Why Bandung's Citarum basin makes 2026 process wastewater a strategic problem, not a compliance footnote
Bandung Regency is the most populous district in Greater Bandung, and its population is projected to keep rising toward Indonesia's 2045 demographic window, increasing domestic water demand on the Citarum/Cikapundung basin that every new fab or data hall must draw from (UKM 2024, citing BPS Bandung Regency 2020–2022). Globally, semiconductor water use doubled between 2012 and 2022 (TNFD, Feb 2026, citing Marcello 2024), and 40% of existing fabs plus over 40% of new fabs announced since 2021 are projected to sit in basins with high or extremely high water-stress risk by 2030 (TNFD 2026, citing Lepawsky 2024).
The TNFD February 2026 case study links that basin stress directly to transition risk—market, reputational and policy exposure stacked on top of physical supply risk—meaning a Bandung facility that discharges an under-designed train faces permit loss, community opposition and customer-ESG scrutiny. Local life-cycle evidence already shows that Bandung wastewater-treatment choice shifts emissions from 0.11 to 0.78 kg CO₂eq per m³ (IOP 2024), so engineering decisions at the Bandung site are materially consequential. The same logic that drives a manufacturing water-reduction engineering guide elsewhere applies here, but the basin multiplier is the differentiator: a Bandung 2026 train must be sized to Citarum carrying capacity and the next tightening of Permen LHK effluent quality, not to a global average.
Two process streams, not one: fab wastewater versus data-hall wastewater
Designing one combined train for a site that produces two chemically different streams is the most common Bandung retrofit error. The fab side generates UPW rinse reject (largest volume, low TDS, high recovery potential), fluoride-bearing spent etch and cleaning waste (HF, BOE, NH₄F residues), high-COD organics (photoresist, solvents, stripper), and CMP slurry wastewater carrying colloidal silica and trace metals. The data-hall side is chemically simpler but larger and more continuous: cooling-tower blowdown with high TDS, scaling ions and residual biocides, humidification bleed-off, and low-pressure RO reject from makeup pretreatment. The volume anchor is unambiguous: a typical data centre uses 25–770 million L/yr while hyperscale facilities exceed 2 billion L/yr, and a single fab uses around 14 billion L of UPW per year, with 1.4–1.6 units of municipal water consumed for each unit of UPW (TNFD, Feb 2026, citing IDE Technologies 2024). That ratio forces every Bandung designer to plan the water balance first and the chemistry second. The working rule: keep fluoride, organics and CMP streams segregated from cooling blowdown; only RO reject and humidification bleed should normally be considered for blending into a reuse loop.
| Parameter | Fab process stream | Data-hall process stream |
|---|---|---|
| Dominant volume | UPW rinse reject (largest), fluoride/etch, CMP slurry, organics | Cooling-tower blowdown, humidification bleed-off, low-pressure RO reject |
| Key contaminant | Fluoride (HF/BOE/NH₄F), COD, colloidal silica, trace metals | TDS, scaling ions (Ca, Mg, Si), biocides |
| Flow profile | Batch, shift-driven, peak during tool campaigns | Continuous, diurnal, tied to ambient wet-bulb |
| Default 2026 reuse path | UPW reject → RO/UF → process or boiler feed; fluoride → precipitation; organics → MBR ± AOP | Cooling-tower blowdown → softening + side-stream RO → cooling makeup; humidification bleed → reuse loop |
| Segregation rule | Never blend with cooling blowdown | May receive RO reject and humidification bleed only |
| Anchor volume (TNFD Feb 2026) | ~14 billion L UPW/yr per fab | 25–770 million L/yr typical; >2 billion L/yr hyperscale |
Designing the 2026 Bandung process train block by block
A defensible Bandung train is built source-segregated, then converges only at polishing. A rotary mechanical bar screen is the standard headworks choice because Bandung's combined-sewer inflows during rain events carry rags, plastics and fibrous debris that will blind fine screens and upset downstream chemistry. From headworks, the fab-side flow goes to source-segregated equalization with PLC-controlled chemical dosing for pH balancing ahead of fluoride precipitation; the variable shift pattern of a fab makes EQ mandatory. Fluoride and metals removal involves chemical precipitation plus a lamella clarifier to drop calcium fluoride and metal hydroxides, followed by multimedia filtration to protect downstream membranes from carry-over solids. Organics and COD removal utilizes an MBR membrane bioreactor as the 2026 Bandung default: it offers a small footprint, tolerance of variable fab organic loads, and effluent quality close to reuse without a tertiary step. Add an AOP stage (Fenton or ozone) only when the influent carries recalcitrant resist or stripper residues that the MBR cannot fully oxidise. Polishing and reuse come last: UF followed by a high-recovery industrial RO system recovers UPW rinse reject for process or boiler feed, while cooling-tower makeup uses UF plus a chemical dosing program sized to the Bandung supply-water profile. The pH and reagent side of the train is anchored on an automatic chemical dosing system tied to the same PLC/SCADA platform that reports ESG water metrics. On the discharge side, design to PP 22/2021 and the relevant Permen LHK effluent quality standards for the Citarum basin, but size for the next tightening—and where reuse is constrained, plan ZLD or high-recovery evaporation so a tightening basin allocation does not strand the asset.
| Train block | Function | 2026 Bandung rationale |
|---|---|---|
| Headworks — rotary bar screen | Remove rags, plastics, fibrous debris | Combined-sewer debris during rain events; protects downstream chemistry |
| Equalization + pH control (PLC) | Buffer batch flows; dose acid/base | Fab shift pattern makes EQ mandatory ahead of precipitation and biology |
| Fluoride/metal precipitation + lamella clarifier | Drop CaF₂ and metal hydroxides | Segregated from cooling blowdown; protect downstream membranes |
| MBR (membrane bioreactor) | COD/BOD removal to near-reuse quality | Small footprint, handles variable fab organics; default 2026 Bandung block |
| AOP (Fenton or ozone) — only if needed | Oxidise recalcitrant resist/stripper residues | Add when MBR effluent cannot meet reuse COD target |
| UF + RO polishing | Recover UPW reject; produce cooling makeup | Anchor on the 1.4–1.6× UPW ratio (TNFD Feb 2026) to justify high-recovery design |
| Discharge / ZLD fallback | Meet PP 22/2021 and Permen LHK; protect against basin tightening | Design for the next standard revision, not today's limit |
Stream-by-stream technology selection matrix for 2026 Bandung facilities
The matrix below is the working tool for a Bandung engineer auditing an existing train or specifying a new one. It links each stream to the technology that fits it, preventing over-specification. UPW rinse reject goes to multimedia filter plus RO with a high-recovery design envelope, then back to process or boiler feed—this is the largest reuse win on the site and the easiest to justify financially against raw water costs from the stressed Citarum system. Fluoride and etch waste goes to calcium-based precipitation, a lamella clarifier, and a plate-and-frame filter press for sludge dewatering; it must never go directly to a biological plant because fluoride shocks will kill the biomass. Photoresist, stripper and high-COD streams go to an MBR as baseline, with an AOP (Fenton or ozone-based) added for refractory loads; ozone requires extra air-permit work in Bandung's urban airshed, so check the local air permit envelope first. Cooling-tower blowdown is softened and passed through side-stream RO with a chemical dosing program sized to the Bandung makeup profile, then reused for cooling makeup, scrub blowdown or—where basin rules allow—irrigation. CMP slurry waste is segregated to a dedicated precipitation, lamella and UF train and never commingled with the MBR, because colloidal silica and trace metals will upset biology and foul the membranes. A UV steriliser closes the polishing train where reuse loops need a non-chemical disinfection barrier.
| Stream | Primary treatment | Solids handling / polishing | Reuse or discharge path |
|---|---|---|---|
| UPW rinse reject | Multimedia filter + RO (high-recovery) | — | Process rinse or boiler feed |
| Fluoride / etch waste | Ca-based precipitation + lamella clarifier | Plate-and-frame filter press for CaF₂ sludge | Discharge to Citarum under PP 22/2021 / Permen LHK |
| Photoresist / stripper / COD | MBR baseline; AOP (Fenton or ozone) for refractory loads | UV steriliser if reused in contact loop | Reuse or discharge; check Bandung air permit for ozone |
| Cooling-tower blowdown | Softening + side-stream RO + chemical dosing | — | Cooling makeup; irrigation where basin rules allow |
| CMP slurry waste | Dedicated precipitation + lamella + UF (segregated) | Sludge to plate-and-frame filter press | Discharge under fab effluent limits; never to MBR |
Compliance, reuse targets and the Bandung-specific 2026 shortlist
The design is only as good as the procurement and compliance envelope around it. Anchor the train to PP 22/2021 and the relevant Permen LHK effluent quality standards for the Citarum basin, plus any province-level tightening, and design for the next revision of those standards. Set a numeric reuse target so the train has a measurable goal tied to TNFD-style water-stress disclosure expectations in 2026: a working Bandung target is recovery of at least 70% of fab UPW reject and recycling of at least 50% of cooling-tower blowdown, with the actual number set against the basin's most recent allocation. Shortlist suppliers on four Bandung-specific criteria: documented UPW-reject and fluoride experience, a basin-discharge compliance track record, a service footprint that can reach a Java site without long travel, and the ability to deliver automated PLC/SCADA packages that integrate directly with the facility's ESG reporting. For small sites or retrofit-constrained projects, a packaged integrated water purification skid with an MBR membrane bioreactor block plus a chemical-programmed disinfection stage such as a chlorine dioxide generator lowers civil works and the GHG-per-m³ exposure that the IOP 2024 Bandung LCA flagged. The Bandung reader can apply the same logic to peer sites by comparing against the Mumbai fab and data-hall process wastewater guide or the Pune fab and data-hall process wastewater guide, but the basin numbers, the fluoride chemistry and the local discharge envelope remain Bandung-specific.
Frequently Asked Questions
What discharge and reuse targets should a Bandung fab or data hall set for 2026?
A defensible Bandung 2026
Frequently Asked Questions
What is the recommended process wastewater treatment train for a semiconductor fab in Bandung in 2026?
The recommended treatment train for a Bandung-based semiconductor facility consists of segregated collection followed by specialized pretreatment: acid/alkali neutralization, fluoride precipitation (using calcium chloride to reach < 10 mg/L), and specialized heavy metal removal for copper and nickel. This must be followed by a centralized biological treatment stage using Membrane Bioreactor (MBR) technology to handle high organic loading, and a final Reverse Osmosis (RO) stage to ensure compliance with stringent water recycling targets and local discharge limits.
How do data hall cooling-tower blowdown and humidification bleed differ from fab wastewater in Bandung, and can they share a treatment train?
Data hall cooling-tower blowdown is typically characterized by high cycles of concentration, resulting in elevated Total Dissolved Solids (TDS), hardness, and biocides, whereas fab wastewater contains complex fluorides, solvents, and heavy metals. Due to the risk of chemical cross-contamination and the disparity in treatment requirements, it is technically inadvisable to share a treatment train; data hall streams are best managed via side-stream filtration and softening, while fab streams require dedicated chemical-physical separation.
Which Indonesian and Citarum-basin discharge standards apply to a 2026 fab or data hall in Bandung?
Facilities must comply with the national Permen LHK No. 5 of 2014 regarding industrial wastewater quality standards. However, due to the critical status of the Citarum River, all Bandung-based facilities are subject to the significantly stricter regional requirements under Pergub Jawa Barat No. 39 of 2019, which mandates lower concentration limits for Chemical Oxygen Demand (COD), Ammonia, and Total Suspended Solids (TSS) compared to national averages.
What is the realistic CAPEX and footprint for an MBR plus RO package serving a mid-size Bandung fab in 2026?
For a mid-size fab processing approximately 1,000 to 2,000 m3/day, a containerized or modular MBR plus RO package requires an estimated footprint of 400 to 600 square meters, including equalization tanks and chemical dosing skids. The estimated 2026 CAPEX, accounting for regional logistics and specialized corrosion-resistant materials, ranges between $1.5 million and $2.8 million USD, excluding civil works and site-specific piping infrastructure.
How should a Bandung facility shortlist wastewater equipment suppliers for a 2026 fab or data-hall retrofit, and what lead time should be expected?
Shortlisting should prioritize vendors with proven local track records in West Java to ensure rapid access to spare parts and localized technical support for membrane maintenance. Suppliers must demonstrate compliance with Indonesian SNI standards and provide references for high-purity water recovery. Given current global supply chain volatility, expected lead times for specialized membrane skids and high-grade instrumentation range from 24 to 36 weeks from the date of purchase order issuance.
