Decarbonising Biomanufacturing: Which Levers, in What Order

Decarbonising biomanufacturing is usually presented as a list: buy renewable power, cut HVAC, raise titer, go single-use, electrify heat. Every item on that decarbonisation list is real. What the list hides is that the items interact, so a decarbonisation roadmap that ranks projects by their standalone savings and adds them up will get both the total and the priority order wrong.

This article runs a practical decarbonisation programme for a biologics site through one model, one lever at a time and then in combination, so the interactions become visible as numbers. The case is the cluster's standard 2,000 L single-use CHO fed-batch, run in the free bioprocess LCA calculator, with a stainless-steel twin and a 10,000 L E. coli process for the questions where they answer differently. You will come away with a decarbonisation sequence, the arithmetic that justifies it, and the two places where a well-meant project increases emissions.

Start with where the carbon actually is

In a biologics suite, the carbon is almost entirely electricity, and the electricity is almost entirely air handling. Any plan for decarbonising biomanufacturing that does not start from that split will spend its first decarbonisation budget on the wrong things.

The modelled batch makes 3.024 kg of purified antibody and emits 12,036 kg CO₂e cradle-to-gate at a 400 g CO₂e/kWh grid, which is 3,980 kg CO₂e per kg. The breakdown is lopsided enough to state as a rule of thumb.

Two consequences follow before any decarbonisation lever is modelled. First, projects that touch only media, water or waste can never deliver more than about 4% on this batch, however well they are executed. Second, anything that changes the carbon content of a kilowatt-hour or the number of kilowatt-hours the air handlers draw is operating on 95.6% of the footprint. That is where a decarbonisation plan starts.

The split is specific to a long, cleanroom-bound mammalian process. A 1.5-day E. coli batch has a very different shape: electricity 71.2%, media 16.3%, steam 7.7%, because the cleanroom is occupied for a tenth as long. The reference figures for six processes give the breakdown for each, and it is worth checking which one your site resembles before reading any lever as general.

The four decarbonisation levers, one at a time

Modelled individually, electricity supply is worth 83.6%, titer 40.0%, HVAC setback 22.3% and water-generation energy 1.8%. Each figure is the change from the 3,980 kg CO₂e per kg base case with only that one input moved.

A few notes on what each decarbonisation row does and does not represent.

• Electricity supply multiplies every kilowatt-hour in the batch, which is why it dominates. It does not touch the 532 kg CO₂e per batch that sits outside the electricity line (resin, incineration, water, media), and it raises a question of accounting that the attributional vs consequential comparison works through in detail: a contract that does not cause new low-carbon capacity changes a market-based number without changing the atmosphere.
• Titer changes the denominator. Every input per batch is unchanged, so the per-batch total is untouched and the per-kg figure falls in exact inverse proportion: 3/5 = 0.600, a 40.0% cut. That property matters enough to get its own section below.
• HVAC setback is modelled as an average power figure because that is how the calculator takes it. The operational reality (reduced air changes when the suite is at rest, validated recovery times) is covered in the HVAC section.
• Water-generation energy covers the move from multi-effect distillation towards membrane-based water for injection, which the European Pharmacopoeia has permitted since 2017. On this batch it is a rounding error, as the water footprint analysis found: water generation is 3.1% of the electricity.

Why decarbonisation order matters

Pull all four levers together and the footprint falls 92.0%, to 318.9 kg CO₂e per kg. Add up their standalone savings and you get 147.7%. The difference is not an error in either calculation. It is the arithmetic of decarbonisation levers that multiply.

A cleaner grid lowers the carbon value of every kilowatt-hour, so any decarbonisation project that saves kilowatt-hours is worth less after the grid changes. A higher titer divides everything by a larger product mass, so any project measured per kilogram is worth less after titer rises. Each lever erodes the others. The practical consequence is that a lever does not have a single value. It has a value that depends on what has already been done.

The HVAC and water-generation decarbonisation projects both lose about 93% of their value when moved from first to last (889 to 67, and 71 to 5), because they save kilowatt-hours, and the grid lever has already made kilowatt-hours cheap in carbon. Titer loses 87% (1,592 to 213). Even the grid lever loses 55%, because by the time it arrives the other three have already removed a quarter of the kilowatt-hours and spread the rest over 67% more product.

The order-averaged figures are the fairest single number to put against each decarbonisation project in a business case:

• Electricity supply: 2,355 kg CO₂e per kg, 64.3% of the combined saving
• Titer 3 → 5 g/L: 846, 23.1%
• HVAC setback: 426, 11.6%
• Water generation: 34, 0.9%

The ranking is identical whichever way it is measured, which is reassuring: electricity supply first, then titer, then HVAC, then water. What changes is the size of each decarbonisation claim. If three teams each report their decarbonisation project's standalone saving to the same sustainability board, the board will believe it has cut the site's footprint by nearly one and a half times its size.

Worked example: decarbonising one batch

Applied in the order HVAC, water, titer, power, the four levers take the batch from 12,036 to 1,607 kg CO₂e and from 3,980 to 319 kg CO₂e per kg. Every step below reproduces in the calculator.

Base (2,000 L, 80% working, 3 g/L, 70% DSP yield, 90% success, 14 d, 400 g/kWh)
  Product      = 1,600 L x 3 g/L x 0.70 x 0.90      = 3.024 kg
  Electricity  = 28,759 kWh x 0.400                = 11,503.7 kg CO2e
  Non-elec.    = 203.5 + 242.7 + 51.4 + 34.6       =    532.2 kg CO2e
  Batch        = 12,035.9 kg CO2e  -> 3,980.1 kg CO2e/kg

Step 1  HVAC 80 -> 60 kW over 336 h
  HVAC energy  26,880 -> 20,160 kWh  (-6,720 kWh x 0.400 = -2,688.0)
  Batch        =  9,347.9 kg CO2e    -> 3,091.3 kg CO2e/kg   (saves 888.9)

Step 2  Water generation 15 -> 6 kWh/m3 on 60 m3
  WFI energy      900 -> 360 kWh     (-540 kWh x 0.400 = -216.0)
  Batch        =  9,131.9 kg CO2e    -> 3,019.8 kg CO2e/kg   (saves 71.4)

Step 3  Titer 3 -> 5 g/L
  Product      = 1,600 x 5 / 1,000 x 0.70 x 0.90   = 5.040 kg
  Batch        =  9,131.9 kg CO2e (unchanged)
                                     -> 1,811.9 kg CO2e/kg   (saves 1,207.9)

Step 4  Supply 400 -> 50 g CO2e/kWh
  Electricity  = 21,499 kWh x 0.050                =  1,075.0 kg CO2e
  Batch        =  1,075.0 + 532.2  = 1,607.2 kg CO2e
                                     ->   318.9 kg CO2e/kg   (saves 1,493.0)

Result: -86.65% per batch, -91.99% per kg

Two things stand out. The titer step saves more in this sequence (1,208) than it would going last (213) and less than going first (1,592), exactly as the ordering argument predicts. And the non-electrical 532.2 kg never moves: none of the four levers touches it. At the start it was 4.4% of the batch. At the end it is 33.1%. The next phase of decarbonising the site has to be about that residue, which is why the order of the final sections matters.

How much does cleanroom HVAC setback save?

Every 10% cut in cleanroom HVAC energy removes about 8.9% of the modelled CHO batch's footprint at 400 g CO₂e/kWh, and a setback schedule that halves suite load for 12 hours a day is worth 22.3%. On a 50 g/kWh supply the same project is worth 17.1%, and in absolute terms it saves 111 kg CO₂e per kg instead of 889, eight times less.

HVAC setback is the practice of reducing air change rates, and therefore fan, cooling and reheat load, when a classified area is at rest, and restoring them with a validated recovery time before operations resume. The case for it rests on a simple mismatch: a fed-batch suite runs 24 hours a day because the bioreactor does, but it is occupied, opened and disturbed for a fraction of that time, and occupancy is what generates the particles the air changes exist to remove.

The field evidence is now substantial. Loomans and colleagues studied three ventilation strategies for pharmaceutical cleanrooms and found that demand-controlled filtration based on occupancy could be implemented with negligible effect on cleanliness requirements, with modelled savings of up to 93.6% in their case-study facilities. At commercial scale, Roche/Genentech reported cutting ISO 8 areas from 25 to between 12 and 16 air changes an hour and ISO 7 high-risk areas from 48 to 28, which removed 165,000 cfm of airflow and cut site energy use and greenhouse gas emissions by 14%.

Three practical points decide whether the carbon arrives.

1. Setback is a validation project before it is an engineering one. The at-rest state has to be defined, recovery to in-operation limits demonstrated, and the change controlled. ISO 14644-16 on cleanroom energy efficiency gives the framework, and the air-change reduction case above was explicitly a joint manufacturing, quality, validation and engineering effort.
2. Fan energy falls faster than airflow. By the fan affinity laws, shaft power scales with the cube of flow, so halving airflow through a variable-speed fan cuts its power to about an eighth. The 25% HVAC energy cut modelled here is therefore conservative for fan-dominated systems and optimistic for systems dominated by reheat or dehumidification. Meter the air handler before claiming a figure.
3. Its carbon value is tied to the grid, and so to the timing. HVAC setback delivered while the site is on grid power is worth 8.9% per 10% of HVAC energy. Delivered after a low-carbon supply contract, it is worth a small fraction of that in carbon, though it keeps its full value in energy cost and in the kilowatt-hours the contract has to cover.

That last point is the strongest argument for putting HVAC work early in any decarbonisation sequence. It is also why the life cycle costing analysis of the same batch found HVAC to be a negative-cost abatement lever: it saves money whatever the grid, so its cost per tonne only looks worse once someone else has already claimed the carbon. The free cleanroom classification calculator checks particle counts against ISO 14644-1 limits, which is the evidence that has to hold before and after air changes are reduced.

Intensity targets versus absolute targets

Titer is the second-largest decarbonisation lever per kilogram and exactly zero per batch. Whether it contributes to decarbonising a site depends entirely on whether the site then runs fewer batches.

That distinction maps onto the two kinds of decarbonisation target a site will be held to. An intensity target (kg CO₂e per kg of drug substance, per dose, per unit revenue) rewards titer in full. An absolute target (tonnes per year, which is what a science-based target and most corporate net-zero pledges are built on) rewards it only through a reduction in operating hours.

A site pursuing decarbonisation that uses a titer gain to make more product will report a 40% intensity improvement and an unchanged absolute footprint, and both statements are true. Neither is misleading on its own. The error is to count the titer saving towards an absolute decarbonisation target, which is a common way for a site roadmap to look finished on paper while its reported tonnes stay flat.

The same logic applies to every lever that works only through the denominator: downstream yield, batch success rate and, for perfusion processes, the number of volumes harvested. They are among the best things a process team can do, and the LCA worked example shows why titer dominates intensity. They belong in the intensity column of the plan, with the absolute effect stated separately and conditioned on the demand assumption.

When to electrify steam

Replacing a gas steam boiler with an electrode boiler lowers emissions only on a grid below 224.1 g CO₂e/kWh. On the 400 g/kWh base grid it raises them, so electrification belongs after the power supply in the sequence, not before it.

The break-even is a single line of arithmetic. The calculator's steam model uses 2.2 MJ per kg of steam, an 80% boiler and the natural gas combustion factor of 0.0503 kg CO₂e per MJ, which gives 226.4 g CO₂e per kWh of steam heat. An electric route delivering that heat at efficiency or coefficient of performance η matches it at a grid of 226.4 × η.

On the stainless CHO twin, steam is only 1.6% of the batch, so the effect is small either way: an electrode boiler adds 1.3% at 400 g/kWh and removes 8.3% at 50. In the E. coli process steam is 7.7% of the batch, nearly five times the share, and there the decision is visible.

The lesson for decarbonising biomanufacturing generalises beyond boilers. Any project that swaps a fuel for electricity (steam, hot water, space heating, autoclaves) transfers emissions from Scope 1 to Scope 2, and whether that transfer is a reduction depends on the electricity. A site that electrifies first and contracts clean power second will report an emissions increase in the intervening years, which is a reporting problem even when the end state is right.

What is left once the power is decarbonised

Once the grid is clean, single-use plastic becomes the largest non-electrical term, and the single-use against stainless decision starts to matter for the first time. After the full stack, resin production and incineration are 27.8% of the batch, against 3.7% in the base case.

This is the point where decarbonisation and earlier findings in the cluster come together. At the 400 g/kWh base grid, the single-use vs stainless environmental comparison found the two within 0.24% of each other, with the verdict crossing at 479.2 g CO₂e/kWh. That near-tie was produced by electricity dominating both. Clean the electricity and the tie breaks:

So the later phase of decarbonisation has three candidates, in rough order of carbon value on this batch:

• Polymer end of life. Incineration without energy recovery is 243 kg CO₂e per batch, the largest single non-electrical item. Energy recovery reduces it by roughly a third and cannot eliminate it, since the fossil carbon in the polymer is released either way. Take-back and recycling schemes for single-use assemblies are still immature, which is why the tool rates its polymer data as the weakest in the set.
• Resin choice and assembly mass. 204 kg CO₂e per batch from resin production. Lighter assemblies and fewer redundant components reduce it directly.
• Process format. For a new facility decided on a clean-power basis, the stainless option's lower residue is a real point in its favour, to be weighed against flexibility, capital and changeover. For an operating single-use site, it is not a retrofit lever.

Media and water remain small throughout (2.2% and 3.2% of the stacked batch). They matter more for microbial processes, where media alone is 16.3% of the E. coli base case, and there feedstock choice moves up the list.

A sequenced playbook for decarbonising biomanufacturing

Measure energy intensity first, cut kilowatt-hours while they still carry carbon, then clean the supply, electrify heat only after that, and finish on plastics and media. Titer runs in parallel throughout and is reported against intensity, not absolute, targets.

Text version of Figure 3: Step 1, measure kilowatt-hours per kilogram and the grid factor. Step 2, cut load with HVAC setback, worth 22.3% if done first. Step 3, contract additional low-carbon power, worth 83.6% alone and 64% of the combined saving. Step 4, electrify steam and heat, only once the grid is below 224 g CO2e/kWh times the heating COP. Step 5, address the residue of plastics, media and end of life. A parallel track improves titer, yield and batch success, which cut intensity but not the per-batch total.

Each step of decarbonising biomanufacturing in more detail:

1. Measure the right two numbers before any decarbonisation spend. Energy intensity in kWh per kg and the grid factor behind the carbon figure. With those, every later saving can be restated on any grid, as the benchmark article shows. Without them, the site cannot tell a process improvement from a change in the grid mix.
2. Cut kilowatt-hours while they still carry carbon. HVAC setback first, because it is the largest load and its carbon value shrinks fastest once the supply changes. Water generation and cooling efficiency here too, knowing they are small for a mammalian suite and larger for a microbial one.
3. Contract low-carbon power, and check it is additional. This is the largest single decarbonisation lever by a wide margin. Bjørn and colleagues showed that unbundled renewable energy certificates can let companies report reductions that do not correspond to real ones, so favour supply that demonstrably causes new capacity. Cutting load in step 2 also shrinks the contract step 3 has to buy.
4. Electrify heat only once the supply clears the break-even. 224.1 g CO₂e/kWh for a resistance or electrode boiler, scaling with COP for a heat pump. Earlier than that, it moves emissions from Scope 1 to Scope 2 and increases them.
5. Work the residue, the last stage of decarbonisation. Single-use end of life and assembly mass for a mammalian single-use site, media and feedstock for a microbial one. These were irrelevant at the start and dominate at the end.

Titer, yield and batch success run alongside all five decarbonisation steps. They are the best value the process team can add and they should be reported against intensity targets, with any absolute claim conditioned on the site running fewer batches.

Four limits apply to everything above, and to any decarbonisation roadmap built from it. The numbers are screening estimates from one open model, not a verified assessment of any facility. The HVAC and supply inputs are stated assumptions chosen to be realistic, not measurements. The analysis is cradle-to-gate for drug substance, so fill-finish, cold chain and distribution are outside it. And the grid lever is modelled attributionally: a consequential assessment of the same supply contract can give a very different answer, which is worth settling before the contract is signed.

The short version: decarbonising biomanufacturing is a sequencing problem more than a shopping list. The same four decarbonisation levers deliver 92% whichever order they arrive in, but the credit each one earns, and whether two of them raise emissions on the way, depends entirely on the order. Plan the decarbonisation sequence first and the business cases for each project follow from it.

Frequently asked questions

What is the biggest lever for decarbonising a biomanufacturing site?

Electricity supply. On a modelled 2,000 L single-use CHO fed-batch, electricity is 95.6% of the cradle-to-gate footprint, so moving from a 400 to a 50 g CO₂e/kWh supply cuts it by 83.6% on its own, from 3,980 to 652 kg CO₂e per kg. Averaged over every possible order of the four levers modelled here, electricity supply delivers 64% of the combined saving.

How much carbon does cleanroom HVAC setback save?

Cleanroom HVAC is 93.5% of the modelled CHO batch's electricity and 89.3% of its total footprint at 400 g CO₂e/kWh, so every 10% cut in HVAC energy removes about 8.9% of the batch. Running the suite at half load for 12 hours a day, a 25% cut in HVAC energy, saves 22.3%. On a 50 g CO₂e/kWh supply the same project saves only 17.1%, and 8 times less carbon in absolute terms.

Does increasing titer reduce a site's emissions?

It reduces emissions per kilogram, not necessarily per year. Raising titer from 3 to 5 g/L cuts the modelled footprint per kg by exactly 40% and per batch by exactly 0%. A site that keeps running 16 batches a year emits the same 192.6 t CO₂e and makes 67% more product. Absolute emissions fall only if demand is fixed and the site runs fewer batches, here 9.6 a year for 115.5 t.

Should a biomanufacturing site electrify its steam boilers?

Only once its electricity is clean enough. Steam from a gas boiler at 80% efficiency carries 226.4 g CO₂e per kWh of heat. An electrode boiler therefore only lowers emissions on a grid below 224.1 g CO₂e/kWh, which only 1 of the 26 US eGRID subregions meets. A heat pump with a coefficient of performance of 2 moves the break-even to 452.7 g/kWh. Electrify after the power supply, not before.

Why don't carbon savings from different projects add up?

Because the levers multiply rather than add. A cleaner grid shrinks the value of every kilowatt-hour saved, and a higher titer shrinks the value of everything per kilogram. Modelled on one CHO batch, the four levers save 83.6%, 40.0%, 22.3% and 1.8% individually, which sums to 148%, while together they save 92.0%. A roadmap that adds the individual savings will promise more than it can deliver.

What is left after the electricity is decarbonised?

Mostly plastic. After clean power, HVAC setback, higher titer and lower water-generation energy, the modelled single-use CHO batch falls to 1,607 kg CO₂e, of which single-use resin and its incineration are 27.8%. At a zero-carbon grid a matched stainless-steel process is 32.4% lower than single-use, against 0.24% at 400 g CO₂e/kWh, so the single-use decision starts to matter only once power is clean.

References

1. Loomans M.G.L.C., Molenaar P.C.A., Kort H.S.M., Joosten P.H.J. (2019). Energy demand reduction in pharmaceutical cleanrooms through optimization of ventilation. Energy and Buildings 202:109346. doi:10.1016/j.enbuild.2019.109346
2. Bjørn A., Lloyd S.M., Brander M., Matthews H.D. (2022). Renewable energy certificates threaten the integrity of corporate science-based targets. Nature Climate Change 12:539–546. doi:10.1038/s41558-022-01379-5
3. Arpagaus C., Bless F., Uhlmann M., Schiffmann J., Bertsch S.S. (2018). High temperature heat pumps: Market overview, state of the art, research status, refrigerants, and application potentials. Energy 152:985–1010. doi:10.1016/j.energy.2018.03.166
4. Belkhir L., Elmeligi A. (2019). Carbon footprint of the global pharmaceutical industry and relative impact of its major players. Journal of Cleaner Production 214:185–194. doi:10.1016/j.jclepro.2018.11.204
5. 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
6. Amasawa E., Kuroda H., Okamura K., Badr S., Sugiyama H. (2021). Cost–Benefit Analysis of Monoclonal Antibody Cultivation Scenarios in Terms of Life Cycle Environmental Impact and Operating Cost. ACS Sustainable Chemistry & Engineering 9:14012–14021. doi:10.1021/acssuschemeng.1c01435