7%；而且压力一低，全系统每个漏点也跟着变小，实际节省更多。
Most plants run high to feed one thirsty machine at the end of an undersized pipe. Fix that one pipe, and the other 99% of the plant can live 1 bar lower.
多数工厂把压力开高，只是为了喂饱末端某台"渴"的设备。把那一段管径改好，其余 99% 的设备完全可以低 1 bar 运行。
Log pressure at the furthest machine, not at the compressor. Lower the setpoint 0.2 bar per week until someone complains — then go back one step.
记录最远端设备处的压力，而不是空压机出口。每周下调 0.2 bar，直到有人反映，再退回一档。
Compressed air压缩空气03 · Recover the compressor heat03 · 回收压缩热
up to 90% of input可达输入 90%1–2 yr年
03 · Recover the compressor heat03 · 回收压缩热
Nearly all the electricity a compressor eats becomes heat. If you have any hot water or space-heating demand nearby, that heat is already paid for.
空压机吃进的电几乎全变成了热。只要附近有热水或采暖需求，这份热等于已经付过钱了。
Recovery projects fail on the demand side, not the source side. Find the cold user first — wash bay, boiler make-up, staff showers — then size the heat exchanger.
余热项目通常不是败在热源，而是败在需求端。先找到冷用户——清洗、锅炉补水、员工淋浴——再去选换热器。
Is the compressor room the hottest place in the plant in winter, with the extract fan running? That is the whole business case, standing there.
冬天空压站是不是全厂最热、排风扇还在拼命转？——商业论证就站在那儿。
Motors电机04 · Put a VSD on fans and pumps04 · 风机水泵上变频
20–50%1–3 yr年
04 · Put a VSD on fans and pumps04 · 风机水泵上变频
Power rises with the cube of speed. Slow a centrifugal machine 20% and you use roughly half the energy — for the same job, because the job never needed full flow.
功率随转速的三次方变化。离心式设备降速 20%，能耗约剩一半——活儿还是那些活儿，因为本来就不需要满流量。
Wherever there is a throttling valve or a damper, there is a VSD business case. The valve is literally a device for converting your electricity into noise.
哪里有节流阀或风门，哪里就有变频器的商业理由。那个阀门本质上是一台"把电变成噪音"的机器。
Beware constant-pressure duties (lifting water high, positive-displacement pumps) — there the cube law does not apply and savings are much smaller.
注意恒压工况（高扬程提升、容积泵）——那里三次方定律不成立，节能量小得多。
Motors电机05 · Right-size and go IE4 at failure05 · 按实际选型，坏了就上 IE4
2–6% / motor / 台at replacement随更换发生
05 · Right-size and go IE4 at failure05 · 按实际选型，坏了就上 IE4
A motor costs its own purchase price in electricity every few weeks of running. Efficiency class is the cheapest decision you will ever make about it.
一台电机跑上几周的电费，就等于它自己的售价。效率等级，是你对它做出的最便宜的一个决定。
Never rewind the same motor twice — each rewind typically costs 0.5–1% of efficiency. Pre-approve IE4 replacements so the night shift doesn't fit whatever is on the shelf.
同一台电机别重绕两次——每次重绕通常损失 0.5–1% 效率。提前批准 IE4 备件清单，免得夜班随手装上货架上那台。
Clamp the current: a motor loaded below 40% of nameplate is oversized, running at poor efficiency and poor power factor.
用钳表测电流：负载率低于铭牌 40% 的电机就是选大了，效率和功率因数都很差。
Steam & heat蒸汽与热06 · Survey and repair steam traps06 · 疏水阀普查与修复
5–10% of steam 蒸汽< 1 yr年
06 · Survey and repair steam traps06 · 疏水阀普查与修复
A single trap stuck open can blow away thousands of dollars of steam a year, silently, inside a lagged pipe where nobody sees it.
一只卡在敞开位的疏水阀，一年就能悄悄吹掉几千美元的蒸汽——藏在保温层里，谁也看不见。
Traps are consumables, not fittings. Build a numbered register with test dates, exactly like you do for pressure vessels.
疏水阀是易损件，不是管件。像管压力容器那样，建一本带编号和检测日期的台账。
Ultrasonic probe plus surface temperature. Failure rate above 10% means the population is out of control; best practice is under 3%.
超声波探头配合表面测温。失效率超过 10% 说明整体失控；良好实践在 3% 以下。
Steam & heat蒸汽与热07 · Economizer + O₂ trim on the boiler07 · 锅炉加省煤器＋氧量调节
5–8% fuel 燃料1–2 yr年
07 · Economizer + O₂ trim on the boiler07 · 锅炉加省煤器＋氧量调节
Two independent gains from the same flue: burn with less excess air (1–3%), and take the leftover heat into the feedwater (4–6%).
同一条烟道上两笔独立收益：把过量空气降下来（1–3%），再把余热送进给水（4–6%）。
Combustion tuning is an operating cost, not a project — it drifts back within a year. Automatic O₂ trim keeps it, manual tuning does not.
燃烧调整属于运行成本，不是一次性项目——一年内就会漂回去。自动氧量调节能守住，手动调整守不住。
Read stack O₂ and temperature. Natural gas boilers should sit at 2–4% O₂; above 6% you are heating the neighbourhood.
测烟气含氧和排烟温度。燃气锅炉应在 2–4% 含氧；超过 6%，你在给整条街供暖。
Steam & heat蒸汽与热08 · Insulate bare valves, flanges and lines08 · 给裸阀门、法兰、管道保温
3–5% fuel 燃料< 1 yr年
08 · Insulate bare valves, flanges and lines08 · 给裸阀门、法兰、管道保温
One uninsulated 150 mm valve on a 10 bar steam line can waste the energy of several households a year — and there are usually dozens of them.
10 bar 蒸汽管上一只 DN150 裸阀，一年浪费的能量抵得上好几户人家——而这种阀门通常有几十只。
Use removable insulation jackets. Hard-lagged valves get stripped at the next repair and never re-covered — jackets come off and go back in minutes.
用可拆卸保温套。硬做的保温一次检修就被扒掉，从此再没包回去；保温套几分钟就能拆装。
Walk the line with a thermal camera. Anything you cannot comfortably touch is an invoice you are paying every hour.
拿热像仪沿管走一遍。凡是手不敢摸的地方，都是一张按小时结算的账单。
Steam & heat蒸汽与热09 · Return the condensate09 · 冷凝水回收
10–15% fuel 燃料1–2 yr年
09 · Return the condensate09 · 冷凝水回收
Condensate is hot, already treated water. Sending it to drain means you pay twice: once for the heat, once for the chemicals and make-up water.
冷凝水是又热又已处理过的水。排掉等于付两次钱：一次买热，一次买药剂和补水。
Every 10 °C of feedwater temperature is about 1% of boiler fuel. Condensate is the cheapest way to raise it.
给水温度每升 10 °C，约省 1% 锅炉燃料。冷凝水是把它升上去最便宜的办法。
Measure condensate return rate. Under 40% is common and poor; 70% is good; over 90% is world class where the process allows it.
测冷凝水回收率。低于 40% 很常见但很差；70% 算好；工艺允许时超过 90% 属世界级。
Cooling制冷10 · Raise chilled water, clean the condenser10 · 提高冷冻水温、清洗冷凝器
5–20%weeks数周
10 · Raise chilled water, clean the condenser10 · 提高冷冻水温、清洗冷凝器
A chiller's work is proportional to the lift between evaporator and condenser. Every degree you close that gap is 2–3% off the compressor, for free.
冷机做的功正比于蒸发与冷凝之间的温升。这个温差每缩小 1 °C，压缩机就少 2–3% 的功——不花钱。
Ask what the process actually requires, not what the design drawing said in 1998. Many plants make 6 °C water for a 12 °C duty.
问一句工艺真正需要多少度，而不是 1998 年那张设计图写了多少。很多工厂在为 12 °C 的需求制 6 °C 的水。
Log condenser approach temperature weekly. Above 2 °C and rising = fouling. It is a maintenance KPI, not a comfort complaint.
每周记录冷凝器逼近温差。高于 2 °C 且持续上升＝结垢。这是维修 KPI，不是"有人喊热"的投诉。
Cooling制冷11 · Use free cooling when the air is cold11 · 室外冷的时候用自然冷却
10–30% of cooling 冷量1–3 yr年
11 · Use free cooling when the air is cold11 · 室外冷的时候用自然冷却
For a large part of the year, outside air or tower water is already colder than your chilled water setpoint. Running a compressor against that is paying for winter twice.
一年中有相当长时间，室外空气或冷却塔水本来就比你的冷冻水设定温度还低。此时还开压缩机，等于把冬天买了两遍。
Count the hours first. Pull a year of local weather data, count hours below your changeover temperature, and the saving calculates itself.
先数小时数。调一年的当地气象数据，数出低于切换温度的小时数，节能量自己就算出来了。
If a plate heat exchanger already exists but the changeover is manual, it is probably never used. Automate it.
如果板换早就装了、但切换靠人工，那它多半从没用过。把它自动化。
Utilities公用12 · LED plus occupancy and daylight control12 · LED＋感应与自然光控制
50–70%1–2 yr年
12 · LED plus occupancy and daylight control12 · LED＋感应与自然光控制
The lamp swap gets 40–60%. The controls on top get another 20–30% — and in warehouses and plant rooms, controls often save more than the lamps did.
换灯本身省 40–60%，再加控制还能省 20–30%——在仓库和设备间，控制省下的往往比换灯还多。
Sell it on light quality and safety, not on kWh. Better light in an inspection area pays back in defects avoided long before the electricity does.
用光质量和安全去说服，而不是用度电。检验区的光变好，靠减少缺陷带来的回报，远早于电费。
Measure lighting power density (W/m²) before and after, and lux at the task. Over 12 W/m² is old technology; under 5 W/m² is excellent.
改造前后测照明功率密度（W/m²）和作业面照度。超过 12 W/m² 是老技术；低于 5 W/m² 很优秀。
Utilities公用13 · Shutdown discipline — nights, weekends, breaks13 · 停机纪律 — 夜间、周末、休息
2–8% of site 全厂no cost零成本
13 · Shutdown discipline — nights, weekends, breaks13 · 停机纪律 — 夜间、周末、休息
Most plants draw 15–30% of their peak load when nothing is being produced. It is the single largest zero-investment saving in this whole list.
多数工厂在不生产时仍在用掉峰值负荷的 15–30%。这是整张清单里最大的一笔零投资节能。
Write a shutdown sequence like a start-up checklist, laminated, at the machine, with a named owner per shift. Habit beats exhortation.
像开机检查表一样写一份停机顺序，塑封挂在设备旁，每班指定责任人。习惯胜过号召。
Plot 15-minute interval data for a public holiday. Every plateau above zero is an asset nobody switched off.
把一个法定假日的 15 分钟负荷曲线画出来。每一段高于零的平台，都是一台没人关的设备。
Utilities公用14 · Sub-meter, then review it every month14 · 装分表，并且每月复盘
enables everything一切的前提1–2 yr年
14 · Sub-meter, then review it every month14 · 装分表，并且每月复盘
Metering saves nothing by itself. But plants that track SEC monthly typically hold 1–3% per year of continuous improvement that untracked plants give straight back.
计量本身不省一度电。但坚持每月跟踪单位产品能耗的工厂，通常每年能守住 1–3% 的持续改进——不跟踪的工厂则原样退回去。
Meter the four biggest users first. Perfect coverage is a trap; 80% of the energy on 20% of the meters is the goal.
先给四个最大用户装表。追求全覆盖是个陷阱；用 20% 的表看住 80% 的能量才是目标。
Does energy appear on the same weekly report as output, quality and safety? If not, it is not managed — it is just measured.
能耗有没有和产量、质量、安全出现在同一张周报上？没有的话，它只是被"测量"了，并没有被"管理"。
Walk your air-conditioning and chiller plant. Tick what is already true.
走一遍你的空调与冷冻站。已经做到的，打勾。
Ten checkpoints, in the order an auditor would use them. Everything you cannot tick is an opportunity — the panel ranks what is left by how much it is typically worth. This is exactly how a systematic evaluation works: not a guess, a gap list.
十个检查点，按审核员的走查顺序排列。凡是打不上勾的，就是机会——右侧面板会按典型价值把剩下的排序。系统性评估就是这么做的：不是猜，而是一张缺口清单。
Still on the table尚未拿到的部分
Ranges are typical published values for HVAC energy in industrial buildings. Savings do not simply add — the estimate is capped and indicative only.区间取自工业建筑空调能耗的公开典型值。各项节能不能简单相加——此处已设上限，仅供指示。
Two calculations that change how a plant argues about energy.
两笔账，能改变一个工厂讨论能源的方式。
Click Next step and watch the number build. Nothing here is harder than multiplication — the difficulty was only ever that nobody wrote it down.
点下一步，看着数字一层层长出来。这里没有比乘法更难的东西——难只难在从来没人把它写下来。
A. What does one small air leak cost?A. 一个小漏点，一年要多少钱？
A 1.6 mm hole. You would not bend down to look at it.一个 1.6 毫米的小孔，你甚至不会弯腰去看它。
- 01
Start with the hole. A 1.6 mm leak in a 7 bar system flows about 6.5 cfm of free air, continuously.
先看这个孔。7 bar 系统上一个 1.6 mm 的漏点，持续泄漏约 6.5 cfm 自由空气。
1.6 mm @ 7 bar → 6.5 cfm - 02
Convert air to electricity. A typical plant compressor needs about 0.2 kW per cfm (that is 20 kW per 100 cfm).
把空气换成电。典型工厂空压机大约需要 0.2 kW/cfm（即每 100 cfm 耗 20 kW）。
6.5 cfm × 0.2 kW/cfm = 1.3 kW - 03
Leaks don't take breaks. If the compressor runs 8,000 hours a year, that 1.3 kW never stops.
漏气不休息。空压机一年运行 8,000 小时，这 1.3 kW 一刻不停。
1.3 kW × 8,000 h = 10,400 kWh - 04
Put a price on it at 0.10 per kWh.
按每度 0.10 计价。
10,400 kWh × 0.10 = 1,040 / year年 - 05
Now count. A mid-sized plant that has never run a leak program typically has 40–60 leaks of this size or bigger.
现在数一数。一个从没做过查漏的中型工厂，通常有 40–60 个这样大小甚至更大的漏点。
50 × 1,040 ≈ 52,000 / year年 - 06
And the fix? A fitting, a hose clamp, twenty minutes of a fitter's time. Payback is measured in days — which is why leak hunting is the first thing every energy program does.
而修好它？一个接头、一个卡箍、钳工二十分钟。回收期以天计——这就是为什么每个能效项目都从查漏开始。
cost to fix ≈ 20 min · payback ≈ days修复成本 ≈ 20 分钟 · 回收期 ≈ 几天
B. Why slowing a fan 20% saves half the powerB. 风机降速 20%，为什么能省一半电
The affinity law, in six clicks.六次点击，讲清相似定律。
- 01
The law first. For a centrifugal fan or pump: flow follows speed, pressure follows speed squared, and power follows speed cubed.
先说定律。对离心风机或水泵：流量与转速成正比，压头与转速平方成正比，功率与转速三次方成正比。
Q ∝ n · H ∝ n² · P ∝ n³ - 02
Take a real fan: 75 kW, running flat out, 6,000 hours a year. The process only needs 80% of the air.
拿一台真实风机：75 kW，全速运行，一年 6,000 小时。工艺其实只要 80% 的风量。
75 kW · 6,000 h/yr年 · need需求 80% - 03
Today you close a damper to get there. The fan still spins at full speed and only drops to roughly 90% of full power — you are burning electricity in a piece of sheet metal.
现在的做法是关小风门。风机仍在全速转，功率只降到约 90%——你是在用一块铁皮烧电。
75 kW × 0.90 ≈ 67 kW - 04
Now open the damper fully and slow the motor to 80% speed instead. Cube the ratio.
改成：风门全开，把电机降到 80% 转速。把这个比值取三次方。
0.8³ = 0.512 → 75 × 0.512 ≈ 38 kW - 05
Same air delivered. The difference is what the damper was destroying.
送出的风一样多。差额就是那块风门在毁掉的东西。
67 − 38 = 29 kW × 6,000 h ≈ 174,000 kWh - 06
At 0.10 per kWh that is about 17,400 a year from one fan. A VSD and installation at this size runs roughly 15,000–20,000.
按每度 0.10 计，仅这一台风机就是每年约 17,400。这个容量的变频器加安装，大约 15,000–20,000。
payback ≈ 1 year, then free回收期 ≈ 1 年，之后全是净赚
Typical, good, outstanding. Find your row, then move one column right.
一般、良好、卓越。找到你那一行，然后往右挪一格。
A number without a benchmark is just a number. Open a row to see how to measure it — most of these need nothing more expensive than a clamp meter, a thermometer and an hour.
没有基准的数字只是数字。点开任意一行，看怎么测——这里大多数指标，一把钳表、一支温度计、一个小时就够了。
Compressed air specific power @ 7 bar压缩空气比功率 @ 7 bar
24 kW/100cfm
19
< 17
Measure kW at the compressor panel and flow at the receiver outlet over one steady hour. Include dryer and cooling fans — the honest number is always worse than the sales brochure.
在空压机柜测功率，在储气罐出口测流量，取稳定运行一小时。要把干燥机和冷却风机算进去——诚实的数字总比样本差。
Compressed air leak rate压缩空气泄漏率
> 25%
10–15%
< 7%
Stop production, keep the compressor running, time load and unload cycles for 30 minutes. Load time ÷ total time = leak rate.
停产、保持空压机运行，记录 30 分钟内的加载与卸载时间。加载时间 ÷ 总时间 ＝ 泄漏率。
Boiler efficiency & stack O₂ (gas)锅炉效率与烟气含氧（燃气）
75% · O₂ >6%
82% · 3–4%
> 88% · 2–3%
A flue gas analyser gives O₂, CO and stack temperature in ten minutes. Efficiency above 88% means you are condensing — check the economizer and the flue material first.
一台烟气分析仪十分钟就能给出含氧、CO 和排烟温度。效率超过 88% 意味着已进入冷凝区——先确认省煤器和烟道材质。
Steam trap failure rate疏水阀失效率
> 15%
5–8%
< 3%
Ultrasonic probe plus surface temperature on every trap, once a year, recorded against a tag number. The register matters more than the survey.
每年一次，逐台用超声波探头加表面测温，按位号记录。台账比这次普查本身更重要。
Condensate return rate冷凝水回收率
< 40%
~70%
> 90%
Compare boiler make-up water volume to steam produced. High make-up means condensate, and its heat and chemicals, are going down a drain somewhere.
用锅炉补水量对比产汽量。补水量大，说明冷凝水连同它的热量和药剂，正从某个地沟流走。
Chiller efficiency (full load)冷水机组效率（满负荷）
0.75 kW/ton
0.55
< 0.45
kW from the panel, tons from flow × ΔT on the chilled water side. Then repeat it at 50% load — that is where the machine actually spends its life.
功率读电柜，冷量由冷冻水侧流量 × 温差算得。再在 50% 负荷下测一次——机组一生大部分时间都待在那里。
Lighting power density照明功率密度
> 12 W/m²
7–9
< 5
Count fittings × wattage ÷ floor area, and check lux at the working plane. Over-lighting an aisle is as expensive as using the wrong lamp.
灯具数 × 功率 ÷ 面积，再测作业面照度。通道照过头，和用错灯具一样贵。
Base load (night / non-production)基础负荷（夜间／非生产）
25–30% of peak峰值
15%
< 10%
Plot 15-minute interval data across a public holiday. Every flat plateau above zero is a machine, a fan or a compressor nobody switched off.
把法定假日的 15 分钟负荷曲线画出来。每一段高于零的平台，都是一台没人关的设备、风机或空压机。
Management system maturity管理体系成熟度
ad-hoc projects零散项目
ISO 50001
50001 + M&V
Outstanding means SEC is reported monthly against a regression baseline, every measure is verified after installation, and the plant manager owns the number.
卓越意味着：单位产品能耗按回归基线按月汇报、每项措施实施后都做验证、并且这个数字由厂长本人负责。
Energy saved is margin, not cost avoidance.
省下来的能源是利润，不是"少花的钱"。
Sales teams have to win new orders to make profit. An energy saving lands on the bottom line whole — no discount, no freight, no bad debt. This is the sentence that gets an EEP funded.
销售要拿下新订单才能赚到利润；而一笔节能是整份落到利润上的——没有折扣、没有运费、没有坏账。这句话，就是 EEP 拿到预算的理由。
The sales-equivalent test"等值销售额"测试
Pick your net margin, and see how much new revenue your sales team would have to win to match one unit of energy saved.选择你的净利率，看看销售团队要拿下多少新营收，才能抵得上省下的这一块钱能源。
So a 200,000 annual energy saving at a 5% net margin is worth the same to the P&L as 4,000,000 of new orders — won without a single new customer, price negotiation or delivery risk.因此，在 5% 净利率下，每年省下 20 万的能源，对利润表的贡献等同于 400 万的新订单——而且不需要新客户、不需要谈价、没有交付风险。
Sustainability可持续Does it actually make the business sustainable?它真的能让生意可持续吗？
Does it actually make the business sustainable?它真的能让生意可持续吗？
Every kWh not used is Scope 2 that never happened, and every m³ of gas not burned is Scope 1 removed at the source. No offsets, no accounting argument — and the abatement cost is usually negative: it pays you.
每一度没用掉的电，都是从未发生的范围二；每一立方没烧掉的气，都是在源头消掉的范围一。不需要抵消、不需要口径之争——而且减排成本通常是负的：它反过来给你钱。
Large buyers now cascade their own targets down the supply chain through CDP questionnaires, supplier scorecards and audits. An ISO 50001 system with real data is the difference between staying on the list and quietly dropping off it.
大买家正通过 CDP 问卷、供应商记分卡和现场审核，把自己的目标逐级传导到供应链。一套有真实数据的 ISO 50001 体系，决定你是留在名单上，还是悄悄被划掉。
A plant that needs 30% less energy per tonne is 30% less exposed to the next price spike, the next carbon levy, the next border adjustment. Efficiency is the cheapest hedge you can buy — and it never expires.
单位产品能耗低 30% 的工厂，对下一次价格暴涨、下一轮碳费、下一个边境调节机制的敞口也低 30%。能效是你能买到的最便宜的对冲——而且永不到期。
And the side effects are the ones operations actually cares about: fewer leaks, cleaner condensers, tuned burners and right-sized motors mean less downtime, less noise, steadier quality and a cooler building in August.
而它的"副作用"，恰恰是生产最在乎的：漏点少了、冷凝器干净了、燃烧器调好了、电机不再选大——停机更少、噪音更低、质量更稳，八月的车间也更凉快。
Where these figures come from.
这些数字从哪里来。
All percentages on this page are typical published ranges from the organisations below. They are a hypothesis for your plant, never a guarantee — your own meter is the only dataset that counts.
本页所有百分比均为下列机构公开发布的典型区间。它们是对你工厂的假设，绝非承诺——你自己的电表才是唯一算数的数据集。