Every cleaning cycle in a biomanufacturing plant is a mass flow that nobody weighs. It arrives as purified water, leaves as effluent, and appears on no bill of materials. Cleaning mass intensity is the metric that makes it visible: the kilograms of cleaning water and cleaning chemicals consumed per kilogram of product released.
This article works the metric properly for a real process, compares stainless against single-use on it, and then does the more useful thing, which is to ask whether the metric deserves the decisions people want to make with it. Every figure comes from the bioprocess LCA calculator running the same modelled 2000 L CHO batch used throughout this cluster, so the arithmetic can be reproduced rather than believed.
What cleaning mass intensity measures
Cleaning mass intensity (CMI) is the total mass of cleaning inputs divided by the mass of product, in kg/kg. It is a decomposition of process mass intensity that isolates the cleaning cycle, in the same way that a cost model isolates utilities from raw materials.
The parent metric has a clear lineage. Jimenez-Gonzalez and colleagues set out in 2011 why the pharmaceutical industry converged on process mass intensity rather than E-factor or atom economy, and Budzinski and colleagues adapted it to biologics in 2019, where water rather than solvent dominates. Benison and Payne then extended the same accounting explicitly into plant cleaning in 2022. The metric sits in that line of work.
One caveat belongs at the front rather than buried at the end. CMI is not an ISO-standardised indicator. There is no governing definition telling you whether pre-rinse water counts, whether to include the water used to clean transfer lines between suites, or whether a campaign of six batches divides by six. Those are your boundary choices, and a CMI quoted without them is not comparable to anyone else's. State the boundary in writing, exactly as any life cycle boundary has to be stated.
Calculate CMI for your own process
Enter vessel volume, titer, downstream yield and your CIP water volume. The calculator returns process mass intensity, its water component and the carbon each one carries.
How to calculate CMI
Sum every litre of cleaning water, add the dry mass of cleaning chemicals, and divide by the product the cleaning enabled. One litre of water is taken as one kilogram, which is accurate to about 0.2% at the ambient temperature CIP skids meter at. Hot water is up to 3% less dense, so record volumes at the meter rather than at the vessel.
The modelled batch is a 2000 L stainless CHO fed-batch at 80% working volume, 3 g/L titer, 70% downstream yield and 90% batch success, which releases 3.024 kg of product. Its clean-in-place cycle consumes 26,000 L.
A 26,000 litre clean-in-place cycle for a 2000 litre stainless bioreactor is broken into pre-rinse 4,000 litres, caustic wash 6,000 litres, intermediate rinse 4,000 litres, acid wash 4,000 litres and final rinse 8,000 litres. Divided by 3.024 kilograms of product this gives a cleaning mass intensity of 8,598 kilograms per kilogram, which is 30.93 percent of total process mass intensity. The same cycle carries 371.9 kg CO2e, made up of 193.7 kilograms from sterilisation steam, 156 kilograms from pharmaceutical water generation electricity and 22.3 kilograms from water supply and effluent treatment, which is only 3.1 percent of the batch carbon footprint.
Worked example: CMI of the stainless batch
Step 1. Product mass. 2000 L × 80% working = 1,600 L. At 3 g/L that is 4.8 kg upstream. Apply 70% downstream yield and 90% batch success: 4.8 × 0.70 × 0.90 = 3.024 kg.
Step 2. Cleaning input mass. 26,000 L of CIP water at 1 kg/L = 26,000 kg. Cleaning chemicals, assuming a 6,000 L caustic wash at 1.0% w/v and a 4,000 L acid wash at 0.5% w/v, add 60 kg of sodium hydroxide and 20 kg of phosphoric acid, so 80 kg.
Step 3. Divide. 26,000 ÷ 3.024 = 8,598 kg/kg water-only. Including chemicals, 26,080 ÷ 3.024 = 8,624 kg/kg.
Step 4. Sanity check the chemicals. The 80 kg of caustic and acid is 0.31% of the cleaning mass. The metric is, to within a third of a percent, a water number. Chemical-reduction projects do not move it; water-reduction projects do.
The chemical concentrations in step 2 are stated assumptions, not calculator outputs, because the tool does not model CIP chemistry. That does not matter much here precisely because of step 4: any plausible concentration leaves CMI a water metric.
Stainless vs single-use: 13× on mass, nothing on carbon
The metric separates the two facility types more sharply than any other indicator, and then the carbon footprint refuses to follow it. The calculator ships a matched pair of presets, identical in volume, titer, yield, batch length, cleanroom load and grid, differing only in cleaning regime and plastic inventory.
| Quantity | Single-use | Stainless | Ratio |
|---|---|---|---|
| CIP water per batch | 2,000 L | 26,000 L | 13.0× |
| Cleaning mass intensity | 661 kg/kg | 8,598 kg/kg | 13.0× |
| Process mass intensity | 19,884 kg/kg | 27,797 kg/kg | 1.40× |
| CMI as share of PMI | 3.33% | 30.93% | 9.3× |
| Cleaning carbon per batch | 13.7 kg CO2e | 371.9 kg CO2e | 27.1× |
| Cleaning share of batch carbon | 0.11% | 3.10% | 28.2× |
| Total footprint | 3,980 kg CO2e/kg | 3,971 kg CO2e/kg | 0.998× |
The 13.0× ratio is not an artefact of this particular batch. Both routes share the same denominator, so the ratio is simply 26,000 ÷ 2,000. It is invariant in titer, in downstream yield, in batch success and in vessel scale. Any facility comparison run on it will return that same number wherever you set the process parameters, which makes it a very stable metric and a very poor tiebreaker.
The reason the totals converge anyway is in the full single-use versus stainless comparison: the stainless cleaning premium of +358 kg CO2e per batch is very nearly cancelled by the single-use plastic premium of +387 kg, leaving a net 28 kg in favour of stainless. Cleaning mass intensity sees the first of those two premiums and is structurally blind to the second, because polymer film is a consumable rather than a cleaning input.
31% of the mass, 3% of the carbon, 0.03% of the cost
Score the stainless batch's cleaning cycle against three sustainability metrics and you get three incompatible answers about how important it is. This is the same structural finding the water footprint analysis reached, sharpened by isolating cleaning alone.
| Metric | Cleaning contribution | Share of batch total | Verdict it implies |
|---|---|---|---|
| Process mass intensity | 8,598 of 27,797 kg/kg | 30.93% | Cleaning is a headline issue |
| Carbon footprint | 371.9 of 12,007 kg CO2e | 3.10% | Cleaning is a minor line |
| Cost of goods | $150.80 of $468,000 | 0.032% | Cleaning is invisible |
The mass and cost views disagree by a factor of 966. That is not a defect in any of the three metrics; it is the reason ISO 14044 forbids publishing a single weighted score in comparative assertions disclosed to the public. Whoever picks the metric picks the answer, and cleaning is the clearest example of it in a bioprocess because water is simultaneously the heaviest, cheapest and least carbon-intensive thing in the plant.
There is a hard consequence hiding in the middle row. Cleaning contributes 3.10% of batch carbon, so 3.10% is the absolute ceiling on what any cleaning-side decarbonisation project can deliver, including abolishing cleaning entirely. A programme promising more than that is either counting something outside the cleaning boundary or is wrong.
Which levers actually move CMI
Six levers, scored on the metric and on carbon side by side. The pattern is that almost nothing moves both, and the two that do are not cleaning levers at all.
| Lever | CMI (kg/kg) | Δ CMI | Δ carbon |
|---|---|---|---|
| Halve CIP water to 13,000 L | 4,299 | −50.0% | −0.74% |
| Campaign 5 batches per full clean | 1,720 | −80.0% | −2.48% |
| Eliminate CIP water, keep SIP steam | 0 | −100% | −1.48% |
| Switch to single-use | 661 | −92.3% | −2.98% |
| Titer 3 → 5 g/L | 5,159 | −40.0% | −40.0% |
| Downstream yield 70 → 85% | 7,081 | −17.6% | −17.7% |
Three things in that table are worth pulling out.
Eliminating CIP water entirely saves less carbon than campaigning does. Zeroing the water removes 100% of CMI and 1.48% of carbon, while campaigning five batches removes 80% of CMI and 2.48% of carbon. The reason is that campaigning divides the sterilisation steam as well as the water, and steam is 52% of the cleaning carbon. A project targeting litres alone leaves the larger half of the cleaning footprint untouched.
The campaign lever has a closed-form ceiling. Running N batches between full cleans divides the per-batch cleaning inventory by N, so carbon saving tends to the cleaning share as N grows: 1.55% at N = 2, 2.48% at N = 5, 2.79% at N = 10, 2.94% at N = 20 and 3.10% in the limit. Ninety percent of the achievable benefit arrives by the tenth batch, which is well inside what changeover and cleaning validation constraints usually permit.
Titer and yield move both metrics by exactly the same percentage. That is not coincidence. Both appear only in the denominator of both metrics, so the metric is exactly inverse in titer, just as the carbon footprint is: 1 g/L gives 25,794 kg/kg, 3 g/L gives 8,598, 5 g/L gives 5,159 and 10 g/L gives 2,579. A cell line project is a cleaning-water project, and it is a better one than most cleaning-water projects. You can reproduce the whole column by changing titer in the LCA model behind these figures and watching mass and carbon fall together.
Check the cost side of the same levers
Cleaning water is 0.03% of batch cost. The economics calculator shows which of these levers survive once labour, facility time and media are priced in.
Why several small rinses beat one large rinse
Splitting a fixed rinse volume into more, smaller rinses removes dramatically more residue for the same water, because rinse dilution is geometric rather than linear. This is the one genuinely large, genuinely free lever on cleaning water, and it is routinely left on the table.
After draining, a vessel retains a holdup volume Vh as film and low points. Filling with rinse volume Vr, mixing to homogeneity and draining again multiplies the residue concentration by Vh / (Vh + Vr). After n rinses the residue is that factor raised to the nth power, so log removal is linear in the number of rinses and only logarithmic in the size of each.
Worked example: 2,000 L of rinse water, spent two ways
Take a 2000 L vessel retaining Vh = 10 L after draining, which is 0.5% of volume.
One 2,000 L rinse. Dilution factor = 10 ÷ (10 + 2,000) = 0.004975. Log removal = −log10(0.004975) = 2.30 log.
Ten 200 L rinses. Dilution factor per rinse = 10 ÷ (10 + 200) = 0.04762, giving 1.322 log each. Over ten rinses: 10 × 1.322 = 13.22 log.
Same 2,000 L of water. Nearly eleven extra logs of residue removal.
Read the other way round, to hit a 6-log target you need 5 × 200 L = 1,000 L in small rinses against 3 × 2,000 L = 6,000 L in full-volume rinses. That is a 6× water reduction at identical cleaning performance, worth 1,653 kg/kg on this batch.
| Rinse volume | % of vessel | log10 removed per rinse | Water per log (L) | Rinses for 6 log | Total water (L) |
|---|---|---|---|---|---|
| 100 L | 5% | 1.041 | 96 | 6 | 600 |
| 200 L | 10% | 1.322 | 151 | 5 | 1,000 |
| 400 L | 20% | 1.613 | 248 | 4 | 1,600 |
| 1,000 L | 50% | 2.004 | 499 | 3 | 3,000 |
| 2,000 L | 100% | 2.303 | 868 | 3 | 6,000 |
| 4,000 L | 200% | 2.603 | 1,537 | 3 | 12,000 |
That floor is worth stating precisely. As rinse volume shrinks, water per log tends to Vh × ln(10) = 2.303 Vh. The lower bound on rinse water is set by the holdup, not by the vessel. Halving retained holdup, through better drainability, slope, and elimination of dead legs, halves the floor for every rinse the vessel will ever receive. On this vessel, cutting holdup from 10 L to 5 L takes a 6-log rinse at 200 L from 5 rinses to 4.
Two practical limits stop this being free money all the way down. Spray devices need a minimum flow and volume to wet the full surface, so below roughly 5% of vessel volume you stop achieving coverage and the dilution model no longer describes what is happening. And each additional rinse costs a drain-and-fill cycle, so the saving is in water and effluent, not in time. Fan, Phinney and Heldman measured exactly this trade-off between CIP parameters and rinse effectiveness in a food-processing context, and the geometry carries over directly.
Where CMI collides with cleaning validation
Cleaning mass intensity and cleaning validation pull in opposite directions, and validation wins every time. That is the correct outcome, and it defines where the metric may legitimately be used.
A cleaning validation programme sets a maximum allowable carryover, converts it to a surface limit and then to a rinse-sample or swab limit, and requires demonstrated, reproducible achievement of it. Tightening that limit demands more log reduction, and more log reduction demands more rinse water. The metric therefore rises with the stringency of the acceptance criterion, which means a facility can improve its CMI by cleaning less thoroughly, and that is not an improvement.
This is why the rinse-granularity result in the previous section matters more than any of the volume levers. Splitting a rinse changes the water needed to reach a fixed log reduction; it does not change the log reduction. It is the only water lever that is unambiguously neutral with respect to the CIP and SIP validation position, because the acceptance criterion is met identically either way. Any lever that reduces cleaning water by reducing cleaning has to be requalified, and most will fail.
The useful framing for a site, then, is a constrained one: hold the validated acceptance limit fixed, and minimise water subject to it. Under that constraint the available moves are rinse granularity, holdup reduction through better-drained equipment and lines, final-rinse water recovery into the next pre-rinse, and campaigning. Everything else on a typical CIP water-reduction list is either a validation change or a rounding error.
What the metric does not tell you
Cleaning mass intensity is a useful diagnostic and a bad decision rule. Four specific failures are worth knowing before quoting one.
- It is blind to everything that is not cleaning. On the matched pair, CMI is 13.0× worse for stainless and the total footprint is 0.2% better. Any facility decision made on cleaning mass intensity alone would get the environmental answer backwards.
- It weights water at its mass, which is where water matters least. Water is the heaviest, cheapest and least carbon-intensive input in the plant. Mass is the one axis on which it can dominate, and the metric is built entirely on that axis.
- It has no boundary convention. Pre-rinse inclusion, transfer-line cleaning, campaign division and whether SIP condensate counts are all unlegislated. Two honest engineers can differ threefold on the same plant.
- It can be improved by cleaning worse. No metric that rewards a weaker acceptance limit should be used as a target without the validation position pinned alongside it.
Used properly, it answers one question well: within a fixed cleaning regime, where is the water going and what would a change be worth? That is a real question, and the rinse arithmetic above shows the answers can be large. It simply is not the question of whether the process is a good one, which needs the full carbon footprint and, honestly, the cost model next to it.
Set the acceptance limit first
The cleaning validation calculator works MACO through to swab and rinse limits, which is the constraint every CMI reduction has to respect.
Frequently asked questions
What is cleaning mass intensity (CMI)?
Cleaning mass intensity is the total mass of cleaning water and cleaning chemicals consumed per kilogram of product, expressed in kg/kg. It is a decomposition of process mass intensity that isolates the cleaning cycle rather than the process. On a modelled 2000 L stainless CHO fed-batch producing 3.024 kg, 26,000 L of CIP water gives a CMI of 8,598 kg/kg; the matched single-use batch uses 2,000 L and scores 661 kg/kg.
How do you calculate cleaning mass intensity?
Divide the total cleaning input mass by the product mass. Sum every litre of pre-rinse, wash, intermediate rinse and final rinse water, treating 1 litre of water as 1 kilogram, add the dry mass of caustic and acid used, then divide by the kilograms of purified product the campaign released. Cleaning chemicals are typically under 0.5% of the total, so CMI is in practice a water number.
Is cleaning mass intensity an ISO standard metric?
No. Process mass intensity is the established metric, promoted by the ACS Green Chemistry Institute Pharmaceutical Roundtable and adapted to biologics by Budzinski and colleagues in 2019. Cleaning mass intensity is a decomposition of PMI rather than a separately standardised indicator, though Benison and Payne extended the PMI concept explicitly into plant cleaning in 2022. Define your boundary in writing before quoting a number.
Does reducing cleaning water reduce carbon emissions?
Barely. On the modelled stainless batch, cleaning accounts for 30.93% of process mass intensity but only 3.1% of the carbon footprint, so eliminating cleaning water entirely would cut the carbon footprint 1.48% and eliminating the whole cleaning cycle including its steam would cut it 3.1%. That 3.1% is a hard ceiling on what any cleaning-side project can deliver.
Do several small rinses use less water than one large rinse?
Yes, substantially. Rinse dilution is geometric, so each rinse multiplies residue by Vh/(Vh+Vr) where Vh is the retained holdup. With 10 L of holdup, ten 200 L rinses achieve 13.2 log of residue removal while a single 2,000 L rinse achieves 2.3 log using exactly the same water. The water needed per log of removal falls toward a floor of 2.303 times the holdup volume as rinse size shrinks, though spray-device coverage sets a practical minimum.
Does campaigning batches reduce cleaning mass intensity?
Yes, and it is the single largest CMI lever available to a stainless facility. Running five batches between full cleans divides the per-batch cleaning inventory by five, cutting CMI 80% from 8,598 to 1,720 kg/kg. The carbon benefit is far smaller at 2.48%, and it is bounded above by 3.1% no matter how long the campaign runs.
References
- Jimenez-Gonzalez C., Ponder C.S., Broxterman Q.B., Manley J.B. (2011). Using the Right Green Yardstick: Why Process Mass Intensity Is Used in the Pharmaceutical Industry To Drive More Sustainable Processes. Organic Process Research & Development 15:912–917. doi:10.1021/op200097d
- Benison C.H., Payne P.R. (2022). Manufacturing mass intensity: 15 Years of Process Mass Intensity and development of the metric into plant cleaning and beyond. Current Research in Green and Sustainable Chemistry 5:100229. doi:10.1016/j.crgsc.2021.100229
- Budzinski K., Blewis M., Dahlin P., D'Aquila D., Esparza J., Gavin J., et al. (2019). Introduction of a process mass intensity metric for biologics. New Biotechnology 49:37–42. doi:10.1016/j.nbt.2018.07.005
- Fan M., Phinney D.M., Heldman D.R. (2018). The impact of clean-in-place parameters on rinse water effectiveness and efficiency. Journal of Food Engineering 222:276–283. doi:10.1016/j.jfoodeng.2017.11.029
- Pietrzykowski M., Flanagan W., Pizzi V., Brown A., Sinclair A., Monge M. (2013). An environmental life cycle assessment comparison of single-use and conventional process technology for the production of monoclonal antibodies. Journal of Cleaner Production 41:150–162. doi:10.1016/j.jclepro.2012.09.048