Pharmaceutical Wastewater: Six Standards Govern One Industry, and "High-Concentration, Refractory" Is Only Half the Story

All are called "pharmaceutical wastewater," yet amino acid fermentation wastewater has a B/C of 0.52–0.64, making it more biodegradable than most printing and dyeing or petrochemical wastewater; while β-lactam antibiotic wastewater has only 0.20–0.25, not even reaching the aerobic threshold. The six national standard families have table row counts ranging from 9 to 25 rows, and the same "ammonia nitrogen" is 35 mg/L for the fermentation category and 25 mg/L for the chemical synthesis category, yet reverses when it comes to Total Nitrogen. Trying to make one article or one process package treat all pharmaceutical wastewater is itself a flawed premise.

Gaowutong · Industrial Water Treatment Technology Series · For industry technical personnel · All data are cited from original public standards, peer-reviewed literature, and public engineering sources

I. First distinguish "which category it is": the six standards' table row counts range from 9 to 25 rows

The six standards divide responsibilities by product process, not by scale, not by plant location, and not by discharge destination. The fermentation category governs the process of "producing antibiotics or other active ingredients through microbial fermentation, followed by separation, purification, and refining"; the chemical synthesis category governs the process of "producing pharmaceutical active ingredients through one or a series of chemical reactions." It is very common for the same plant site to have both fermentation workshops and synthesis workshops, in which case both standards are simultaneously effective.

The table below juxtaposes the "new enterprise limits" of the six standards (i.e., Table 2 in each standard, implemented for new enterprises from 2008 年 8 月 1 日, and for existing enterprises from 2010 年 7 月 1 日). The first thing to look at is not comparing values but comparing row counts: the mixed preparation category has only 9 items, while the chemical synthesis category has 25 items—more than double the difference.

Read these 6 rows as 6 sentences:

First, the same parameter has different values in the two standards, and the direction is even opposite. Ammonia nitrogen is 35 mg/L for fermentation-based and 25 mg/L for chemical synthesis-based; Total Nitrogen is 70 mg/L for fermentation-based and 35 mg/L for chemical synthesis-based. Fermentation-based is more lenient on ammonia nitrogen by 10 mg/L, and more lenient on Total Nitrogen by a factor of two. For a single plant site where the two types of workshops discharge in a mixed manner, the strictest value must be taken item by item: ammonia nitrogen according to the chemical synthesis-based 25, Total Nitrogen also according to the chemical synthesis-based 35, while COD is 120 for both. This is not an inference—Article 4.2.3 of the fermentation-based standard explicitly states: "Where the production facilities of an enterprise simultaneously produce more than two products, different discharge control requirements or national water pollutant discharge standards for different industries may apply, and the wastewater generated by the production facilities is treated and discharged in a mixed manner, the strictest concentration limits specified in the discharge standards shall be implemented, and the discharge concentration based on the benchmark wastewater volume of water pollutants shall be converted according to the formula."

Second, the water from the formulation line is the cleanest, yet the limits are the strictest. The COD limit for the mixed formulation category is 80 mg/L, the lowest among the six categories (extraction-based and traditional Chinese medicine-based 100, fermentation-based and chemical synthesis-based 120). Biological engineering-based is also 80. Intuition would suggest that "mixed formulation is just packaging and tableting, the water is very clean, so it should be lenient"—but the standard gives the opposite answer. Looking at the brackets makes it clear: the bracketed limits in Table 2 for the fermentation-based category (COD 100, BOD₅ 30, ammonia nitrogen 25, Total Nitrogen 50, TOC 30) apply to "integrated production enterprises that simultaneously produce fermentation-based active pharmaceutical ingredients and mixed formulations"; the same applies to the chemical synthesis-based category. In other words, as long as an enterprise has a single formulation line, the limits for the active pharmaceutical ingredient tier must be tightened by one more tier. The logic is not hard to understand: the formulation line discharges relatively clean water, which would dilute the high-concentration water from the active pharmaceutical ingredient line, and dilution cannot be counted as treatment.

Third, "three tiers" are easier to misremember than "two tiers." All six standards have Table 1 (transition period for existing enterprises 2009-01-01～2010-06-30), Table 2 (new enterprises + existing enterprises after 2010-07-01), and Table 3 (special emission limits, applicable to designated areas such as the Taihu Lake Basin). Table 1 is a buffer left for old plants put into operation before 2008 年, and has been invalidated after 2010 年 7 月 1 日. If anyone today still cites "COD 200" when talking about fermentation-based pharmaceutical manufacturing, they are referring to this already invalidated Table 1.

II. Before Applying the Label of "High Concentration and Refractory," Calculate It Yourself First

"Pharmaceutical wastewater = high concentration, refractory, and toxic" is almost an industry consensus. This statement fully holds for antibiotics, but it does not hold when applied to the entire pharmaceutical industry. The verification method is very simple: take the two appendix tables of HJ 2044—2014 "Technical Specifications for Wastewater Treatment Engineering of Fermentation-based Pharmaceutical Industry," divide BOD₅ by COD, and calculate the biodegradability yourself.

The conventional threshold for Aerobic biological treatment is B/C ≥ 0.3. Among the 7 rows, the three rows of β-lactams (0.20～0.25), other antibiotics (0.25～0.27), and vitamins (0.25～0.26) fall below the threshold; the three rows of aminoglycosides (0.39～0.60), macrolides (0.43～0.50), and amino acids (0.52～0.64) are above the threshold. The biodegradability of amino acid fermentation wastewater is better than that of most Printing and Dyeing Wastewater, petrochemical, and papermaking wastewater.

Therefore, the statement "pharmaceutical wastewater is refractory" only holds true for antibiotics (especially β-lactams) and vitamins. Applying the label "pharmaceutical wastewater" to amino acid fermentation broth is a mislabeling—and a wrong label directly leads to a wrong process route: water that should go directly to Anaerobic + A/O is given an unnecessary Advanced Oxidation stage, and the extra cost is measured in yuan per ton of water.

There is something even more intriguing here: the upper limit of COD (28000) and the upper limit of BOD₅ (7000) for β-lactams both appear in the highest tier across all 7 rows, yet their B/C falls into the lowest tier 0.20. "High concentration" and "refractory" are two independent things—high concentration only determines tank volume and energy consumption, while B/C determines whether Hydrolysis Acidification is needed and whether Advanced Oxidation should be placed upstream. Conflating the two during process selection leads to the outcome of "a very large tank built, yet B/C still at 0.2."

Incidentally, one more point relevant to later sections: the sulfate column for β-lactams is 500–8000 mg/L, a span of 16 times, the widest row in the entire table; while "other antibiotics" is only 150–300 mg/L. The boundary of the Anaerobic stage for fermentation-based pharmaceutical wastewater should be determined by sulfate rather than by COD, for reasons explained in Section 5.

III. Baseline Discharge Volume per Unit Product: A Conversion Formula That Only Penalizes, Never Rewards

The most unique design of the pharmaceutical standard family is that "discharge volume" is also written into the compliance determination. Six standards (excluding chemical synthesis) all provide a "baseline discharge volume per unit product," in units of m³ water/t product; and the conversion formula is the same:

ρbase = Qtotal ÷ (Σ Yi × Qibase) × ρactual

Where Qtotal is the measured total discharge volume, Yi is the output of the i-th product, Qibase is the baseline discharge volume per unit product for that product, and ρactual is the measured concentration. The key lies in the final additional rule: if the ratio of Qtotal to Σ(Yi·Qibase) is less than 1, the measured concentration shall be used as the basis for determination.

This sentence reads plainly but carries heavy engineering implications: if actual discharge volume is below the baseline, no "conversion reduction" is granted, and the measured value is used directly for determination; if actual discharge volume exceeds the baseline, the measured concentration is multiplied by a factor before determination. This is a one-sided penalty—saving water earns no credit, while wasting water results in multiplied penalties. Here is an example that can be calculated in one pass: a plant produces 1 t of penicillin per day, baseline 1000 m³/t, actual discharge 3000 m³/d, measured COD at the total outlet 150 mg/L. Ratio 3000÷1000 = 3, then ρbase = 3 × 150 = 450 mg/L, exceeding the 120 mg/L of GB 21903 Table 2 by 2.75 times. Even if it reduces discharge to 800 m³/d (ratio 0.8 < 1, determined directly by measured value), 150 mg/L still exceeds the standard—this formula does not reward water conservation; it merely prevents "excessive discharge" from becoming a dilution tactic.

The table below combines the current baseline discharge volumes with the "engineering design reference discharge volume ranges" from the annexes of HJ 2044–2014. The two sets of figures are not the same thing: one is the upper limit for compliance determination, the other is the actual engineering level. Viewed together, one exception can be identified.

Of the 8 rows, 7 rows fall within the engineering range, with only the determination upper limit for penicillin (1000) exceeding the upper bound of the engineering practice range (800). This indicates that when this limit was originally set, leeway was reserved for water volume management of penicillin-type products. However, in Table 2 of the revised draft publicly solicited for comments by the Ministry of Ecology and Environment on 2026 年 5 月, β-lactams were precisely split into "Penicillin G 200" and "Penicillin V Potassium 1000", with the baseline water discharge for Penicillin G dropping directly from 1000 to 200—less than half of the existing engineering practice range of 400～800; Vitamin C was also changed from 300 to 110. Public reports describe this revision as "modifying the baseline water discharge per unit product for penicillin and Vitamin C." Once these two figures take effect, the water balance of existing penicillin production lines will need to be redone entirely.

IV. Antibiotics: Potency Is Not a Discharge Indicator, EC₅₀ Is

In the pharmaceutical wastewater engineering documents retrieved, almost every one contains two statements: "influent residual penicillin potency 200–500 IU/mL" and "antibiotic activity in effluent completely eliminated (detection limit 0.01 IU/mL)." These two statements are not wrong in themselves, but there is a fact that is easily overlooked: among the 12 items in GB 21903, there is no item for "antibiotic potency," and among the 25 items in GB 21904, there is none either. Potency is a process control indicator for fermentation and extraction processes, not a discharge indicator. Using "potency not detected" as the basis for compliance is equivalent to measuring compliance with a ruler that does not exist in the standard.

What truly determines whether a biological system can run is the inhibitory concentration of antibiotics on microorganisms. Halling-Sørensen, in Archives of Environmental Contamination and Toxicology (2001, 40: 451–460), tested 11 antibacterial agents using two standard methods, ISO 15522 (activated sludge growth inhibition) and ISO 9509 (nitrification inhibition), and the results are summarized in the table below. The most memorable thing about this table is the order of magnitude — EC₅₀ falls within 0.1–100 mg/L, not at the g/L level.

Apply this yardstick to the engineering influent: by the international standard definition of penicillin sodium, 1 IU is approximately equivalent to 0.6 μg for conversion; a potency of 200–500 IU/mL converts to a mass concentration of approximately 120–300 mg/L. Compared with the table above, this value is two to three orders of magnitude higher than the EC₅₀ of chlortetracycline and streptomycin (0.4–0.5 mg/L), and has also exceeded the EC₅₀ of penicillin G itself against activated sludge heterotrophic bacteria, 84.6 mg/L. This is the arithmetic basis for the statement that "fermentation wastewater must be detoxified before entering biological treatment", not "COD being too high".

The second thing you must remember: Nitrifying bacteria are more sensitive than heterotrophic bacteria. The article points out that most chemicals are 5–10 times more effective against nitrifying bacteria, with streptomycin showing the most extreme contrast (heterotrophic bacteria 0.47 mg/L vs nitrifying bacteria 0.02 mg/L). The ammonia nitrogen in fermentation-based pharmaceutical wastewater is 100–360 mg/L (HJ 2044, Annex 2), and denitrification itself is the primary task of the entire process. This means inhibition often appears first in the denitrification stage, manifesting on site as "COD is still dropping, but ammonia nitrogen rebounds first" — when you encounter this phenomenon, first check the residual potency and solvents in the influent, and do not first suspect the aeration rate.

The third thing: inhibition not only comes quickly, but also leaves traces. A study on streptomycin sulfate (Environmental Technology, 2022) gave IC₅₀ = 7.5 mg/L for ammonia-oxidizing bacteria (AOB) and IC₅₀ = 6.6 mg/L for nitrite-oxidizing bacteria (NOB); at a dosage of 10 mg/L, the specific ammonia oxidation rate (SAOR) dropped to 3.43±0.52 mg N/(g MLSS·h), and the specific nitrite production rate (SNPR) dropped to 0.24±0.03 mg N/(g MLSS·h); after stopping the dosage, they only recovered to 9.37±0.36 and 1.66±0.49, without returning to the initial levels. During the same period, the relative abundance of the aadA1 resistance gene in the sludge increased by 0.99 orders of magnitude on day 140 天, and the relative abundance of Nitrobacter dropped from 1.30% to 0.25%. In one sentence: antibiotic shock is not "it will pass," but rather "it won't come back even after stopping the dosage, and it leaves behind resistance genes."

V. Sulfate: The Second Pollutant in Fermentation-Based Pharmaceutical Wastewater

In HJ 2044, Annex 2, the sulfate content of β-lactams is 500–8000 mg/L, which differs by an order of magnitude from the 150–300 mg/L of other antibiotics. In the anaerobic stage, sulfate competes with methanogens for electron donors, and there are two sets of quantitative relationships that can be directly applied at this step.

The first set is stoichiometry. Reducing 96 g SO₄²⁻ to sulfide requires 8 electrons, equivalent to 2 m mol O₂ or 64 g COD, so 1 g SO₄²⁻ corresponds to 0.667 g COD. The general criterion in the literature is the COD/SO₄²⁻ ratio: when greater than 0.67, sulfate is limiting and the proportion of sulfate reduction decreases; when greater than 10, sulfate reduction can essentially be ignored; when less than 1, methanogens are outcompeted and eliminated by sulfate-reducing bacteria.

The second set is measured coefficients. For every 1 kg SO₄²⁻ reduced, methane production decreases by approximately 0.23 m³ (standard conditions). Multiplying the two sets together:

0.667 kg COD/kg SO₄²⁻ × 0.35 m³ CH₄/kg COD = 0.233 m³ CH₄/kg SO₄²⁻

This almost coincides with the measured value of 0.23. Two mutually independent sources corroborate the same coefficient, so this coefficient can be confidently used for accounting.

Use this coefficient to check a publicly available case: a large antibiotic pharmaceutical enterprise (annual output of penicillin-type products 2000 t) reported wastewater sulfate of 3000～5000 mg/L and COD of 8000～12000 mg/L. Taking sulfate 4000 mg/L = 4 kg/m³, the COD captured by sulfate reduction is 4000 × 0.667 = 2668 mg/L, accounting for about 27% of the influent COD (taking the median 10000 mg/L); this translates to a methane loss of 4 × 0.233 = 0.93 m³ CH₄/m³ of water. The case also reports "annual reduction of organic pollutant emissions of 4200 t"; back-calculating from the difference 9915 mg/L between influent 10000 mg/L and effluent 85 mg/L, the treatment scale is about 1160 m³/d, so the methane loss caused by sulfate is about 1081 m³ CH₄/d, whereas its reported total biogas production is 55 万 m³/a (about 1507 m³/d, or about 904 m³ CH₄/d at a methane content of 60%). The theoretical methane loss from sulfate alone already exceeds its reported total biogas production—this is not to say its data are falsified: its process flow happens to include a "sulfate-reducing bacteria pretreatment" step, proving that it too knows sulfate must be reduced first. The real value of this accounting is: the design boundary of the anaerobic stage for fermentation-based pharmaceutical wastewater is sulfate, not COD.

There is one more threshold to remember: when hydrogen sulfide concentration increases to the order of 400 mg/L, anaerobic gas production will gradually decline until it stops (the classic report by Lawrence and McCarty); it is free H₂S that exerts the toxic effect, and the lower the pH, the higher the proportion of free H₂S. Therefore, for high-sulfate fermentation wastewater, pH and gas production are a pair of twin knobs—drop the pH, and gas production will collapse first, followed only then by a decline in COD removal rate.

VI. Three Rulers That Do Not Measure Substances: Acute Toxicity, "Not Detectable," and TOC

In the pharmaceutical standard family, there are several indicators whose object of measurement is not the concentration of some substance, but something else. They are most easily treated as a mere formality in acceptance and self-inspection, yet they are precisely the ones most likely to reveal problems.

① Acute Toxicity (HgCl₂ toxicity equivalent) 0.07 mg/L: The Only Indicator Not Tightened

The acute toxicity in Table 2 of the six standards is all 0.07 mg/L, and Table 3 is also all 0.07. The method standard is GB/T 15441—1995 luminescent bacteria method. Its unit is written as mg/L, but what it measures is "toxicity"—it does not say what substance is being measured, only "how toxic" it is, using the reference toxicant HgCl₂ to convert toxicity into a mass concentration. Look at the three-tier comparison table below: COD is tightened from 200 to 50 (4 times), TOC from 60 to 15 (4 times), and ammonia nitrogen from 50 to 5 (10 times), yet acute toxicity remains unchanged across all three tiers; meanwhile, total cyanide is tightened from 0.5 to "not detectable." The loosest item and the strictest item in the table stand side by side—this is a particularly noteworthy design in the pharmaceutical standard family.

② Behind Every "Not Detectable" There Is a Parenthesis

In the last row of the three-tier comparison table above, "not detectable" is followed by "detection limit 0.25 mg/L", meaning that a measured value below 0.25 mg/L is regarded as not detectable. The alkylmercury entries in Table 1 and Table 2 for the chemical synthesis category also state "not detectable", and the detection limit in the footnote is 10 ng/L, i.e., 0.00001 mg/L. Within the same set of standards, the actual meaning of "not detectable" differs by a factor of 25000 across different items. When you see the words "not detectable", you must look for that parenthetical value afterward; otherwise you may either mistakenly judge that you are in compliance or mistakenly judge that you are exceeding the limit. The second type of information was addressed in the 2026 年 revision draft: the alkylmercury discharge limit for the chemical synthesis category was changed directly from "not detectable" to "10 ng/L, i.e., 0.00001 mg/L" — turning a slogan into a number.

③ TOC and COD: One Counts Carbon, One Counts Electrons

All six standards list Total Organic Carbon and COD side by side. Theoretically, complete oxidation of every 1 mol of carbon consumes 1 mol O₂, so the upper-limit ratio of COD/TOC is 2.67 (32÷12, corresponding to substances such as carbohydrates where carbon is at an intermediate oxidation state). Table 2 for the fermentation category is COD 120 / TOC 40, ratio 3.0; the parenthetical version is 100 / 30, ratio 3.3 — both higher than 2.67, indicating that the standards themselves treat them as two separate matters.

Which one triggers the red flag first depends on the oxidation state of the organics in the wastewater: when sugars and alcohols dominate, COD/TOC approaches 2.67, and TOC will hit the limit first; when organic acids dominate (carbon at a higher oxidation state), the ratio is lower, and COD is relatively more lenient; when strongly reducing components are present, the ratio can exceed 2.67, and COD instead hits the limit first. Pollutant discharge permits are item-by-item permits — for example, the permit of Gansu Qilian Mountain Pharmaceutical (published by the Jiuquan Municipal Ecology and Environment Bureau on 2022 年) lists COD 120 mg/L and Total Organic Carbon 40 mg/L side by side in the same permit. Both items must be tested; one cannot be used to replace the other.

VII. Ongoing Revision: Indirect Discharge Limits Will Appear for the First Time

On 2026 年 5 月 20 日, the Ministry of Ecology and Environment publicly solicited opinions on 7 national water pollutant discharge standards via Huanban Biaozheng Han [2026] No. 15 (deadline 2026 年 6 月 20 日), of which 6 are the complete set of revision drafts for the pharmaceutical industry standard family: fermentation category, chemical synthesis category, extraction category, traditional Chinese medicine category, bioengineering category, and mixed preparation category. This is the first overall revision of this standard family since its issuance in 2008 年.

Regarding the background of the revision, the public statement is clearer than any interpretation and is worth reading sentence by sentence: "At present, more than about 60% of pharmaceutical enterprises have moved into industrial parks and discharge wastewater through indirect discharge, but the previously issued series of discharge standards only specified direct discharge limits, did not specify specific limits for indirect discharge, and only proposed that they could be negotiated and agreed upon. During the issuance of pollutant discharge permits, situations arose such as all pollutants of indirect discharge enterprises being permitted according to direct discharge limits, and some pollutants being permitted according to the Grade III standard of the Integrated Wastewater Discharge Standard, resulting in inconsistency. The wastewater treatment costs of enterprises are relatively high, and the treatment efficiency of downstream centralized wastewater treatment facilities has also not been fully brought into play."

Comparing this passage with Articles 2026-10-01 of this series (printing and dyeing wastewater), one will find a very interesting symmetry: the trouble with printing and dyeing is that "indirect discharge has three numbers (COD 80 / 500 / 200), and you have to determine which type of facility you discharge to"; the trouble with pharmaceuticals is that "there is not even this column yet, so it can only be negotiated." The same environmental management system is stuck at different positions in the two industries.

There are five items in this revision directly related to engineering:

First, all six standards add indirect discharge limits for water pollutants. Park enterprises will in the future have a national standard to rely on, rather than relying on agreements; but "having something to rely on" also means "being subject to inspection."

Second, a new "total salt content" item is added (fermentation category), and the chemical synthesis category adds two items, "total salt content" and "perfluorooctanoic acid." Including total salt content in the standard means that desalination, salt separation, and concentrate disposal change from "process selection" to "compliance items" — this can be read in comparison with the salinity framework in Articles 2026-09-19 (salt-tolerant bacteria for high-salinity industrial wastewater).

Third, the comprehensive toxicity control item changes its yardstick: from "acute toxicity (HgCl₂ toxicity equivalent)" to "zebrafish embryo acute toxicity (dilution factor)." It changes from "toxicity equivalent converted using a reference toxicant" to "dilution factor read directly from a biological test." The practical impact on enterprises is that the old and new methods give two different dimensions for the same water sample, and historical data cannot be directly continued.

Fourth, alkylmercury is changed from "not detectable" to 10 ng/L (i.e., 0.00001 mg/L), turning a slogan into a number.

Fifth, the benchmark discharge volume per unit product for penicillin and vitamin C is revised (see Section III).

In addition, the compilation notes provide an order-of-magnitude anchor: 2023 年 the annual emissions of Chemical Oxygen Demand and ammonia nitrogen in the pharmaceutical manufacturing industry decreased by 91% and 94% respectively compared with 2007 年. 2007 年 was the year before the implementation of this set of standards. In other words, this set of standards, which "only has direct discharge limits," reduced the industry's emissions to less than one-tenth of the original after more than ten years of implementation.

VIII. Engineering Ledger and Five Truths

Publicly verifiable engineering data are consolidated in one table. It should be noted that three of the rows come from equipment suppliers' and environmental protection companies' self-reports, have not undergone peer review, and can only be used as directional references and verified by bench tests; another two rows come from government public disclosures and are more credible.

Five engineering truths:

1. Toxicity is the first threshold of the anaerobic stage, not concentration. Three commercial documents give consistent experience: conventional "anaerobic + aerobic" applied directly results in sludge inactivation—at a certain vaccine plant, the biochemical tank sludge activity dropped from 30% to 5%, and COD removal rate was less than 40%. In all three cases, the process chain includes a detoxification stage of "micro-electrolysis / chemical reduction / catalytic oxidation", all positioned before the anaerobic stage.

2. Ammonia nitrogen and antibiotics are natural adversaries, so in engineering, carbon and nitrogen are separated into stages. Fermentation-type pharmaceutical wastewater with ammonia nitrogen 100–1200 mg/L is the norm, and antibiotics happen to inhibit nitrifying bacteria (EC₅₀ 0.02–4 mg/L). The approach at Pucheng Zhengda is that the first-stage A/O focuses on carbon and partial nitrogen, while the second-stage anoxic tank uses sulfur + sulfur autotrophic bacteria for carbon-source-free denitrification—this is another application of the same thing as the sulfur autotrophic denitrification in articles 2026-08-19.

3. Carbon source can be "taken from the raw water". Pucheng Zhengda added untreated raw water to the first-stage anoxic tank to supplement carbon, replacing 1000 kg of glucose per day. Calculate the replacement efficiency: based on the COD equivalent of glucose 1.067 kg COD/kg, 1000 kg/d glucose is approximately 1067 kg COD/d; while raw water COD 15000 mg/L = 15 kg COD/m³, only approximately 71 m³/d of raw water bypass (accounting for 3.6% of 2000 m³/d) is needed to make up for it. The key to "not buying glucose" is often not how advanced the process is, but that there happens to be a stream of high-concentration raw water on hand.

4. Salinity and detoxification must be physically placed upstream, and the order cannot be reversed. In a certain chemical synthetic API case with salinity 7% and B/C of only 0.15, the process places multi-effect evaporation desalination before anaerobic (desalination rate >90%, salinity reduced to <1%), and only then dares to apply UASB. Both salt and toxicity will directly kill granular sludge, and the biological stage can only receive water that has already been detoxified and desalinated.

5. Solid waste is separated by "category". Fermentation-type mycelial residue is HW02 (271-001-02, pharmaceutical waste), waste filter cloth is HW49 (900-041-49), while wastewater station sludge is managed as general industrial solid waste under the pollutant discharge permit. In the same fermentation workshop, four streams of solid waste enter four disposal pathways; if separated incorrectly, hazardous waste is mixed into general solid waste.

References

1. GB 21903—2008 "Discharge Standard of Water Pollutants for Pharmaceutical Industry Fermentation Products Category", Ministry of Environmental Protection, General Administration of Quality Supervision, Inspection and Quarantine, implemented on 2008-08-01 (Table 1, Table 2, Table 3, Table 4 and the baseline water volume conversion in Article 4.2).
2. GB 21904—2008 "Discharge Standard of Water Pollutants for Pharmaceutical Industry Chemical Synthesis Products Category", implemented on 2008-08-01 (Table 1, Table 2, 25 items in total, including alkylmercury detection limit of 10 ng/L).
3. GB 21905—2008 "Discharge Standard of Water Pollutants for Pharmaceutical Industry Extraction Products Category", implemented on 2008-08-01 (Table 1, 11 items in total, baseline discharge volume per unit product 500 m³/t).
4. GB 21906—2008 "Discharge Standard of Water Pollutants for Pharmaceutical Industry Traditional Chinese Medicine Products Category", implemented on 2008-08-01 (Table 1, 14 items in total, baseline discharge volume per unit product 300 m³/t).
5. GB 21907—2008 "Discharge Standard of Water Pollutants for Pharmaceutical Industry Bio-pharmaceutical Products Category", implemented on 2008-08-01; Environmental Impact Report of Zhengzhou Autobio Bioengineering Project (Table 2 cites COD 80, ammonia nitrogen 10, SS 50 mg/L).
6. GB 21908—2008 "Discharge Standard of Water Pollutants for Pharmaceutical Industry Mixed Preparations Products Category", implemented on 2008-08-01 (Table 2, 9 items in total, no characteristic pollutant items, baseline discharge volume per unit product 300 m³/t).
7. HJ 2044—2014 "Technical Specification for Wastewater Treatment Engineering of Pharmaceutical Industry Fermentation Products Category", Ministry of Environmental Protection (Appendix Table 1 discharge volume range per unit product, Appendix Table 2 overview of antibiotic production process wastewater quality, Appendix Table 3 overview of vitamin and amino acid production wastewater quality).
8. "Notice on Public Solicitation of Opinions on 7 National Water Pollutant Discharge Standards Including the 'Discharge Standard of Water Pollutants for Petrochemical Industry (Second Draft for Comments)'", Huanban Biaozheng Han [2026] No. 15, General Office of the Ministry of Ecology and Environment, 2026-05-20.
9. "Discharge Standard of Water Pollutants for Pharmaceutical Industry Fermentation Products Category (Draft for Comments)" and its compilation notes; "Discharge Standard of Water Pollutants for Pharmaceutical Industry Chemical Synthesis Products Category (Draft for Comments)", Ministry of Ecology and Environment 2026-05 (newly added total salt content, acute toxicity changed to zebrafish embryo acute toxicity, alkylmercury changed to 10 ng/L).
10. "Ministry of Ecology and Environment Solicits Public Opinions on 7 National Water Pollutant Discharge Standards", China Environment News, 2026 (including "about 60% or more pharmaceutical enterprises discharge wastewater through indirect discharge" and "2023 年 pharmaceutical manufacturing COD and ammonia nitrogen decreased by 91%, 94% compared with 2007 年").
11. Halling-Sørensen, B. Inhibition of Aerobic Growth and Nitrification of Bacteria in Sewage Sludge by Antibacterial Agents. Archives of Environmental Contamination and Toxicology, 2001, 40(4): 451–460 (EC₅₀ and NOEC of 11 antibacterial agents measured by two methods, ISO 15522 and ISO 9509).
12. Yang, B. et al. Effect of streptomycin sulphate on the nitrification system in activated sludge: insight into nitrification characteristic, antibiotic resistance gene and microbial community. Environmental Technology, 2022 (AOB IC₅₀ 7.5 mg/L, NOB IC₅₀ 6.6 mg/L, changes in SAOR/SNPR and aadA1 abundance).
13. Colleran, E. et al. Report on sulfate reduction causing methane production reduction of about 0.23 m³ (standard state)/kg SO₄²⁻, cited from Reviews in Environmental Science and Bio/Technology, 2015 (Methanogens, sulphate and heavy metals: a complex system).
14. Oude Elferink, S. J. W. H. et al. and Visser, A. et al. Reports on COD/SO₄²⁻ ratio thresholds (0.66–0.67, >10, <1), also see the review in the previous item.
15. Lawrence, A. W. & McCarty, P. L. Report on the decrease of anaerobic gas production until cessation when hydrogen sulfide concentration reaches the order of 400 mg/L, cited from "The Harm of Sulfate in Anaerobic Treatment of Pulp and Paper Wastewater", Wastewater Treatment Engineering Network, 2021-01-04.
16. "Environmental Protection New Quality Productive Forces: Denitrification Technical Renovation Project of Fermentation Pharmaceutical Wastewater of Pucheng Zhengda Biochemical Co., Ltd.", Environmental Protection Online (hbzhan.com), project scale 2000 m³/d, in operation since 2021 年 11 月.
17. "Public Notice on the Proposed Issuance of Pollutant Discharge Permit to Gansu Qilianshan Pharmaceutical Co., Ltd.", Jiuquan Municipal Bureau of Ecology and Environment, 2022-12-07 (GB 21903—2008 Table 2 permitted concentrations and annual discharge amounts, mycelium residue HW02 271-001-02, waste filter cloth HW49 900-041-49).
18. "Public Notice on the Proposed Issuance of Pollutant Discharge Permits to 6 Enterprises Including Gansu Xinhaipeng Chemical Technology Co., Ltd.", Jiuquan Municipal Bureau of Ecology and Environment, 2024 (permitted limits and annual discharge amounts for chemical synthesis pharmaceutical wastewater).
19. "Ozone Catalytic Oxidation-BAF Combined Process for Antibiotic Pharmaceutical Wastewater Treatment", Wastewater Treatment Engineering Network (dowater.com), 2021-08-26 (ozone stage COD removal 43%, B/C 0.12→0.28; BAF effluent COD 46, NH₄⁺-N 4.1 mg/L).
20. "Fermentation and Bio-pharmaceutical Wastewater: High Concentration, High Suspended Solids, Nutrient-rich", Toutiao (non-peer-reviewed, engineering self-report), 2025 (IC loading and biogas/standard coal saving data of the 1000 m³/d case; the biogas item has been independently calculated and verified in this paper).
21. "3 Real Cases: Treatment Solutions and Results for Different Types of Pharmaceutical Wastewater", Toutiao (non-peer-reviewed, engineering self-report), 2025 (chemical synthesis category 300 m³/d with salt content 7% case, traditional Chinese medicine extraction category 200 m³/d case).
22. Treatment engineering data of a large antibiotic pharmaceutical enterprise (annual production of penicillin category 2000 t) and a pharmaceutical enterprise in North China (2000 m³/d), Weilin Environmental Protection public case (non-peer-reviewed, engineering self-report), 2025.
23. GB/T 15441—1995 "Water Quality—Determination of the Acute Toxicity—Luminescent Bacteria Test", as the supporting analytical method for the acute toxicity (HgCl₂ toxicity equivalent) item.
24. WHO penicillin sodium international standard conversion relationship (1 IU approximately equals 0.6 μg), used for the conversion between potency (IU/mL) and mass concentration (mg/L) in Section 4 of this paper.

Gaowutong · Industrial Water Treatment Technology Series | All data in this paper come from the above public standard original texts, peer-reviewed literature and public engineering sources; equipment manufacturers' and environmental companies' self-reported data have been marked item by item as "non-peer-reviewed" in the main text and tables, and are for directional reference only. Please verify by bench-scale or pilot-scale tests before engineering application. All values independently calculated in this paper have been marked as "self-calculated" with the formulas given.