Solvent Recovery Engineering Guide
Vapor Phase VOC Recovery & Emission Control

Executive Summary

This engineering guide explains the solvent recovery process and how Solvent Recovery Units — vapor phase solvent recovery systems also known as solvent vapor recovery systems, industrial solvent recovery systems, solvent recovery solutions and VOC recovery solutions — work, when activated-carbon recovery outperforms thermal oxidizers and other VOC abatement/VOC recovery technologies, which solvents can (and cannot) be recovered, the efficiencies and stack emission levels achievable, and the CAPEX/OPEX, environmental-compliance and safety-engineering factors that drive technology selection. It covers the four-stage adsorption/desorption/condensation cycle at the core of every carbon-based SRU; the three regeneration technologies — nitrogen, steam and vacuum — and the solvent chemistry, utility availability and safety considerations that determine which one fits a given application; a solvent-by-solvent compatibility guide for 17 commonly recovered solvents plus halogenated solvents as a separate case; and a like-for-like comparison of activated carbon adsorption against RTO, DTO, CTO, direct condensation, liquid scrubbing, biofiltration and membrane separation, including where hybrid concentrator/SRU or concentrator/oxidizer configurations make sense. Engineering design sections address airflow and turn-down sizing, humidity, temperature, fouling, corrosion, and cooling-system selection; economics sections cover CAPEX, OPEX, ROI and payback logic; compliance sections map BAT/BREF, IED and EPA regulatory frameworks to system design; and safety-engineering sections cover ATEX/explosion prevention, oxygen and LEL monitoring, fire protection, redundancy and site-layout risk, drawing on independent property-risk-engineering guidance (AXA XL, FM Global) alongside regulatory sources. The guide closes with digitalization/Industry 4.0 practice, DEC's SMS™/CBS™ product range, 60+ FAQs, a 100+ term glossary, and a full bibliography. It is written for process, environmental and EHS engineers, plant managers and risk/insurance engineers evaluating, specifying or auditing a Solvent Recovery Unit.

Key Engineering Takeaways

• SRUs recover, not destroy: activated-carbon adsorption reclaims solvent for reuse, unlike RTO/DTO/CTO thermal oxidation, which converts VOC to CO₂ and water (see §4).
• Recovery efficiency: up to ~99% of solvent input, with TVOC stack values as low as ≤50 mgC/m³ (≤20 mgC/m³ with DEC.ULE™) achievable (see §1).
• Three regeneration technologies — nitrogen (steamless/dry), steam, and vacuum — each suit different solvent chemistries, site utilities and safety profiles; the choice drives CAPEX, OPEX and recovered-solvent purity (see §9).
• Water solubility and flammability, not boiling point alone, are the primary drivers of regeneration-technology selection (see §9.4).
• Not every solvent is safely recoverable: some solvents are flagged in risk-engineering literature as prone to breakdown under adsorption/desorption conditions; DEC manages risk via nitrogen regeneration rather than avoiding it (see §5).
• Two or more adsorber beds are required for continuous operation — one adsorbs while the other regenerates (see §1.2).
• Carbon bed fire protection uses nitrogen inertization preferred to overbed water deluge (see §13.4).
• Regeneration energy demand is the primary operating-cost driver: advanced energy-recovery processes such as DEC.THR_x™ and DEC.ULP™ can reduce energy demand for inert-gas regeneration by up to 40%, while ECOVAP™ can decrease thermal energy demand for steam-regenerated systems from the historical ~3–4 t steam/t solvent to approximately 2 t/t solvent or less (see §10).
• Non-regenerative (throwaway) adsorbers are typically only economical below roughly 10 tonnes/year of total VOC mass loading; above that, a regenerative SRU generally wins on lifecycle cost (see §4.3).

1. What is a Solvent Recovery Unit? Engineering Fundamentals

1.1 Definition

A Solvent Recovery Unit (SRU) — interchangeably called a Solvent Recovery Plant (SRP) or Solvent Recovery System (SRS) (see DEC's terminology explainer on how these terms relate to Solvent Reclaiming Equipment, SRE) — is an industrial air pollution control system that removes volatile organic compounds (VOCs) from a solvent-laden air (SLA) exhaust stream, concentrates them on a sorbent (typically activated carbon), and desorbs them in a form that can be condensed and reused as liquid solvent. Unlike destructive VOC abatement technologies that oxidize solvent into CO₂ and water, a Solvent Recovery Unit is a recovery, not a destruction, technology: the solvent is preserved as a saleable or reusable product.

1.2 The Solvent Recovery Process: How Activated Carbon Solvent Recovery Works — The Four-Stage Cycle

Every carbon-based SRU operates on a repeating four-stage cycle split across at least two adsorber vessels, so that one vessel can adsorb while the other regenerates:

• Adsorption: solvent-laden air passes through a fixed bed of activated carbon; VOC molecules are physically retained (adsorbed) onto the internal pore surface of the carbon by van der Waals forces, while clean air exits the bed and is discharged to the stack or recirculated to the process.
• Bed switching: once a bed approaches its designed breakthrough point (see §8.4), the process air is automatically switched to a second, regenerated bed so that adsorption is continuous while the first bed is regenerated.
• Desorption (regeneration): the loaded bed is stripped of solvent using nitrogen, steam, or hot air/vacuum (see §9), reversing the adsorption equilibrium and releasing a concentrated solvent vapor stream.
• Condensation & recovery: the desorption vapor is cooled in a condenser (direct or indirect, sometimes assisted by refrigeration, brine cooling, or — for particularly stringent residual-emission targets — cryogenic condensation) until the solvent condenses to liquid; if immiscible with water, it is separated by decantation — laminar-flow decanter design, rather than turbulent flow, measurably improves recovered-solvent clarity and reduces emulsion formation at the water/solvent interface — and if miscible, it may require downstream purification (dehydration and/or distillation) to reach the target purity for reuse.

1.3 Adsorption Principles

Adsorption is a surface phenomenon: gas-phase VOC molecules migrate to and are held on the internal surface of a porous solid (the adsorbent) without a chemical reaction taking place. The driving force is the equilibrium partitioning of the VOC between the gas phase and the adsorbed phase, described by adsorption isotherms (e.g., Freundlich or Langmuir-type relationships) that link adsorbed quantity to gas-phase concentration at constant temperature. Physical adsorption (physisorption) is reversible, which is what allows the same carbon bed to be loaded and regenerated thousands of times over its service life. Key variables governing adsorption performance include the carbon's pore size distribution relative to the VOC molecule size, the VOC's molecular weight and polarity, gas velocity through the bed, temperature, and relative humidity of the inlet air.

1.4 Desorption Principles

Desorption is the reverse process: energy is supplied (as heat - TSA, or combined heat and reduced pressure T+VSA) to shift the adsorption equilibrium so VOC molecules leave the carbon surface and return to the gas phase at a much higher concentration than in the original SLA stream. Because desorption reconcentrates the VOC by a large factor (commonly 10–50×, depending on regeneration method and solvent), it enables efficient downstream condensation with a comparatively small heat-exchange duty. Complete desorption is neither achievable nor necessary in practice: a carbon bed retains a "heel" of residual solvent (working capacity is discussed in §8.5), and regeneration is optimized around economic and safety criteria rather than 100% removal.

1.5 Condensation & Solvent Purification

The desorbed vapor stream is condensed to recover liquid solvent. Water-immiscible solvents (e.g., toluene, xylene, hexane) separate cleanly from any condensed process water by gravity in a decanter. Water-miscible solvents (e.g., ethanol, IPOH, acetone, THF) condense together with any process moisture picked up during regeneration, and typically require a downstream separation step — often a dehydration module and/or distillation — to reach a purity suitable for reuse in the originating process. Miscibility alone does not tell the whole story: many solvents form azeotropes — binary (solvent–water) or ternary (solvent–solvent–water, or solvent–solvent) mixtures that boil at a constant composition and cannot be separated by straightforward distillation — so purification design must check the full azeotrope behavior of the recovered mixture, not just whether the solvent is nominally miscible or immiscible with water. Where a problematic azeotrope exists, options include molecular-sieve-based dehydration, pressure-swing distillation (DST-p or DST-v), each selected based on the specific binary or ternary system involved, or entrainer-assisted (azeotropic/extractive) distillation (DST-e). A concrete, well-documented illustration of how quickly this becomes a real constraint rather than a theoretical one: ethyl acetate and water form a binary azeotrope boiling at roughly 70.4°C, at a fixed composition of about 91.5% ethyl acetate to 8.5% water — meaning a wet ethyl acetate stream cannot simply be distilled down to anhydrous solvent past that composition; the azeotrope itself has to be removed overhead and phase-split again to make further progress. Adding a second solvent narrows the achievable purity further still: an ethyl acetate/ethyl alcohol/water blend forms a ternary azeotrope boiling at roughly 70.2°C (approximately 82.6% ethyl acetate, 8.4% ethyl alcohol, 9.0% water), on top of the additional binary azeotropes each pair of these three components forms with each other — illustrating, concretely, why a recovered-solvent stream containing more than roughly single-digit percentages of a second, water-miscible co-solvent and water together can become genuinely difficult to purify by straightforward distillation, rather than merely "needing a slightly bigger column."

Recovered-solvent purity, not just recovery efficiency, is therefore a key design parameter: a plant with 99% VOC capture is only as useful as the reusability of what comes out the other end. As an illustration of what a well-run multi-solvent recovery/distillation train can achieve in practice, documented industrial operating data for solvents including MEK, toluene and cyclohexanone show recovered-solvent GC purity in the 98.5–99%+ range, residual water in the 0.05–0.10% range, and acidity (as acetic or adipic acid, from trace hydrolysis/oxidation byproducts — see §8.6) held below roughly 0.003–0.02%, at a mass-recovery rate close to 100% of net applied solvent — figures worth requesting from a vendor as a specification target and verification benchmark, not just a recovery-efficiency percentage. As a specific illustration of what a well-designed nitrogen-regenerated system can hold to in practice, documented DEC plant performance data for ethyl-acetate-based and mixed MEK/ethyl-acetate/ethanol streams shows recovered-solvent water content below 0.1% (typically averaging around 0.05%) and acetic acid content likewise below 0.1% (typically averaging around 0.05%) — consistent with, and reinforcing, the essentially-eliminated hydrolysis discussed in §8.6 for dry, nitrogen-regenerated service.

1.6 Closed-Loop Recovery Systems

In a closed-loop configuration, the regeneration inert gas (typically nitrogen) is continuously recirculated between the carbon bed, the condenser and back to the bed heater, with no gas vented to atmosphere during desorption. Closed-loop nitrogen systems allow near-total oxygen exclusion, which is important both for solvent stability (many recovered solvents are prone to oxidation) and for keeping the regeneration atmosphere outside the flammable range even at high VOC concentration. Closed-loop design is the basis of nitrogen (inert gas) regeneration systems and is discussed in detail in §9.1.

1.7 Dehydration Sequencing: First-Phase (Gas-Phase) vs. Second-Stage (Liquid-Phase) Dehydration

Where water is removed from the process matters as much as whether it is removed at all, for two independent reasons: energy consumption and byproduct contamination. It is useful to think of dehydration as happening in up to two distinct phases: a first-phase, gas-phase pre-dehydration step, acting on the regeneration vapor before or during condensation, and a second-stage, liquid-phase dehydration step, acting on the already-condensed liquid solvent.

First-phase (gas-phase) pre-dehydration. Rather than only dehydrating the liquid condensate after desorption and condensation are already complete, some SRU designs selectively remove water while the stream is still in the gas/vapor phase, before it reaches that later liquid-phase step. DEC implements this first phase through three related technologies — DEC.ADM_PHD™ (Pulse Heat Dehydration), DEC.ADM_FDP™ (Fractional Dehydration Process) and the gas-phase DEC.ADM_MSU™ (Molecular Sieve Unit), suited to more complex mixed-VOC streams that include ketones — all acting during carbon regeneration itself and driven by the same underlying Kinetic Desorption Algorithm (DEC.KDA™), further tuned by AI/ML, which selectively isolates and condenses water vapor separately from the target solvents. In both cases, the underlying principle is the same: removing water while still in the gas phase reduces the water load that the rest of the process (condensation, and any second-stage dehydration) has to handle, which can lower overall energy consumption and reduce or eliminate the need for a downstream second-stage dehydration step entirely. Pre-dehydration is different from condensing moisture as a preliminary step and then re-evaporating it elsewhere in the air-handling train, which involves two separate phase-change energy penalties (condensation, then re-evaporation) instead of one, and can introduce a mildly acidic, solvent-contaminated water stream (e.g., containing trace acetic acid from ester hydrolysis) into ductwork or filter housings usually not designed for corrosion resistance.

Second-stage (liquid-phase) dehydration and its placement relative to distillation. Where a first-phase step is not sufficient on its own — or is not present — a second-stage, liquid-phase dehydration step [e.g. a DEC.ADM_UDS™ (Ultra-Dehydration System) type module using molecular sieves] removes the remaining water from the already-condensed liquid solvent. When this second-stage dehydration is used ahead of a multi-column distillation train, sequencing still matters within the liquid phase: molecular sieves can act as a mild catalyst supporting the hydrolysis of the processed solvents. Placing the second-stage dehydration step before the full distillation train, rather than between distillation columns, keeps any acid generated out of the distillate light stream and avoids pure solvent contamination. Removing water ahead of distillation also means the columns are fed an already-dry mixture, which is more thermally efficient than distilling a wetter feed and dehydrating partway through the train.

2. Why Recover Solvents? The Environmental and Economic Case

2.1 Environmental Benefits

Solvent recovery directly reduces VOC emissions to atmosphere, which lowers ground-level ozone and photochemical smog formation, reduces exposure to hazardous air pollutants (HAPs) for surrounding communities, and supports compliance with increasingly strict emission limit values under frameworks such as the EU Industrial Emissions Directive (IED) and US EPA regulations (see §12). Because the process recovers material rather than destroying it, it also avoids the combustion-related CO₂, NO_x and fossil fuel (e.g. natural gas, CH₄) consumption associated with thermal oxidation, giving solvent recovery a favorable position in most carbon-footprint comparisons versus destructive VOC abatement.

2.2 Economic Benefits

Recovered solvent directly offsets the cost of virgin solvent purchases, which is often the single largest driver of payback for an SRU (see §11.4). In processes with high solvent consumption — printing, coating, flexible packaging, battery electrode manufacturing — solvent purchase cost can dwarf both the CAPEX and OPEX of the recovery system, making the investment case straightforward once yearly solvent throughput crosses a facility-specific threshold.

2.3 Circular Economy & Decarbonization

Solvent recovery is a textbook circular-economy technology: instead of a linear "purchase virgin solvent → emit / incinerate" model, the solvent is captured, purified, and returned to the same process, cutting both upstream extraction/production emissions embedded in virgin solvent and downstream emissions from destruction technologies. For companies with ESG or net-zero commitments, SRUs are frequently one of the highest-return decarbonization investments available at a process-emissions level, because the reduction in purchased solvent is measurable, auditable, and directly tied to a cost saving rather than only a compliance cost. A rigorous comparison against thermal destruction should account for full lifecycle (Scope 3) emissions, not only the facility's own direct combustion emissions: producing and transporting virgin replacement solvent for a destroyed VOC stream carries its own upstream CO2 footprint, which recovery avoids by keeping the same solvent molecules in use rather than requiring new solvent to be manufactured and shipped in. Published industry case studies applying this full lifecycle accounting to high-value solvents such as toluene have reported substantially lower total CO2e for recovery than for oxidation once virgin-solvent production and logistics are included, alongside sub-one-year payback periods driven by avoided solvent purchase — illustrating that the economic and environmental cases for recovery tend to move together rather than trading off against each other for commodity, high-value solvents (see References, §18).

2.4 Waste Reduction

Beyond air emissions, solvent recovery reduces the volume of spent-solvent waste that would otherwise require off-site disposal or incineration as hazardous waste, lowering waste-handling costs and the regulatory burden associated with hazardous-waste manifesting and transport.

3. Selection Criteria for a Solvent Recovery Unit

When selecting a Solvent Recovery Unit, there are several factors to consider, including:

• The SLA (Solvent-Laden Air) volume that needs to be treated by the VOC emission control system;
• The type of solvents (within chemicals and/or dilutants) being used and their breakdown;
• The yearly volume of solvent being used at site;
• The demanded stack emission limit, in order to comply with applicable local environmental regulations;
• The budget (CAPEX);
• Performance KPIs to specify and verify with the vendor — solvent recovery rate (see glossary), VOC removal efficiency, recovered-solvent purity, specific energy consumption per kg of solvent recovered, and operational stability under variable/turn-down loading (see §7.1) — rather than a single headline efficiency figure;
• The vendor's after-sales capability — preventive/corrective maintenance, operator training, remote monitoring and diagnostics, and revamping/retrofit support for existing installations (see §11.3) — since long-term performance depends as much on ongoing support as on initial engineering (industry surveys of the flexible-packaging printing sector have found a substantial share of the installed SRU base predates 2000, with energy consumption and operating cost cited as the leading concern by a large majority of operators, underscoring how much value well-planned revamping can unlock on older plants);
• Technology roadmap and future readiness — whether the vendor's platform supports hybrid configurations (adsorption plus dehydration/distillation, see §4.2), Industry 4.0 connectivity (see §14), and capacity expansion without a full plant redesign.

3.1 Decision Tree: Which VOC Treatment Technology Should I Choose?

The flow below distills the selection criteria above into a sequential decision path — a fast way to reach a shortlist before working through the detailed comparison in §4 and the quick decision matrix in §4.0.

This decision tree is a starting heuristic for shortlisting, consistent with — but not a replacement for — the full solvent-by-solvent, site-specific engineering evaluation described throughout this guide.

DEC offers a variety of Solvent Recovery Units (SRUs) to meet the needs of different industrial applications: the processes are safe and very efficient, allowing the recovery a wide range of solvents.

Here are some of the benefits of using a DEC.SRU™ (Solvent Recovery Unit):

• most referenced: DEC is rooted back in 1946, the most-referenced, with a succesful delivery of thousand of installations worldwide, tracing back to the world's first activated-carbon solvent recovery plant for the flexible packaging industry — recovering ethyl acetate via the steam-regenerated DEC.SRU_RSV™ process — which began operating in Italy in 1972;
• award winning innovations, effective design, for fitting specific and/or and high demanding tasks;
• different configurations and flexible layouts: SMS™ (Smart Modular Systems) to CBS™ (Custom Built Systems);
• compliance with global environmental regulations and VOC emissions standards: TVOC values down to ≤ 50 mgC/m³ (or ≤ 20 mgC/m³, for SRUs equipped with DEC.ULE™);
• high recovery rates: DEC Solvent Recovery Units can recover up to 99% of the solvent input;
• easy to maintain and operate;
• low operating costs (OPEX): the activated carbon on-site regeneration process is a low-cost way to regenerate the adsorbent;
• full-electric version available (e-SRU™);
• environmentally friendly: Solvent Recovery Units can help to reduce VOC emissions, improve environmental compliance, including sustainability and net-zero targets;
• generate important savings (ROI, improving your bottom-line);
• guaranteed deliverables;
• extended warranty policies available;
• inspection and maintenance contracts, delivered through DEC SERVICE.

4. Technology Comparison: Activated Carbon Adsorption vs Thermal and Biological VOC Abatement

No single VOC abatement technology is universally optimal. The right choice depends on VOC concentration, solvent recoverability/value, flow rate, and whether the objective is recovery (recycling the solvent) or destruction (converting it to CO₂ and water). The table below compares the principal technologies on the criteria engineers actually use to shortlist a solution.

4.0 Quick Decision Matrix

Where the detailed table above compares technologies feature-by-feature, the matrix below scores each option 1 (poor fit) to 5 (excellent fit) against the four criteria engineers weigh most heavily when shortlisting — useful as a fast first pass, to be confirmed against the site-specific factors in §3 before final selection.

Scores are directional engineering heuristics for first-pass shortlisting, not a substitute for the site-specific evaluation in §3 (SLA volume, solvent type and value, throughput, emission limit, CAPEX budget) or the detailed comparison table above.

4.1 When Does Recovery Beat Destruction?

Activated carbon adsorption is generally the preferred technology whenever the recovered solvent has meaningful reuse or resale value and the VOC concentration is compatible with adsorption (not so dilute that bed sizing becomes uneconomic, not so concentrated/near-saturated that explosion-safety margins dominate design). Thermal oxidation (RTO/DTO/CTO) becomes more attractive when the solvent mixture is too complex or contaminated to be usefully recovered, when no local reuse exists for the recovered solvent, or when destruction efficiency requirements are so stringent that recovery alone cannot guarantee compliance. In practice, many multi-line facilities use both: SRUs on lines with high-value, recoverable single solvents (e.g., toluene-based printing), and thermal oxidation on lines with complex or low-value solvent blends. Liquid scrubbing/absorption (see table above) is worth a closer look specifically where the target solvent is both highly volatile and water- or otherwise liquid-soluble: because the absorbent stripping step is a comparatively conventional distillation/stripping operation rather than a full carbon-bed thermal regeneration cycle, published industry comparisons have put its specific steam demand meaningfully below that of carbon adsorption for equivalent duty — an advantage worth weighing against scrubbing's other limitations (a taller, smaller-diameter stripping column than the absorber itself, a more restrictive choice of absorbent liquid to avoid azeotrope formation and excessive volatility, and generally poorer performance against very dilute or highly variable SLA streams than a carbon bed's inherent buffering capacity provides) rather than dismissed purely on technology-maturity grounds.

4.2 Hybrid Configurations

Rotor/bed concentrators — hydrophobic zeolite, HCA™ or activated-carbon media in DEC.XBC_RBC™ (Rotary Bed Concentrator), DEC.XBC_SBC™ (Static Bed Concentrator) or DEC.XBC_FBC™ (Fluidized Bed Concentrator) configurations, part of DEC's XBC™ range — are frequently used upstream of both SRUs and oxidizers to handle very large, dilute air volumes: the concentrator adsorbs VOCs from a large low-concentration flow and desorbs them into a much smaller, higher-concentration stream that can then be fed to a right-sized downstream SRU or oxidizer, substantially reducing the size (and cost) of the final abatement equipment. The choice between rotary, static and fluidized bed formats depends on airflow, dust loading and pressure-drop constraints at the specific site.

4.3 Non-Regenerative ("Throwaway") Carbon Adsorption

Everything discussed so far in this guide assumes a regenerative system: the carbon bed is desorbed and reused thousands of times over its service life (see §9). For applications with genuinely low total annual VOC mass loading, a non-regenerative ("throwaway") carbon adsorber can be a more economical choice: the carbon bed simply adsorbs until saturated, at which point the spent carbon is replaced (and typically sent for off-site regeneration or disposal) rather than regenerated in place. Because there is no regeneration subsystem — no desorption heating, no condenser, no nitrogen/steam/vacuum infrastructure — CAPEX is substantially lower than a regenerative SRU, at the cost of higher recurring carbon replacement OPEX and no solvent recovery value (the adsorbed VOC leaves with the spent carbon rather than being recovered as reusable liquid). As a general industry rule of thumb, non-regenerative adsorbers tend to make economic sense only below roughly 10 tonnes of total annual VOC mass loading; above that threshold, the carbon replacement cost of a throwaway system typically exceeds the full lifecycle cost of a regenerative SRU, especially once solvent recovery value is accounted for (see §11). Non-regenerative adsorbers are also commonly used for pollutants beyond solvent VOCs — including general hazardous air pollutants (HAPs), odor control, and hydrogen sulfide (H₂S) — where the pollutant has no reuse value regardless of concentration, making regeneration irrelevant to the technology choice. Typical examples include skid-mounted vapor recovery on bulk liquid storage tanks (e.g., bitumen or fuel tanks, where VOC and H₂S vapors displaced during tank filling need treatment — see VRU™ vapor recovery) and odor abatement at wastewater treatment facilities, both cases where intermittent or dilute emissions and the absence of solvent reuse value favor a non-regenerative or throwaway approach over a full regenerative SRU. Fuel and petrol vapor recovery specifically — storage terminals, truck/rail/marine loading racks, and vapor-balance systems — is typically specified against its own regulatory family distinct from the general industrial VOC frameworks discussed in §12, including Germany's TA-Luft concentration limits, the EU's petrol-vapor-recovery Directive (94/63/EC and its amendments, covering Stage I/II vapor balancing), and US EPA loading-based limits for gasoline terminals; a well-specified vapor recovery train, potentially with a polishing second stage, should be sized against whichever of these frameworks applies at the installation's jurisdiction rather than a generic VOC target. Odor-driven projects are typically quantified using olfactometry (odor concentration expressed in odor units, OU/m³, via a panel-based dilution method) or GC/MS analysis of the specific compounds responsible, rather than the TVOC/mgC metrics used for solvent recovery; permitted odor limits are jurisdiction-specific but commonly fall in the range of roughly 5 OU/m³ for residential/inhabited areas and 10 OU/m³ for industrial zones, illustrating why odor abatement is evaluated against a different metric and regulatory framework than VOC mass-based solvent recovery targets, even when the same activated carbon technology is used for both. A further, easily overlooked use case is temporary standby coverage: a modest non-regenerative unit kept on hand (or rented) to maintain emission-control continuity while a site's primary regenerative SRU is offline for scheduled maintenance, a carbon change-out, or an unplanned repair — a lower-cost way to avoid a compliance gap during planned downtime than sizing permanent full-capacity redundancy for an event that, ideally, occurs only occasionally (see the redundancy discussion in §15.1 for the case where continuous full-capacity backup is instead the right answer).

5. Solvent Compatibility Guide

Not all solvents behave the same way on activated carbon, and not all are equally suited to each regeneration method. The table below summarizes engineering considerations for 17 commonly recovered solvents.

5.1 How to Read This Table for Design Purposes

Boiling point is the first-order indicator of desorption difficulty: low-boiling solvents (acetone, methanol, THF) desorb easily but achieve lower working capacity per adsorption cycle, so beds must be sized (or cycle times shortened) accordingly. High-boiling solvents (NMP, DMF, cyclohexanone) adsorb strongly — which is good for capture efficiency — but require more thermal energy (or a vacuum assist) to desorb completely, directly affecting regeneration technology choice (see §9). Water miscibility determines whether the recovered condensate can be used directly or needs a purification step — but miscibility is only a starting point: many of the solvents above also form binary (solvent–water) or ternary (solvent–solvent–water) azeotropes, which fix the composition reachable by simple distillation regardless of feed ratio and must be checked explicitly for the actual solvent blend being recovered, not inferred from miscibility alone. Multi-solvent processes are particularly exposed to this, since a ternary azeotrope between two process solvents and water can behave very differently from either solvent's individual binary azeotrope with water. Finally, flammability/peroxide-forming tendencies (acetone, hexane, THF, MEK) push the design toward inert (nitrogen) regeneration rather than open-air thermal desorption.

A related, practical risk in multi-solvent blends is light-component slip-through: many real-world inks and coating formulations include a proportion of light alcohols (ROH; e.g., ethanol, isopropanol) alongside heavier, more strongly adsorbing solvents. Because these alcohols are smaller, more volatile molecules with comparatively lower working capacity on carbon (see the Ethanol and IPOH rows above), an SLA adsorption temperature that is not properly optimized for the specific blend can allow a disproportionate share of the alcohol fraction to pass through the bed unrecovered, even while the heavier solvent components are captured efficiently — with the practical consequence that recovery efficiency figures quoted for the "main" solvent in a blend can look good while the light-alcohol fraction is quietly under-recovered. This is a design and commissioning verification point (adsorption temperature, bed sizing against the full blend composition, not just its dominant component) rather than a fundamental limitation of carbon adsorption itself.

A related nuance worth flagging specifically for the hexane family: not all isomers behave the same on activated carbon. n-Hexane (boiling point ~69°C) adsorbs well and is the reference case discussed elsewhere in this guide; heptane (boiling point ~98°C) has an even higher adsorption affinity, so an SRU already engineered for n-hexane service can generally handle heptane as well, though heptane's stronger affinity means somewhat more thermal energy — and therefore energy cost — is needed to desorb it cleanly. Iso-hexane, by contrast, has a lower boiling point (~60°C) and a correspondingly weaker affinity for activated carbon than the straight-chain isomer, and its use is generally not recommended for carbon-based recovery: the practical consequence of that weaker affinity is lower adsorption efficiency and higher fugitive emissions for the same bed design, not merely a modest efficiency penalty. Where a process specification allows some latitude in which hexane isomer to purchase, n-hexane or heptane are the better fit for activated-carbon recovery.

5.2 Halogenated (Chlorinated) Solvents

The solvents discussed above are all flammable VOCs, but activated carbon adsorption is also routinely used to recover halogenated (chlorinated) solvents — methylene chloride, chloroform, trichloroethylene, perchloroethylene, 1,1,1-trichloroethane and related compounds — which are chemically distinct from the ester/ketone/alcohol/hydrocarbon families covered elsewhere in this guide. The key engineering difference is that most halogenated solvents are non-flammable, which removes flammability and LEL/ATEX design (see §13) as the dominant safety driver for that specific stream — but this does not make halogenated-solvent recovery simpler overall: many are listed hazardous air pollutants (HAPs) with strict toxicity-driven exposure limits, some legacy compounds (e.g., certain CFCs, carbon tetrachloride) are also regulated as ozone-depleting substances under the Montreal Protocol, and chlorinated streams can generate corrosive byproducts (e.g., HCl) during thermal regeneration that drive different materials-of-construction requirements (see §7.8). This corrosivity consideration also plays into the recovery-versus-destruction decision from §4.1: oxidizing a halogenated stream generates acid gas that pushes a thermal oxidizer toward exotic alloys and dedicated downstream acid-gas treatment, whereas carbon-based recovery avoids combustion altogether and so generally needs less extreme materials of construction for the same solvent — an additional, corrosion-driven argument for recovery over destruction specifically for halogenated streams, on top of the solvent-value argument that applies more broadly. Halogenated-solvent recovery should therefore be scoped as its own case, with toxicity/regulatory status and corrosion allowance replacing flammability as the primary design drivers. Where a halogenated compound genuinely isn't economical or practical to recover intact — a trace contaminant in an otherwise low-value stream, for instance — steam reforming (reacting the carbon-sorbed halogenated compound with superheated steam to destroy it rather than desorb it intact) is a destructive alternative sometimes used specifically for spent-carbon or off-gas streams too dilute or mixed to justify recovery; this sits outside DEC's core recovery scope and would typically be evaluated as a separate destructive-treatment path rather than integrated into the same regeneration cycle as recoverable solvent.

6. Industrial Applications

Solvent recovery is applied across a wide range of industries, each with characteristic VOC mixes, emission profiles and preferred technology configurations.

6.1 Flexible Packaging

Flexible packaging converting — laminating, pouch and film production, substrate coating — uses large volumes of ethyl acetate, ethanol, toluene, IPOH and MEK-based adhesives and coatings, generating high-volume, moderate-concentration SLA streams. Activated carbon SRUs are the dominant technology because solvent purchase volumes are large and recovered solvent is routinely reused directly in the process, giving fast payback. Engineering challenges include managing multi-solvent blends and matching regeneration technology to the dominant solvent in the mix.

6.2 Printing

Rotogravure, flexographic and heliogravure printing presses — including publication printing (magazines, catalogues, inserts) alongside packaging and label printing — use ethyl acetate, ethanol, toluene, IPOH and MEK-based inks at high volumes, generating exhaust streams similar in character to flexible packaging but driven by ink formulation rather than substrate coating (see also the rotogravure, flexo and lamination case studies). Publication printing in particular tends to run at large scale with high press utilization, which generally strengthens the payback case for solvent recovery over destruction given the volume of solvent purchased annually. Engineering challenges include managing multi-solvent ink blends and matching regeneration technology to the dominant solvent in the formulation.

6.3 Coating

Industrial and functional coating lines — including coil coating, wire coating, wood coating, textile coating, rubber coating, furniture finishing, magnetic-tape manufacturing, weatherproof sheeting/membrane coating, carbonless (carbon-copy) paper coating and protective/specialty coatings — emit toluene, butyl/propyl acetate and ketone-based (MEK) VOCs, typically at higher and more consistent concentrations than paint spray booths. Activated carbon SRUs recover these solvents efficiently given their strong adsorption behavior and the higher-value reuse case in the coating formulation.

6.4 Adhesives

Solvent-based adhesive manufacturing and application (lamination, tape coating, synthetic leather, industrial belting, composite gaskets) generate ethyl/butyl acetate, toluene and hexane emissions comparable to flexible-packaging and printing operations, with similar SRU suitability.

6.5 Automotive

Automotive OEM and Tier-1 paint shops combine high air volumes from spray booths with adhesive and sealant application processes; VOC abatement is frequently a mix of RTO for booth exhaust and SRUs for concentrated, high-value solvent streams from specific process steps. Friction-material manufacturing (brake pads, clutch facings) is a related automotive-supply-chain application, with its own characteristic solvent-laden exhaust from resin/binder processing well suited to activated carbon recovery.

6.6 Aerospace

Composite layup, adhesive bonding and specialty coating processes in aerospace manufacturing use high-purity solvents (acetone, MEK, IPOH) where both emission control and solvent traceability/purity are important, favoring carbon adsorption with careful purity-focused condensate handling.

6.7 Chemical Industry

Chemical process vents can contain complex, often single-solvent, high-concentration streams (e.g., reactor vents, dryer exhausts, solvent degreasing operations) well suited to solvent recovery, particularly where the same solvent is reused as a process input (e.g., hexane in extraction, toluene in synthesis). Hydrogen peroxide production via the anthraquinone process is a specific example: the process circulates alkylated anthraquinone compounds dissolved in an organic working solution, and recovering solvent carried into the raw material off-gas is a recognized application for activated carbon TSA systems in this sector.

6.8 Pharmaceutical Industry

Pharmaceutical manufacturing frequently uses high-purity solvents (acetonitrile, methanol, THF, acetone, IPOH) in synthesis, purification and cleaning — including tablet coating and surgical/wound-dressing product manufacturing; GMP considerations around cross-contamination and solvent traceability make dedicated, well-characterized recovery trains particularly valuable, and vacuum regeneration is often favored for heat-sensitive or high-boiling solvents. Cross-contamination risk is, in practice, one of the main reasons many pharmaceutical sites default to incinerating or externally destroying waste solvent streams rather than recovering them, even though destruction is typically both more expensive per unit of solvent and a source of avoidable GHG emissions compared with recovery; a dedicated, single-solvent (or well-characterized multi-solvent) recovery train specifically addresses that contamination concern rather than requiring recovery to be abandoned altogether. For multi-solvent waste streams where distillation-based separation is being evaluated, engineering teams sometimes use process-simulation software (e.g., Aspen Plus/Aspen Batch Modeler) to model the separation and validate column design before specifying equipment — a useful due-diligence step for any multi-solvent recovery train, not only in pharma.

6.9 Battery Manufacturing

Lithium-ion electrode coating (NMP-based cathode processes, increasingly aqueous anode processes) generates high-value NMP-laden air streams; NMP's high boiling point makes vacuum-assisted regeneration a common choice, and recovery economics are typically very strong given NMP's cost and consumption volumes (see the NMP solvent recovery case study).

6.10 Electronics

Cleaning and coating processes in electronics and semiconductor manufacturing use IPOH, acetone and other high-purity solvents; low-contamination recovery trains and careful moisture management are priorities given sensitivity to trace residues.

6.11 Paint

Paint application — spray booths for architectural, industrial and general-finishing paints — typically emits xylene, toluene and glycol-ether-based VOCs across large exhaust air volumes but at comparatively low, humidity-affected concentrations. Because booth exhaust volumes are large relative to VOC mass loading, paint lines are among the most common candidates for a concentrator-plus-SRU or concentrator-plus-oxidizer hybrid configuration (see §4.2), substantially reducing the size of the final abatement equipment.

6.12 Composites

Composite manufacturing (styrene-based resin systems, solvent cleaning of tooling) presents polymerization-fouling risks on carbon beds (see §7.7) that must be specifically engineered around, sometimes favoring alternative technologies or specialized carbon-protection measures.

6.13 Refinery & Petrochemical Process Wastewater (BTEX Recovery)

A related but distinct application worth being aware of applies the same vapor-phase carbon adsorption principle to a liquid stream rather than SLA directly: recovering benzene, toluene, ethylbenzene and xylenes (BTEX) that are dissolved — not merely entrained — in refinery and petrochemical process wastewater, most commonly desalter effluent, which at many refineries accounts for a disproportionate share of total wastewater BTEX loading. Because benzene is both a VOC and a listed Hazardous Air Pollutant, U.S. NESHAP rules for benzene waste operations require aromatics-bearing wastewater above a fairly low annual mass threshold to be controlled to a high capture/destruction efficiency, and discharge limits are commonly specified in the sub-ppm range — a combination that rules out simple dilution as a compliance strategy. Where conventional treatment options (desalter emulsion-breaker optimization, liquid-phase activated carbon, steam stripping, or air stripping) each carry their own capital/operating-cost or safety trade-off, one configuration couples a nitrogen-stripping column — inerted rather than air-based specifically to keep the stripped-gas loop outside the flammable envelope — with regenerable vapor-phase carbon adsorbers on a closed nitrogen loop, desorbed in place by live steam. In documented industrial operating experience, systems of this kind have treated wastewater flows from roughly 100 up to several thousand GPM, reducing inlet BTEX loadings on the order of tens of ppmw down to well under 1 ppmw at discharge, with carbon bed change-out driven mainly by pore fouling from higher-boiling co-contaminants and — where hydrogen sulfide is also present in the feed — by chemisorbed elemental sulfur from H₂S reacting with any trace oxygen in the loop, which is itself a further reason to keep the stripping loop inerted rather than air-based. This is a genuinely different unit operation from the air-phase SLA recovery covered throughout the rest of this guide — the feed is a liquid, not an SLA duct — but it shares the same core adsorption/desorption/condensation building blocks, and the wastewater-treatment stripping-column considerations discussed in §9.2 are directly relevant to evaluating it as an option.

7. Engineering Design Considerations

7.1 Airflow & VOC Concentration

System sizing starts from the total SLA airflow (Nm³/h) and the VOC concentration profile (average and peak), which together define both the mass loading the carbon must handle and the bed cross-section needed to keep face velocity — and pressure drop — within acceptable limits. Collection design deliberately targets a specific VOC concentration range at the source: extraction (ventilation) volume is typically controlled so that solvent concentration in the collected SLA is maintained near a constant level compatible with, but safely below, the LEL — commonly around 30–40% of the LEL as an accepted operating design target in Europe — since operating too dilute increases bed size and cost for a given solvent mass, while operating too close to the LEL erodes the safety margin discussed in §13.3. Sizing on peak values alone is not sufficient: the minimum airflow and minimum VOC concentration the plant will actually see in service must also be defined, since these set the required turn-down ratio (the ratio between maximum and minimum design flow/load the SRU must handle reliably). At the low end of that range, face velocity can drop enough to worsen channeling and widen the mass transfer zone (see §8.5), while VOC concentration can fall low enough to weaken breakthrough detection. As a practical minimum, most SRUs are designed to turn down reliably to roughly 10–25% of rated SLA airflow without instability; process lines with wider swings (e.g., batch operations, intermittent dryers) should specify the actual minimum expected airflow and VOC concentration explicitly, rather than assume the plant will scale down linearly from the peak design point, since fan, valve and instrumentation turn-down capability — not just bed sizing — govern how low the system can actually go.

Turning the 30–40%-of-LEL design target into an actual SLA concentration setpoint requires the solvent's LEL expressed as a mass concentration (g/m³), not just a volume percentage — the table below gives illustrative values (LEL in g of solvent per m³ of solvent-vapor/air mixture, at 20°C) for solvents commonly recovered by activated carbon; site-specific design should always confirm the figure against the actual solvent grade and temperature in use rather than relying on a generic table.

Halogenated (chlorinated) solvents — methylene chloride, chloroform, trichloroethylene, perchloroethylene, 1,1,1-trichloroethane, carbon tetrachloride — are generally non-flammable and so are not LEL-limited in the same way (see §5.2); their design driver is toxicity/HAP status and corrosion, not explosion safety. This table is illustrative, not exhaustive — practically all industrially used solvents are recoverable, and site-specific LEL data should be confirmed against the actual solvent grade in use.

7.2 Humidity

Water vapor competes with VOC molecules for adsorption sites on activated carbon, reducing working capacity for hydrophilic carbons at high relative humidity; humidity must therefore be characterized and, where excessive, addressed with pre-cooling/dehumidification or carbon grades selected for humidity tolerance. As a practical benchmark, the effect is non-linear rather than gradual: above roughly 60–70% relative humidity, the amount of water co-adsorbed onto the carbon tends to increase sharply rather than proportionally, displacing VOC adsorption capacity and — if not accounted for in the pre-cooling/dehumidification design — creating a real risk of operational malfunctions rather than just a modest efficiency loss. A related design subtlety: cooling SLA to control temperature (see §7.11) raises its relative humidity for a given absolute moisture content, since cooler air holds less water vapor at saturation — so a pre-cooling stage aimed purely at temperature can inadvertently push RH into the problematic range discussed above even without adding any moisture to the stream. Where this interaction matters, an SLA re-heat stage after cooling (bringing the temperature back up slightly, without adding humidity) can be used to bring relative humidity back down to an acceptable adsorption range, decoupling temperature control from humidity control rather than accepting whichever RH the cooling step happens to produce.

A specific, increasingly common version of this problem arises with so-called "water-based" printing inks in flexography and gravure on non-porous films (plastic substrates), which — despite the label — are frequently better described as hydro-solvent formulations: pure water evaporates too slowly and wets non-porous film surfaces too poorly on its own, so a substantial organic co-solvent fraction (commonly ethanol or propanol) is added to restore printability, drying speed and pigment dispersion, and that organic fraction can be a large share — sometimes the majority — of the volatile content. For an SRU sizing exercise, this means the SLA stream is not the low-VOC stream the "water-based" label implies: it is a water/alcohol mixture in which the water directly competes with the target alcohol for carbon adsorption sites exactly as described above, reducing effective bed capacity for the VOC actually worth recovering, while the co-adsorbed water then has to be separated from the recovered alcohol downstream via dehydration and, where a difficult azeotrope is present, distillation (§1.4). In practice this typically means an SRU sized for a genuinely hydro-solvent stream ends up larger, with a more energy-intensive regeneration and purification train, than a like-for-like unit sized for the same VOC mass loading from a conventional solvent-based (acetate/alcohol) ink — a sizing distinction worth raising explicitly with converters evaluating a line switch (see the water-based ink FAQ below).

7.3 Oxygen Concentration

Oxygen concentration in the regeneration atmosphere is a primary safety parameter for inert (nitrogen) systems: oxygen ingress must be kept below the design threshold to stay outside the flammable envelope at the VOC concentrations reached during desorption (see §13).

7.4 Pressure Drop

Bed depth, particle size and face velocity all trade off against pressure drop, which drives fan sizing and electrical energy consumption; pressure drop typically increases over the carbon's service life as fines accumulate and channeling or partial fouling develops.

7.5 Solvent Mixtures

Multi-solvent streams desorb at different rates and temperatures; regeneration must be designed (temperature profile, purge duration) around the least-easily-desorbed component to avoid progressive accumulation ("heel build-up") of heavier solvents on the carbon over repeated cycles. The underlying mechanism worth understanding, not just the symptom: within a multi-solvent blend, the heavier (higher-molecular-weight, generally less volatile) component is held more strongly by the carbon and will progressively displace a lighter, more volatile co-adsorbed solvent as the bed approaches saturation — so in a blend, it is characteristically the lightest component that breaks through first and is most exposed to under-recovery, which is the same underlying effect behind the light-alcohol slip-through risk discussed in §5.1, not a separate phenomenon.

7.6 Temperature

Inlet air temperature affects adsorption capacity (higher temperature generally reduces physisorption capacity) and must be considered alongside seasonal variation and any upstream process heat. As a practical benchmark, adsorption performance for many solvents falls off sharply above roughly +35°C SLA temperature — on the order of -3% capacity per +1°C increase beyond that point — which makes reliable pre-cooling (see §7.11) a genuine performance issue in warm climates or during summer operation, not just a comfort margin.

7.7 Fouling, Dust & Polymerization

Particulates (dust, overspray, fibers) foul the carbon bed surface and increase pressure drop; some solvents (styrene, unsaturated resins, certain acrylates) can polymerize on the hot carbon surface during regeneration, progressively blinding pores. Upstream filtration and, where polymerization risk exists, temperature-limited regeneration profiles or alternative technologies are used to mitigate this.

7.8 Corrosion

Chlorinated or acidic VOC streams can generate corrosive byproducts (e.g., HCl) during thermal regeneration; materials of construction (stainless steel grades, coatings) must be selected accordingly. Internal components in direct, sustained contact with sulfur- or chloride-bearing vapor — such as the carbon-bed support screens — sometimes warrant higher-alloy materials than the vessel shell itself: nickel alloys such as Incoloy 825 are an established choice for carbon support screens in this service, selected both for resistance to sulfidic/chloride attack and to avoid galvanic corrosion between the carbon and a dissimilar-metal support structure. Vessel shells are frequently carbon steel with a defined corrosion allowance rather than solid stainless steel, which is often adequate provided the specific corrosion mechanisms present (chlorides, sulfur compounds, condensate pH) have been characterized for the actual solvent mix rather than assumed generically — mild (carbon) steel is generally a satisfactory choice for hydrocarbon-only service, while wetted parts handling ketones and esters specifically warrant stainless steel given those solvents' tendency toward acidic breakdown products in service (see the acetic-acid hydrolysis and MEK-oxidation discussion in §8.6). A related, easily overlooked corrosion pathway is specific to inhibited chlorinated solvents: many chlorinated hydrocarbons are supplied with a stabilizing inhibitor package precisely because the uninhibited compound is corrosive in the presence of water, and if that inhibitor is depleted, stripped out, or simply not replenished somewhere in the recovery/water-contacting train, the chlorinated solvent can become meaningfully more corrosive than its data sheet would suggest — non-metallic linings are sometimes warranted for the specific vessels or piping runs most exposed to this risk, and inhibitor fate through the full adsorption/desorption/water-contacting cycle is worth tracking explicitly for these solvents rather than assumed to remain intact throughout.

7.9 Utility Availability

Available site utilities — electrical capacity, cooling water/chilled water (CWS/BCS), thermal fluid (TFS, hot oil or steam), nitrogen supply/generation (N2X) — directly influence which regeneration technology (§9) is most cost-effective at a given site.

7.10 Activated Carbon Quantity & Sizing

More carbon is not automatically better. A carbon bed is sized against a specific SLA airflow and VOC concentration profile, and the resulting face velocity and residence time through the bed are part of that calculation, not an afterthought. Oversizing the carbon quantity for a given vessel cross-section and airflow reduces the linear velocity of the SLA through the bed and increases residence time, which sounds protective but can actually reduce the effective utilization of the carbon's adsorption capacity relative to a correctly sized bed, while also increasing pressure drop (see §7.4) and therefore fan power consumption. Undersizing, conversely, risks early breakthrough (see §8.4) under peak or upset conditions. Because flow rate, VOC concentration and temperature can all vary meaningfully in real operation — flexible packaging converting lines are a typical example — carbon quantity and bed geometry should be sized (and, where conditions vary widely, controlled) against the actual expected operating envelope, including its minimum and maximum (see turn-down ratio, §7.1), rather than against a single nominal design point with an arbitrary safety margin added on top. A supplier quoting extra carbon as a blanket margin is worth a direct question: is the added quantity compensating for a specific, named design gap (e.g., humidity control, SLA pre-cooling, carbon quality) that should instead be addressed directly? To make these figures concrete rather than abstract, a documented large multi-adsorber installation (six vessels: four adsorbing, one regenerating, one cooling/standby at any given time, treating on the order of 75,000–100,000+ SCFM total SLA) has operated with roughly 37,000–38,000 lbs of carbon and a bed depth on the order of 40 inches per vessel, cycling through adsorption (on the order of two hours), steam regeneration (on the order of 40 minutes) and cooling (on the order of 20–25 minutes) before returning to service — illustrating the order of magnitude a large industrial installation's bed inventory and cycle timing actually falls into, for benchmarking a vendor's proposed design against. As a much rougher, order-of-magnitude rule of thumb useful only for an early feasibility estimate (not final sizing), a twin-bed steam-regenerated installation running roughly 8,000 hours per year has been documented needing on the order of a few tonnes of installed carbon per 1,000 tonnes/year of solvent throughput handled — a starting point for a first-pass carbon-inventory estimate before a proper mass-and-energy balance is run against the actual airflow, concentration and cycle-time profile. Bed orientation is also worth flagging explicitly at the sizing stage: activated carbon is a comparatively light material, and a bed design that routes gas flow upward through a loose or lightly-packed carbon bed risks fluidizing (lifting and disturbing) the carbon rather than keeping it as a stable fixed bed — this is one of the reasons fixed-bed SRU adsorbers are engineered around a specific flow direction and bed-retention design (support screen, hold-down provisions) rather than assumed to tolerate flow reversal without consequence.

7.11 Cooling System Selection

An SRU typically has several distinct cooling duties, not just one: SLA pre-cooling ahead of adsorption (bringing incoming air from a typical +50–70°C down to the +30–35°C range that favors adsorption, depending on relative humidity), pre-cooling of the regeneration loop (nitrogen or vapor) ahead of final condensation, cooling for any molecular-sieve regeneration loop, condensation duty for a distillation column's overhead vapor, and the condenser-side cooling load for a chiller (BCS) where refrigeration is used. Common cooling system options — collectively falling under DEC's CWS (Cooling Water Systems) product code — include open cooling towers, closed-circuit (dry) coolers, air-cooled chillers, and adiabatic (hybrid wet/dry) systems, each with different water consumption, footprint, and climate sensitivity. Open cooling towers are generally the most space- and energy-efficient option in warm climates, but consume make-up water and require water treatment; dry coolers and air-cooled chillers avoid water consumption but their achievable approach temperature is climate-limited — a dry-air cooler typically cannot cool a water circuit closer than about 10–15°C above ambient temperature, so in warm/hot regions (ambient above roughly +20°C) they may not reliably reach the target SLA or condensation temperatures, with direct knock-on effects on adsorption capacity (see §7.6) — particularly for multi-solvent blends where under-cooling can allow lighter components to slip through unrecovered (see §5.1). The right choice is climate- and site-specific rather than universal, and should be evaluated against local water availability and ambient temperature profile, not selected solely to minimize water use on paper.

7.12 SLA Filtration

Particulate filtration of the SLA stream upstream of the carbon bed protects downstream components (heat exchangers, fans) and extends carbon service life by preventing particulate fouling of the bed surface (see §7.7). Static (cell-type) filters are simple but require periodic manual replacement, which — depending on filter housing design — can require entry into a confined filtration enclosure; automatic or continuous roll-media filtration systems avoid this by advancing a fresh section of filter media without stopping the process or requiring operator entry, and typically offer a substantially longer service interval between media changes than static cells. Where static filters are used, confined-space entry procedures should account for the risk of localized oxygen depletion inside enclosed filtration housings, particularly on nitrogen-regenerated systems (see §13.2).

7.13 Quick-Reference: Typical Engineering Ranges

The table below consolidates the numerical benchmarks discussed throughout this guide into a single reference; each figure is a general engineering illustration to be confirmed against the specific solvent, carbon grade and site conditions in play — not a substitute for a project-specific mass and energy balance.

8. Activated Carbon Engineering

8.1 Pore Structure

Activated carbon's performance is governed by its pore structure: micropores (<2 nm) provide most of the adsorption surface area and dominate capacity for small VOC molecules, mesopores (2–50 nm) assist transport into the micropore network, and macropores (>50 nm) act as feeder channels. Carbon grades are selected/activated to match the pore-size distribution to the target VOC molecule size.

8.2 BET Surface Area

BET (Brunauer–Emmett–Teller) surface area, typically 800–1,500+ m²/g for VOC-recovery grade carbons, quantifies the total available adsorption surface per unit mass and correlates broadly with adsorption capacity, though pore-size match to the target molecule matters as much as raw surface area.

It is worth being explicit that "activated carbon" is not one product spec: regenerative solvent-recovery duty and once-through/throwaway air-and-gas-cleaning duty (see §4.3) call for measurably different carbon grades, even though both are gas-phase adsorbents. Solvent-recovery-grade carbon is typically formulated (coal, wood or coconut-shell based, extruded into ~2–4 mm cylindrical pellets to minimize flow resistance) toward a well-balanced micro/meso/macropore structure that favors both high loading and clean desorbability, with a correspondingly lighter bulk density (roughly 350–430 kg/m³) and a benzene-from-air adsorption capacity (isotherm basis, p/p₀ ≈ 0.9) often in the 40–60 wt% range. Once-through gas-cleaning-grade carbon, by contrast, is optimized purely for retention (no desorption cycle to design around), typically bituminous-coal or coconut-shell based, denser (roughly 470–570 kg/m³) with a correspondingly lower benzene-adsorption capacity in the 20–40 wt% range and a lower CTC activity band (roughly 35–60 wt% versus ~65–90 wt% for recovery-grade carbon). Specifying a once-through-style carbon into a regenerative SRU (or vice versa) is a common and costly sizing error worth flagging explicitly to a carbon supplier at the quotation stage, rather than assuming any "activated carbon" line item is interchangeable.

8.3 Adsorption Index (Vapor-Phase Activity)

The iodine number is a widely quoted carbon-quality figure, but it is a liquid-phase test and is not the appropriate index for VOC-recovery (vapor-phase/gas-phase) activated carbon; using it to specify or compare gas-phase adsorbents can be misleading. For gas-phase adsorbents, DEC specifies and benchmarks carbon quality using vapor-phase adsorption indices instead: the butane index (ASTM D5742), the CTC (carbon tetrachloride) index (ASTM D3467 — historically the most common and accurate KPI, though carbon tetrachloride's status as a known carcinogen means it is no longer used as a standard reference test), the ETAC (ethyl acetate) index, and the benzene index (JIS K 1474). A higher adsorption index indicates higher activity and a greater capacity for adsorbing VOC molecules. As a general benchmark, solvent-recovery-grade pelletized carbon is typically specified with a CTC index in the range of roughly 55–60%+ (higher for heavier solvents such as toluene) and a butane working capacity of roughly 10 g/100 mL or more, though the appropriate target for a given installation should be verified against the specific solvent(s) and duty rather than assumed generically. See DEC's activated carbon performance reference for the full methodology behind each index. Where a design or performance evaluation is based on adsorption/isotherm testing rather than published index values alone, it's worth confirming whether the test was run on virgin or regenerated carbon: regenerated carbon typically costs less than virgin material but can have measurably lower adsorptive capacity, so a capacity figure generated on virgin carbon can overstate what a plant running on regenerated carbon will actually achieve in service — testing (or at minimum specifying) against the carbon condition the plant will actually operate on gives a more realistic design basis than virgin-carbon data alone.

8.4 Breakthrough Curves

A breakthrough curve plots outlet VOC concentration versus time (or bed volumes treated) for a bed under constant inlet conditions; it starts near zero, then rises as the bed's adsorption capacity is consumed, defining the "breakthrough point" at which the bed must be switched to regeneration to avoid VOC slip to atmosphere. Breakthrough curve shape depends on the mass transfer zone (§8.5), flow rate, and adsorption kinetics.

8.5 Mass Transfer Zone (MTZ) & Working Capacity

The Mass Transfer Zone is the portion of the bed actively transferring VOC from gas to solid phase at any moment; upstream of the MTZ the carbon is saturated, downstream it is still fresh. A narrow MTZ (fast kinetics, well-matched pore structure) allows a bed to be used closer to its full equilibrium capacity before breakthrough. "Working capacity" — the capacity actually usable in cyclic service, always lower than fresh-carbon equilibrium capacity — is the number that should drive bed sizing, not laboratory-measured maximum capacity.

8.6 Carbon Aging & Replacement

Carbon capacity declines over years of service due to irreversible fouling (heavy residues, polymerization products, oxidation byproducts) that block pores and cannot be removed by normal regeneration; this is distinct from the reversible loading/unloading of the operating cycle. Aging rate depends on solvent chemistry, regeneration completeness, and any contaminants in the SLA stream; carbon replacement intervals typically range from several years to a decade-plus depending on service severity — illustratively on the order of thousands of regeneration cycles for well-maintained pelletized carbon in typical solvent-recovery service — and should be verified periodically against breakthrough performance rather than assumed on a fixed calendar. When replacement is needed, exhausted carbon does not necessarily need to be discarded entirely for virgin material: extraction, screening and off-site reactivation (thermal reprocessing, typically in a rotary kiln at several hundred degrees Celsius, that can restore a large majority of the carbon's original adsorption capacity) — services of the kind delivered through adsorbent maintenance services (AMS) — is a circular-economy-consistent alternative worth evaluating, subject to the specific carbon grade and contamination history supporting reactivation. A related but distinct mechanism, sometimes called adsorption hysteresis, occurs with certain solvents — MEK is a well-documented example (see §5) — that are not fully stripped by a standard regeneration cycle even when the carbon itself is not yet fouled; the residual buildup this leaves behind, cycle after cycle, progressively erodes usable capacity. The residue driving this buildup is not always just unreacted solvent: the very high surface area that makes activated carbon an effective adsorbent can also let it act as a mild catalyst for side reactions among adsorbed species, generating higher-boiling reaction byproducts in place on the carbon — acetic acid formed from the hydrolysis of acetate esters (ethyl acetate, isopropyl acetate, n-propyl acetate) is a well-documented example (see the molecular-sieve hydrolysis discussion in §1.7, where the same underlying catalytic mechanism applies to a different adsorbent) — for ethyl acetate specifically, the reaction is CH₃COOC₂H₅ + H₂O → CH₃COOH + C₂H₅OH, and it is water availability, not just temperature, that drives it: documented steam-regenerated plant experience has found average acetic acid content in the process water on the order of 1% by weight, which on its own corresponds to a Chemical Oxygen Demand (COD) well above 10,000 mg/l — a genuinely significant water-quality burden, and one reason the wastewater treatment cost discussed for steam regeneration in §9.2 is driven as much by hydrolysis byproducts as by residual unreacted solvent. As a rough order-of-magnitude conversion, roughly 6% of the ethyl acetate throughput ends up hydrolyzed for every 1% (by weight) of acetic acid generated in the process water — a real yield loss on top of the water-treatment cost, and one that is essentially eliminated under nitrogen (dry-gas) regeneration, since without steam's water content the hydrolysis reaction has little to react with even at a materially higher regeneration temperature. These higher-boiling byproducts are typically harder to desorb than the parent solvent, so they contribute disproportionately to residual buildup even when the bulk solvent load is being adequately recovered. Some carbon-catalyzed side reactions are also exothermic rather than heat-neutral — peer-reviewed research on MEK specifically has documented an oxidative degradation pathway on the carbon surface (MEK + O₂ → 2,3-butanedione [BDO, i.e. diacetyl] + H₂O) that occurs under oxygen-containing, elevated-temperature conditions (typically ≥100°C), is thermodynamically favorable, and releases roughly 56 kJ/mol (~780 J/g) of adsorption energy in the process — notably higher than the roughly 20 kJ/mol (~5 kcal/mol) order-of-magnitude heat of adsorption typical of ordinary, non-reactive physisorption on activated carbon, consistent with this being genuine reaction exotherm layered on top of ordinary adsorption heat rather than physisorption alone — which is a second, distinct reason these reactions matter beyond capacity loss: BDO is itself a yellow-to-yellow-green, odor-causing contaminant that degrades recovered-solvent quality, and the localized heat released can contribute to hot-spot formation within the bed, feeding into the fire-risk considerations discussed in §13.4, on top of the residue itself narrowing the working capacity available for the next adsorption cycle. The same research found that switching from steam to nitrogen as the desorption and purge medium substantially reduced BDO formation compared with air-purged steam desorption — in controlled testing, BDO concentration in the recovered MEK fell from roughly 0.1–0.2 wt% under conventional steam/air-purge conditions to as low as 0.01–0.04 wt% under nitrogen desorption and purging, an order-of-magnitude improvement in recovered-solvent quality from the regeneration-medium choice alone — consistent with why DEC's own compatibility guidance (§5) favors nitrogen regeneration for MEK — and that commercial carbon grades vary meaningfully in how much BDO they generate, with activated-carbon-fiber media producing markedly less than conventional granular carbon, though at higher material cost. To put rough numbers on that trade-off: unmodified commercial gas-phase carbons typically test with an ignition point in the roughly 450–500°C range (measured by thermogravimetric analysis) — though this laboratory figure characterizes clean carbon under controlled test conditions, and industry field experience with ketone-driven hot spots specifically has long flagged a materially lower practical ignition threshold, on the order of 370°C, for a carbon bed already primed by localized reactive self-heating; the gap between the two figures is itself a useful engineering reminder that a clean-carbon TGA ignition point is a starting reference, not a hot-spot safety margin to design an interlock setpoint against directly, and among modifiers evaluated in the peer-reviewed research, magnesium oxide raised the ignition point by roughly 10–15°C while amine-based modifiers shifted it only marginally — but potassium nitrate, despite being a plausible-sounding flame-retardant-style additive, lowered the ignition point by close to 30°C, illustrating that a modifier's effect on ignition safety cannot be assumed from its chemistry alone and should be verified experimentally rather than taken on a supplier's claim. The same body of research also carries a more severe specific safety caution worth noting directly: certain carbon surface modifications intended to improve performance can backfire badly on ignition safety — one heavily KOH-loaded modified carbon tested dropped to an ignition point in the region of roughly 280°C, low enough that it visibly auto-ignited (smoke observed at the bed outlet) simply from being cooled with room-temperature compressed air after use, underscoring that not every carbon-modification approach marketed as a performance or quality improvement should be assumed safe for ketone service without independent verification. Older approaches to this problem required unpacking the carbon and reconditioning it in a special external furnace at very high temperature — costly, slow, and requiring the bed to be taken fully out of service. Within the nitrogen-regenerated RSG™ process, DEC addresses this in place with DEC.DTD™ (Deep Thermal Desorption): a controlled, elevated-temperature desorption cycle — beyond the temperature used in normal regeneration — that recovers solvent molecules strongly retained within the activated carbon, including those associated with adsorption hysteresis, directly within the existing RSG™ vessel and without special furnace equipment or extended downtime. This aims to restore the carbon's original adsorption capacity and extends adsorbent service life by reducing replacement frequency.

8.7 Alternative Adsorbent Media

Activated carbon is not the only viable adsorbent for VOC recovery. Molecular sieves (zeolites) offer a highly uniform, hydrophobic pore structure well suited to dehydration and to VOC streams where humidity tolerance matters more than raw capacity (see §1.7). Macroporous polymer resin adsorbents (PRA) are another alternative: their weaker van der Waals adsorption bonds relative to activated carbon mean less energy is required to desorb the same mass of solvent, which can lower regeneration energy consumption, at the trade-off of generally lower adsorption capacity and different selectivity characteristics than carbon for a given solvent. Beyond material type, the physical form factor of the carbon itself is a further variable: activated-carbon fiber (in mat or cloth form) has also been used as an alternative to conventional granular or pelletized carbon, offering faster adsorption/desorption kinetics from its finer structure, though it is a more specialized and typically higher-cost option than granular carbon for most conventional SRU duties. The right adsorbent choice is therefore solvent- and objective-specific — maximum capacity and cost-per-kg favor activated carbon in most conventional applications, while minimum desorption energy or specific humidity/selectivity requirements can favor molecular sieves or polymer resins instead — and should be evaluated against the actual target VOC(s) rather than assumed by default.

9. Regeneration Technologies: Nitrogen, Steam & Vacuum

Regeneration technology selection is one of the most consequential SRU design decisions, driving CAPEX, OPEX, solvent quality, and safety architecture.

9.1 Nitrogen (Inert Gas) Regeneration — TSA with N₂ (e.g., DEC.SRU_RSG™)

Nitrogen regeneration is also referred to as steamless regeneration or dry regeneration, since — unlike steam regeneration (§9.2) — no water is introduced into the desorption loop at any point in the cycle. A closed loop of heated nitrogen strips solvent from the carbon bed; the nitrogen is then cooled to condense the solvent and recirculated. Advantages: excellent for flammable, oxidation-sensitive or peroxide-forming solvents (acetone, MEK, THF, hexane) since the atmosphere stays inert throughout the cycle; produces a dry, non-aqueous condensate for water-immiscible solvents, easing purification; the same inert nitrogen loop also enables in-place high-boiling-residue reconditioning via DEC.DTD™ when needed (see §8.6), without special furnace equipment; unlike steam regeneration (§9.2), there is no need to produce steam on-site, since heating can be supplied indirectly through a Thermal Fluid System (TFS) or, in the full-electric e-SRU™ configuration, entirely by electrical heating — avoiding the boiler, resin-exchanger and makeup-water dependencies described for steam regeneration (see §9.2). A trade-off worth naming explicitly is that, at a given regeneration temperature, hot-gas desorption is generally somewhat less complete than live-steam desorption, since steam's condensing latent heat and its own displacement of adsorbed species both aid stripping in ways a hot inert gas alone does not fully replicate — a difference that mainly matters where residual carryover between distinct solvent campaigns on the same shared carbon is a specific concern (e.g., toll or contract recovery operations processing different customers' solvents through the same equipment), and is one more reason a plant handling frequent, distinct solvent campaigns should verify actual inter-campaign carryover experimentally rather than assume it from the base regeneration technology's typical single-solvent performance. Disadvantages: requires a nitrogen supply or on-site generation; note that since the RSG™ loop runs slightly overpressurized with nitrogen, leak-tightness is not a critical safety issue in the way it is for vacuum systems — any leakage tends to show up as higher nitrogen make-up consumption rather than oxygen ingress — though oxygen-ingress monitoring is still maintained as a standard safeguard (see §13.2). Operating costs: driven by regeneration heating energy and nitrogen make-up losses; generally moderate.

The first step of a nitrogen regeneration cycle, for solvents with a stable behaviour and an adequately high autoignition temperature, is a dedicated gas-phase pre-dehydration stage — at DEC, DEC.ADM_PHD™ (DEC Pulse Heat Dehydration), running a proprietary selective-desorption kinetic algorithm (DEC.KDA™): over a few minutes, the bed is heated with hot SLA (rather than the main regeneration nitrogen) to drive off most of the water the carbon adsorbed from ambient humidity, before the main desorption step begins, with the process continuously LEL-monitored to keep solvent concentration in that circuit under control, while monitoring the temperature trend and cycle time. Doing this in the gas phase, ahead of the main condensation step, is a deliberate design choice: an alternative process DEC.ADM_FDP™ (DEC Fractional Dehydration Process), used for unstable solvents, pre-condenses this water and then re-evaporates it back into the filtration room using DEC.ADM_WPA™ module, which — compared to DEC.ADM_PHD™ does — pays the energy cost of condensation for water that was never going to be recovered as product in the first place. The practical payoff of doing it correctly is a substantially drier raw condensate reaching the main condenser, and therefore less energy spent condensing water that is not the target solvent. For solvents with a comparatively low autoignition temperature such as hexane (see below), DEC.ADM_PHD™ is deliberately skipped in favor of DEC.ADM_FDP™, starting the cycle directly with nitrogen inertization, trading some of the energy inefficiency for a more conservative safety margin.

On the condensate side, a second DEC-specific system — DEC.ADM_WPA™ — closes the loop for water-immiscible solvents: residual decanter water (containing a small amount of solvent) is routed back to the SLA duct entering the filtration room rather than to a dedicated wastewater treatment plant, with the small solvent fraction it carries re-adsorbed on the next adsorption cycle and the water vapor ultimately discharged through the stack. Combined with DEC.ADM_PHD™, this "double dehydration" approach is the basis for a claim worth stating plainly because it changes a real line item in a project's CAPEX and OPEX: for solvents handled this way, a dedicated wastewater treatment system is not required at all, in contrast to steam regeneration (§9.2), where the water fraction of the desorbate (commonly on the order of 4–5 kg of water per kg of recovered solvent) does require dedicated treatment. Heat recovery on the regeneration loop itself is handled by a further DEC-specific system, DEC.THR_x™: a dual-module arrangement providing closed-loop heat recovery on both the cooling and heating sides of the regeneration cycle (rather than only one), which — alongside DEC.ULP™ within the broader ERS™ Energy Recovery Solutions range (see §10) — is part of how nitrogen-regenerated systems reach the lower net regeneration-energy figures referenced in §1.

Solvent-specific engineering detail illustrates how these design choices play out in practice. Toluene, with a comparatively high autoignition temperature (on the order of 480°C), needs no additional safety measures beyond the standard RSG™ configuration; being water-immiscible, its condensate is decanted directly, with the water fraction handled by DEC.ADM_WPA™, and condensation is typically achieved with a brine chiller (BCS) in the region of −12°C (or lower). Hexane, by contrast, has a markedly lower autoignition temperature (roughly 225°C — see §5), which changes the regeneration design in three specific ways: the DEC.ADM_PHD™ pre-dehydration stage is skipped in favor of DEC.ADM_FDP™, beginning the cycle directly with hot-nitrogen inertization; a redundant oxygen control loop (DEC.ODP₂, on top of the standard single-loop DEC.ODP₁) is commonly added given the tighter safety margin; and condensation requires a colder duty — typically glycol water at around −20°C (or lower) rather than the brine chiller (BCS) used for toluene — reflecting hexane's higher volatility. Nitrogen heating itself can be supplied either fully electrically (heating nitrogen to roughly 190°C) or via a hybrid thermal-oil-plus-electric arrangement (oil at roughly 210°C raising the nitrogen to about 160°C, with an electric booster carrying it the rest of the way to 190°C) — the same full-electric-versus-hybrid choice discussed for e-SRU™ configurations elsewhere in this guide (see §10), applied here specifically to the nitrogen-heating duty. The decanter itself, in a nitrogen-regenerated system, is typically sized in the range of 2–3 m³ depending on plant capacity — a comparatively compact vessel that, like RSG™'s atmospheric adsorbers (§11.1), can often be fabricated locally under DEC's design supervision rather than requiring specialist international fabrication.

9.2 Steam Regeneration — TSA with Steam (e.g., DEC.SRU_RSV™)

Live steam is passed through the loaded bed, providing both heat and a stripping medium; the steam/solvent vapor mixture is then condensed, and water-immiscible solvents separate from the condensed steam. As a concrete illustration of the utility conditions involved, a typical steam-regenerated design takes pressurized steam at roughly 5–7 bar from the boiler, drops it through a pressure reducer local to the SRU, and regenerates the bed with saturated steam at a comparatively low 1.1–1.5 bar (100–120°C) fed counter-current to the SLA flow — a full regeneration cycle on the order of 40–50 minutes is typical. Because the steam supply to each adsorber is pressurized, a correctly sized rupture disk is a standard, non-optional pressure-protection fitting on every steam-regenerated adsorber vessel, distinct from the LEL/oxygen/temperature interlocks discussed in §13. Advantages: very effective, well-proven desorption for a wide range of solvents where steam supply is already available on-site (common in process industries); rapid, efficient heat transfer; because the steam itself tops out at a comparatively modest ~180°C, steam regeneration also has an inherent safety margin for solvents with a low autoignition temperature (e.g., hexane, at roughly 225°C — see §5) that is worth weighing against nitrogen regeneration's other advantages for that solvent family, rather than assuming nitrogen is always the safer default purely because it is inert. As an illustrative real-world benchmark, a large multi-adsorber steam-regenerated plant recovering a mixed MEK/toluene/cyclohexanone stream has been documented achieving roughly 4–10% carbon working capacity and consuming on the order of 5–8 kg of steam per kg of recovered solvent — a notably wider (and, on the high end, less favorable) range than the generic 3–4 tonnes-per-tonne figure cited in §10, illustrating how much actual steam demand can vary with solvent mix, cycle time and bed design, and why a vendor's quoted specific-steam-consumption figure should always be checked against the specific solvent blend rather than assumed from a generic industry average. Cooling-water duty for both the regeneration and drying-phase condensers is comparatively mild — typically only 25–32°C cooling water from a cooling tower, with no chiller required — which is itself a genuine steam-regeneration advantage over the brine- or glycol-chilled condensation duties nitrogen regeneration often needs for more volatile solvents (see §9.1).

Disadvantages: introduces water into the condensate, complicating recovery of water-miscible solvents (ethanol, IPOH, acetone) which then require downstream dehydration; not preferred for hydrolysis-sensitive solvents. For particularly volatile solvents, or where stack limits are especially stringent, an additional carbon drying stage — using heated air immediately after steam regeneration, followed by ambient-air cooling before the bed returns to adsorption service — helps ensure residual moisture in the bed does not compromise the next adsorption cycle's capacity. This drying step is a balance rather than a simple maximization: driving the bed completely dry is neither necessary nor even fully desirable, since a modest amount of residual moisture left on a freshly cooled bed is displaced by the more strongly adsorbed solvent early in the next adsorption cycle, and the heat this displacement consumes (the water's own heat of desorption) offsets some of the heat the incoming solvent's adsorption would otherwise release — a small but genuine bed-temperature-moderation effect worth factoring into drying-stage targets rather than treating "as dry as possible" as automatically the correct goal. A refinement of this cooling step routes the cooling airflow through the hot, just-regenerated bed and then discharges it via a second, actively-adsorbing bed rather than venting it directly — which both improves next-cycle adsorption performance by ensuring the bed is properly cooled before returning to service, and substantially reduces the visible water-vapor plume that direct venting after each steaming would otherwise produce, avoiding a common source of neighbor/community visual-nuisance complaints near the plant. Where steam is not already available on-site, it must be generated specifically for the SRU, which brings its own dependencies: a dedicated boiler (with associated permitting, burner fuel supply and stack emissions of its own), a resin exchanger (water softening/demineralization train) to condition boiler feedwater and prevent scale build-up and fouling in the boiler and heat-exchange surfaces, and a reliable, adequately sized source of makeup water — a real constraint in water-stressed regions or sites without existing process water infrastructure. These dependencies should be evaluated alongside the regeneration technology choice itself, since they can add meaningfully to both CAPEX (boiler package, resin exchanger skid, water treatment) and OPEX (boiler fuel, resin regeneration chemicals, water consumption and any associated effluent). Operating costs: driven by steam consumption; economical where low-cost steam is already generated on-site — but a further cost that is often underestimated at the specification stage is wastewater treatment for the aqueous (desorbate) fraction: even after decantation, the separated water phase typically retains a residual solvent load (governed by that solvent's water solubility and any azeotrope behavior, see glossary) — for a water-immiscible solvent this decanted water fraction commonly amounts to roughly 4–5 kg of water per kg of solvent recovered — so this stream generally cannot be discharged untreated and must be routed to on-site or third-party wastewater treatment — commonly an air-stripping or steam-stripping column (using ambient/recycled process air or live steam to strip residual solvent from the condensate), liquid-liquid extraction (using an immiscible extracting solvent to pull residual solvent out of the aqueous phase, an alternative worth considering particularly for high-water-content or dilute streams where stripping alone is less efficient), optionally followed by a polishing activated-carbon filtration stage for streams requiring very low residual solvent concentration — adding a recurring OPEX line that should be budgeted alongside steam consumption when comparing regeneration technologies — though this cost can be reduced at the source, rather than only downstream, by a thermocompression heat-recovery process such as DEC's ECOVAP™ (see §10), which purifies the decanted water enough for direct boiler-feed reuse as part of recovering the desorbate's condensation heat.

9.3 Vacuum Regeneration — T+VSA (e.g., DEC.SRU_RTV™)

Vacuum-assisted regeneration is fundamentally a combined temperature-and-vacuum-swing cycle (T+VSA), not a vacuum-only process: reducing pressure lowers the adsorbed solvent's effective boiling point, which helps desorption, but vacuum by itself is not sufficient to cleanly strip a loaded bed. Heat still has to be applied — without it, desorption stalls well short of completion and the bed accumulates an abnormal "heel" of residual solvent cycle after cycle, exactly the fouling/hysteresis mechanism discussed in §8.6. In practice, the bed temperature vacuum regeneration actually needs to reach is close to what the RSG™ nitrogen process requires at atmospheric pressure — the pressure reduction buys a real but comparatively modest reduction in required temperature, not a fundamentally different thermal regime — so the overall thermal energy balance of RTV™ ends up very similar to RSG™'s. This matters directly for technology selection: since RTV™ carries essentially the same heating burden as RSG™ plus the additional electrical and mechanical burden of creating and maintaining the vacuum itself (vacuum-pump power, tighter sealing, the equipment and compliance overhead detailed below), the combination puts RTV™ at a structural cost and complexity disadvantage relative to RSG™ for the wide range of solvents both technologies can handle — vacuum regeneration's case rests on the narrower set of applications (see §9.4) where its lower-pressure desorption is genuinely decisive, such as particularly high-boiling or thermally sensitive solvents (NMP, DMF, cyclohexanone), rather than on a general temperature advantage that does not really materialize in practice. It remains not ideal for other solvent families such as esters and alcohols (ROH; e.g., ethyl acetate, ethanol, isopropanol), where vacuum alone is, again, not sufficient to fully desorb the compound and higher regeneration temperatures — equivalent to a plain TSA cycle as used in the RSG™ nitrogen process — still need to be applied.

Advantages: the achievable reduction in regeneration temperature, modest as it is, still meaningfully reduces thermal degradation risk and can improve recovered-solvent purity for heat-sensitive compounds; effective for high-boiling solvents that are difficult to desorb by heat alone at atmospheric pressure. This same lower-temperature operation is also a genuine trade-off rather than a pure benefit: vacuum regeneration is often run at relatively modest temperatures (on the order of 150–180°C) specifically to save regeneration energy, but if that temperature is not sufficient to fully desorb the highest-boiling components present in a given solvent mixture, incompletely-removed residues accumulate on the carbon cycle after cycle, which can shorten effective carbon service life compared with a regeneration cycle run hot enough to fully strip the bed — so the energy saved on lower-temperature operation needs to be weighed against any resulting increase in carbon replacement frequency (see §8.6).

Disadvantages: vacuum equipment (vacuum pumps, tight vessel and valve sealing) adds mechanical complexity; because the adsorber vessels operate under vacuum rather than at atmospheric pressure like RSG™, they are pressure-retaining equipment subject to PED/ASME pressure-equipment design codes, adding engineering, documentation and fabrication cost that a plain vessel-sizing comparison would miss. This also has a practical sourcing consequence: RSG™ adsorbers, operating at atmospheric pressure, can in many cases be fabricated locally — even within the customer's own scope, close to the installation site — under DEC's design and technical supervision, whereas RTV™'s PED/ASME-certified vacuum adsorbers usually cannot be sourced locally in the same way and require fabrication at specifically certified manufacturing sites; for many project locations this means international fabrication and shipment of large pressure vessels, and the resulting freight, export packing, customs clearance and extended transit time can add substantial logistics cost and lead time on top of the fabrication premium itself. Even though condensate water content is low under vacuum regeneration, a second-stage dehydration step is mandatory to reach reuse-grade purity, adding a further process stage beyond condensation; vacuum systems are inherently more leak-sensitive than atmospheric or slightly pressurized systems, and locating small air-ingress leaks (through valve stems, flanges, instrumentation ports) can require extended, methodical troubleshooting because the symptom — gradual oxygen creep or capacity loss — is not always immediately traceable to a specific fitting. This added mechanical complexity carries a real cost beyond the base equipment price: vacuum trains generally require higher-skilled maintenance and supervision staff than atmospheric or nitrogen-loop systems, since diagnosing vacuum-integrity issues is a more specialized competency than routine valve/seal upkeep, and this can mean additional training or a dedicated service contract rather than relying solely on general plant maintenance staff. Spare-parts strategy is also more demanding: because unplanned vacuum-pump downtime stops regeneration (and therefore adsorption capacity) entirely, a spare vacuum pump held on standby is effectively mandatory rather than optional for continuous-duty installations, adding to both the initial spares inventory cost and ongoing spare-parts carrying cost compared with nitrogen or steam regeneration trains. Operating costs: driven by vacuum-pump electrical consumption plus, since RTV™ is a combined temperature-and-vacuum-swing (T+VSA) process, the same heating and nitrogen make-up costs relevant to the RSG™ process, together with higher maintenance attention to seals; certain corrosion-resistant vacuum pumps and compressors (e.g. specific bellows-sealed scroll-type designs above defined inlet-flow and pressure-ratio thresholds, with wetted surfaces in materials such as stainless steel or aluminum alloys) can fall within the EU Dual-Use Regulation (EU) 2021/821's Annex I materials-processing control entries, or equivalent control lists in other jurisdictions, and so require export-control classification and, potentially, an authorization before shipment; project teams should verify current classification with their procurement/compliance function during specification, since control list thresholds and entries are periodically amended.

9.4 Comparing the Three Regeneration Methods

As a quick-reference starting heuristic (to be confirmed against the specific solvent properties in §5, not used as a substitute for them): water-immiscible solvents (toluene, xylene, hexane) often default toward steam regeneration, since it is typically the simpler and more economical route where they're the dominant component; water-miscible solvents (ethyl acetate, ethanol, IPOH) more often push toward nitrogen regeneration, both to avoid contaminating the aqueous condensate and, for oxidation-sensitive or flammable species, for the safety reasons discussed in §9.1. Flammability and oxidation-sensitivity remain the more decisive driver per solvent, so this heuristic should be treated as a starting point for discussion rather than a final answer.

Practical takeaway: on paper, vacuum regeneration (RTV™) may look like a refined option — lower regeneration temperature, reduced thermal degradation risk, and a route to handling high-boiling solvents that are hard to desorb by heat alone. In practice, the core issue is that the temperature reduction vacuum actually buys is modest, not transformative: RTV™'s required desorption temperature and overall thermal energy balance end up close to RSG™'s (see §9.3), so RTV™ is not trading heat for vacuum so much as adding vacuum on top of essentially the same heat requirement. Once the full picture is accounted for — that near-equivalent thermal burden, plus added mechanical complexity, higher-skilled maintenance and supervision requirements, a mandatory spare vacuum pump, a mandatory second-stage dehydration step, higher leak sensitivity, vacuum-pump electricity on top of comparable heating and nitrogen make-up costs, PED/ASME-certified vessels that usually cannot be sourced locally (unlike RSG™'s atmospheric adsorbers) and so add international fabrication and shipping logistics cost, and reduced effectiveness on esters and alcohols (ROH; e.g., ethyl acetate, ethanol, isopropanol) where TSA-equivalent temperatures are needed anyway — nitrogen regeneration (RSG™) is clearly the superior choice over RTV™ for the great majority of real-world applications, with vacuum regeneration reserved for the narrow, genuinely decisive cases (particularly high-boiling or thermally sensitive solvents such as NMP, DMF or cyclohexanone) rather than treated as a generally lower-energy alternative to nitrogen regeneration.

10. Energy Efficiency

Regeneration heating is normally the largest single energy consumer in an SRU, which makes energy recovery the single highest-leverage efficiency lever available to plant designers. Heat recovery between the hot regeneration gas leaving the bed and the cold gas returning from the condenser can substantially cut net energy input, and is the design principle behind DEC's ULP™ (UltraLoop) energy recovery technology within the ERS™ Energy Recovery Solutions range: by recovering thermal energy that would otherwise be rejected at the condenser and returning it to the regeneration loop, ULP™ directly cuts the specific energy consumption of the recovery cycle, lowering both OPEX and the carbon intensity of the recovered solvent. Thermal Fluid Systems (TFS) provide a stable, closely-controlled heating medium for indirect bed heaters and are common where precise temperature control (avoiding local overheating/polymerization) matters. Electrical heating, used in full-electric (e-SRU™-type) configurations, avoids on-site combustion and can align with facilities decarbonizing their utility mix, though electricity cost per kWh should be compared against gas/steam alternatives — and pairing electrical heating with ULP™-style heat recovery further improves the economics of the full-electric route. The efficiency mechanism is worth understanding directly: electrical heating elements can deliver thermal energy straight into the regeneration air stream, on demand, whereas a Thermal Fluid System requires a secondary heat-transfer fluid to be kept circulating and at temperature so it is ready whenever heat is needed — meaning some thermal energy is spent maintaining the TFS loop even during periods when no regeneration cycle is actually calling for heat, a standing loss that direct electrical heating avoids. Heat pumps are increasingly used to recover low-grade waste heat from condenser cooling duties and upgrade it for use elsewhere in the regeneration cycle, a role that overlaps directly with ULP™'s energy-recovery function; concretely, this class of technique can include recovering condenser-side heat as reusable hot water (e.g., in the 90°C range, suitable for production processes, space heating or boiler feedwater preheating) or recovering heat from the desorbate stream itself for steam-regenerated (RSV™) systems — a falling-film condenser-evaporator, boosted by thermocompression with a steam ejector, recovers the heat of condensation from the steam/solvent desorbate mixture and uses it to directly generate steam that is added back into the primary steam supply, reducing the boiler steam otherwise required for desorption. At DEC this specific process is coded ECOVAP™: conventional steam regeneration without this heat recovery typically requires on the order of 3–4 tonnes of steam per tonne of solvent recovered, and independent industry reporting on this class of thermocompression heat-recovery process has documented reductions to roughly 2 tonnes per tonne or less — a meaningfully sharper number than a loose "half" estimate, though the achievable figure still depends on the specific solvent and plant conditions; beyond the direct steam-demand reduction, it also allows the decanted process water from regeneration to be purified enough for direct reuse as boiler feed, addressing the wastewater-treatment cost discussed in §9.2 at the source rather than downstream of it — and, being fully skid-mounted, it is suited to retrofit onto existing steam-regenerated plants with minimal shutdown, not only new-build systems. A related but distinct energy-saving technique, more commonly encountered on the downstream dehydration/distillation side (see §1.4) than in the regeneration loop itself, is mechanical vapor recompression (MVR): rather than using a steam ejector to boost the desorbate's own condensation heat (as thermocompression does), MVR uses a mechanical compressor to raise the pressure — and therefore the condensing temperature — of the column's own overhead vapor, so it can be reused as the reboiler's heating medium instead of consuming fresh steam; MVR tends to suit larger, more continuous distillation duties where the compressor's capital and electrical-power cost are justified by the resulting steam-demand reduction, and is worth evaluating alongside thermocompression specifically where a solvent-recovery installation's purification step is a substantial distillation column rather than a smaller regeneration-condensate polish. Specific energy consumption (kWh per kg of solvent recovered, or per Nm³ treated) is the standard metric for comparing SRU energy efficiency across designs and vendors, and it is the metric on which energy-recovery investments like ULP™ should be justified and tracked. As an illustrative real-world range for nitrogen-regenerated systems specifically — where energy input is split across electrical power, fuel gas for indirect thermal-fluid heating, and nitrogen make-up rather than concentrated in a single steam figure — documented plant performance data across several installations of varying scale (SLA capacity roughly 30,000 to 125,000+ Nm³/h, solvent loadings from ethyl acetate-only through mixed MEK/ethyl acetate/ethanol streams) has shown specific consumption in the region of roughly 0.9–1.3 kWh of electrical energy, 0.2–0.4 Nm³ of fuel gas, and 0.25–0.4 Nm³ of nitrogen make-up per kg of solvent recovered — figures worth requesting from a vendor, broken out by utility, as a like-for-like comparison point against a proposed design's guaranteed consumption, rather than accepting a single blended "energy cost" figure that obscures which utility actually drives OPEX. On multi-adsorber plants, some vendors pursue a different route to the same energy-efficiency goal: overlapping or tightly staggering successive beds' regeneration cycles, rather than recovering and reusing heat as ULP™ does. This approach can smooth utility demand across the cycle, but it comes with its own trade-offs that are easy to understate: closely overlapping regeneration windows generally requires larger, more complex regeneration-loop hardware and control logic (to manage two or more beds desorbing at once, rather than one at a time), and — because it targets cycle-timing efficiency rather than actually recovering and reusing thermal energy — it does not reduce the underlying regeneration heat demand the way heat recovery does, so it can end up more energy-intensive overall for a comparable duty rather than less. When evaluating a revamp of an aging, timer-sequenced plant (see §11.3), it is worth asking a vendor proposing cycle-overlap optimization specifically how it compares, on net specific energy consumption, to a heat-recovery retrofit instead.

11. Economics: CAPEX, OPEX, ROI & Payback

11.1 CAPEX

Capital cost is driven primarily by airflow (vessel and fan sizing), regeneration technology (vacuum and steam infrastructure carry different cost profiles than nitrogen skids), materials of construction (corrosion allowance), and whether the plant is delivered as a pre-built modular skid (typically lower installation cost and faster deployment) or a fully custom-built system (higher engineering content, suited to constrained layouts or unusual process requirements). Regeneration technology choice also affects which pressure-equipment design code (PED/ASME) applies: RSG™ vessels operate at essentially atmospheric pressure (the loop's slight nitrogen overpressure, see §9.1, keeps them below typical PED/ASME thresholds and therefore generally exempt), while RTV™ vessels, operating under vacuum, are pressure-retaining equipment subject to PED/ASME design, fabrication and inspection requirements — adding engineering, documentation and fabrication cost that a simple vessel/fan sizing comparison would miss. This also affects sourcing: atmospheric RSG™ adsorbers can, in many cases, be fabricated locally — even within the customer's own scope, close to the installation site — under DEC's design and technical supervision, while RTV™'s PED/ASME-certified vacuum adsorbers usually cannot be sourced locally in the same way and must be built at specifically certified manufacturing sites, which for many project locations means international fabrication and shipment of large pressure vessels: freight, export packing, customs clearance and extended transit time can add substantial logistics cost and lead time on top of the fabrication premium itself, a factor worth weighing alongside the regeneration technology's process advantages.

11.2 OPEX

Ongoing costs include regeneration utility consumption (nitrogen, steam, or electricity), wastewater treatment for the solvent-laden condensate fraction on steam-regenerated systems, electrical power for fans and, where applicable, vacuum pumps, routine maintenance labor, and periodic activated carbon replacement. Well-designed SRUs are engineered to minimize OPEX per unit of solvent recovered, since OPEX directly erodes the economic benefit of solvent recovery.

11.3 Maintenance & Carbon Replacement Costs

Maintenance costs include routine inspection, valve and seal servicing (particularly for vacuum systems), instrumentation calibration, and periodic carbon replacement when breakthrough performance degrades below the design specification (see §8.6). Vacuum-regenerated systems carry a further consideration: they generally warrant higher-skilled maintenance and supervision staff than atmospheric or nitrogen-loop designs, and continuous-duty installations typically need a spare vacuum pump held on standby, adding to spares inventory cost. Structured adsorbent maintenance services (such as those delivered through DEC SERVICE) help ensure carbon life and mechanical reliability are tracked proactively rather than reactively.

11.4 Solvent Savings, ROI & Payback

The dominant economic variable is almost always the value of recovered solvent: yearly solvent throughput multiplied by recovery efficiency and current solvent purchase price. For high-throughput, high-value-solvent operations, payback periods can be short because avoided purchase cost alone offsets CAPEX; for lower-throughput or lower-value-solvent operations, the investment case rests more on compliance necessity and long-term OPEX savings versus destructive alternatives. A rigorous ROI evaluation should include CAPEX, installation, OPEX, carbon replacement reserve, avoided solvent purchase, and — where relevant — avoided hazardous-waste disposal cost, compared over the expected plant service life.

As a rule-of-thumb lower bound rather than a precise threshold, activated-carbon solvent recovery tends to remain economically viable down to fairly dilute inlet loadings — on the order of ~0.3 g solvent per Nm³ of exhaust air, depending on the specific solvent's value and the plant's complexity — below which the value of recovered solvent may no longer justify a dedicated regenerative SRU relative to a simpler or destructive alternative (see the decision tree in §3.1). Within that viable range, payback periods as short as roughly one year are achievable at the higher end of solvent concentration and value, though the site-specific calculation described above should always be run rather than relying on a single illustrative figure. As a real-world illustration at the larger end of the scale, a documented multi-decade industrial solvent-recovery installation recovering roughly 24 million lbs/year of a mixed MEK/toluene/cyclohexanone solvent blend, with a total installed and later MACT-upgraded capital cost in the low tens of millions of dollars and annual operating cost in the low single-digit millions, generated recovered-solvent value substantially above its annual operating cost — illustrating, at scale, the same value-of-recovered-solvent-versus-OPEX dynamic described above, and consistent with the multi-year (rather than sub-one-year) payback profile typical of large, capital-intensive installations even where the underlying operating economics are strongly favorable. This economic profile is not a new or unproven one: fixed-bed activated-carbon solvent recovery has been in continuous industrial use since the 1920s, and the installed base worldwide today numbers in the thousands of plants across rotogravure and flexographic printing, coating, fiber and film production, and adhesive-tape manufacturing — a long operating history that gives buyers a substantial base of real-world performance and reliability data to evaluate a proposed SRU against, beyond vendor-quoted design figures alone.

12. Environmental Compliance: BAT, BREF, IED & EPA

12.1 BAT & BREF (EU)

Under the EU framework, Best Available Techniques (BAT) are the reference techniques regulators expect installations to apply, and BAT Reference Documents (BREF) — including the Surface Treatment using Organic Solvents (STS) BREF, which most directly covers VOC recovery and abatement for printing, coating and adhesive-application processes of the kind this guide addresses, the Common Waste Water and Waste Gas Treatment/Management Systems (CWW) BREF, and other sector-specific BREFs — document the associated emission levels (BAT-AELs) that permitted installations must demonstrate. Activated carbon adsorption is explicitly recognized within BAT/BREF documentation as a technique for VOC recovery and abatement, alongside thermal and catalytic oxidation and other techniques. EU member states implement these requirements through national permitting frameworks — for example, Italy's Autorizzazione Integrata Ambientale (AIA), the national implementation of the EU's Integrated Environmental Authorization (IEA) regime — so the specific permit conditions an SRU must satisfy are set at the national/regional level even though the underlying BAT/BREF reference framework is EU-wide.

12.2 Industrial Emissions Directive (IED)

The IED is the EU's overarching legal framework requiring installations within its scope to operate under permits that set emission limit values informed by BAT conclusions; VOC/TVOC stack limits (commonly expressed as mgC/Nm³) are set at a permit level and are a primary driver of SRU emission-performance specifications, including whether supplementary polishing technology — sometimes referred to industry-wide as a "guard bed" or finishing adsorption stage (e.g., DEC.ULE™) — is needed to reach very low TVOC targets.

12.3 US EPA Guidance

In the United States, VOC emissions are regulated under the Clean Air Act framework, with facility-specific limits set through New Source Performance Standards (NSPS), National Emission Standards for Hazardous Air Pollutants (NESHAP) — which for many organic-HAP source categories specify a Maximum Achievable Control Technology (MACT) emission standard — and state/local permitting under the US Clean Air Act (CAA); EPA technical guidance historically published through resources such as the P2 InfoHouse documents carbon adsorption as a recognized Reasonably Available Control Technology (RACT) / Best Available Control Technology (BACT) / MACT-compliant option for solvent-laden exhaust streams.

12.4 Sustainability, ESG & Circular Economy Reporting

Because solvent recovery reduces both emissions and purchased-material consumption, SRU performance data (recovery rate, tonnes of solvent recovered per year, avoided emissions) is directly usable in ESG and sustainability reporting, and is increasingly requested by customers and investors as evidence of decarbonization action at the process level.

13. Safety Engineering

13.1 ATEX & Explosion Prevention

SRUs handling flammable solvents are designed to ATEX (or equivalent) area-classification and explosion-proof equipment principles: zoning of hazardous areas, selection of certified electrical and instrumentation equipment for the classified zones, and process design that keeps the regeneration atmosphere outside the flammable envelope by controlling VOC concentration and oxygen level rather than relying on ignition-source elimination alone. Classified-zone equipment typically extends beyond the adsorber vessels themselves to the condenser and raw-recovery circuit, crude/raw-solvent storage tanks, distillation columns, and any high-boiling-residue or azeotrope tanks — each of these should be explicitly identified and zoned in the area-classification study, not assumed to be covered implicitly by classifying the main adsorption/regeneration equipment. Static-electricity control is a further, often underestimated ignition-source risk in flammable-liquid handling: equipotential bonding and grounding of tanks, piping and mobile equipment (to prevent static discharge as an ignition source during liquid transfer or agitation) and lightning protection for outdoor tanks and structures are standard requirements alongside the electrical area classification itself, and should be verified as part of the same hazardous-area engineering package rather than treated as a separate civil/electrical afterthought. Beyond equipment-level area classification, ignitable-liquid engineering practice (see, e.g., FM Global Data Sheet 7-2) also classifies rooms and buildings themselves against a distinct room explosion hazard criterion, applicable to any space handling solvent-laden air or crude/raw solvent generally — not a carbon-bed-specific test — under which a room/building explosion hazard is considered present where the liquid is processed at or above its atmospheric boiling point with a closed-cup flash point at or below roughly 425°F (218°C), or where the liquid's own boiling point is at or below roughly 100°F (38°C). Where this criterion is met, damage-limiting construction and containment/emergency-drainage provisions for the room or building are evaluated in addition to, not instead of, the equipment-level ATEX/hazardous-area measures described above.

13.2 Oxygen Monitoring & Nitrogen Safety

Continuous oxygen monitoring in closed-loop nitrogen systems is a primary safety control, since oxygen ingress — through leaking valves, flanges, seals or instrumentation connections — can allow the regeneration atmosphere to drift toward flammable conditions as VOC concentration rises during desorption. Because inert-loop leaks are not always immediately obvious (a slow oxygen creep can persist for some time before an alarm threshold is reached), plants require a defined leak-detection and isolation procedure, and engineering teams should budget for the possibility that locating a specific leak point in a closed-loop system can require systematic, section-by-section isolation testing rather than a single quick check. Nitrogen systems also require standard inert-gas safety measures (confined-space and asphyxiation-risk controls) around vessels and enclosures where nitrogen could displace breathable air.

The oxygen alarm/interlock setpoint itself is not an arbitrary number: it is referenced to the Maximum Oxygen Content (MOC) — also called Maximum Oxygen Concentration, and closely related to the Limiting Oxygen Concentration (LOC) terminology used in the flammability literature — for the specific solvent(s) and inert gas in service, defined as the highest oxygen concentration (vol.%) at which combustion cannot be sustained regardless of fuel vapor concentration, under the specific pressure, temperature and inert-gas composition of the process. MOC is solvent-specific, and design margins are normally set well below it rather than at it (NFPA 69 specifies a minimum safety margin, commonly 2 vol.% below the LOC where the LOC is ≥5 vol.%). Published MOC/LOC values are conventionally tabulated near ambient temperature — typically a 20–25°C reference (denoted MOC₂₅ below) — which is the convention used in the table that follows, distinct from the 0°C Normal Conditions reference used elsewhere in this guide for gas volumes (Nm³).

Because actual regeneration temperatures run well above this ambient reference (illustratively up to roughly 250°C at the top end of the RSG™/RTV™ range), MOC₂₅ cannot be used directly as the operating interlock setpoint — MOC decreases as temperature rises. The engineering approach to estimating this decrease, based on the Limiting Oxygen Concentration theory originally developed by Zabetakis (U.S. Bureau of Mines) and later refined by Coward & Jones and reflected in NFPA 69, follows four steps: (1) obtain the MOC/LOC at the reference temperature, typically 20–25°C; (2) calculate the adiabatic flame temperature and associated heat losses; (3) correct the flammability limit for initial gas temperature, oxygen concentration, inert-gas heat capacity and pressure; and (4) validate experimentally wherever the target temperature exceeds the range covered by published data. For hydrocarbon and oxygenated solvents in nitrogen, engineering practice commonly applies a simplified linear approximation:

MOC(T) = MOC₂₅ × [1 − k(T − 25)]

where T is temperature in °C and k ≈ (1.0–2.0)×10⁻⁴ °C⁻¹ for nitrogen inerting — valid only as an engineering approximation, and, per step 4 above, not a substitute for experimental validation at temperatures well outside the range the correlation was fitted against. The reference (MOC₂₅) and illustrative 250°C estimates below follow this convention:

The 250°C column is calculated solely by applying the linear approximation formula above (mid-range k = 1.5×10⁻⁴ °C⁻¹ bounding the k = 1.0–2.0×10⁻⁴ °C⁻¹ range) to the MOC₂₅ figures, as a worked illustration of the method — it is not experimentally validated test data. A linear extrapolation across a 225°C span from the MOC₂₅ reference is well beyond the range most published low-temperature correlations were fitted against, which is precisely the condition step 4 of the methodology above flags for mandatory experimental validation. These figures must not be used as an actual safety interlock setpoint without such validation, or without solvent-specific test data (e.g., per ASTM E2079) at the real process temperature and pressure.

MOC is the third leg of a correlated safety check alongside oxygen and LEL monitoring, not an independent, standalone limit: the actual VOC concentration in the loop (tracked against LEL, see §13.3), the oxygen concentration (tracked against MOC), and process temperature all interact, since MOC itself shifts with temperature and pressure and is not a single fixed number across all operating conditions. Interlock logic and alarm setpoints should therefore be engineered around the specific solvent's MOC with an appropriate safety margin, verified against actual process temperature and pressure, rather than a single generic oxygen percentage applied uniformly across different solvents or operating conditions.

13.3 LEL Monitoring

Lower Explosive Limit (LEL) sensors — like the EMS™ stack analyzers discussed in §14, part of DEC's broader ATX™ analytical systems portfolio — are used at key points in the process (regeneration loop, condenser off-gas, enclosures) to provide continuous verification that VOC concentration remains within the safe operating envelope, feeding automated interlocks that can dilute, isolate or shut down the affected section if the LEL setpoint is approached.

13.4 Fire Protection & Activated Carbon Ignition

Activated carbon can support smoldering combustion under certain conditions (notably with strongly oxidizing adsorbates or hot-spot formation from uneven regeneration), so temperature monitoring within the bed, controlled regeneration temperature profiles as part of a broader fire-prevention strategy, and dedicated fire-detection/suppression provisions appropriate to the specific solvent chemistry are standard design elements. Hot spots are not only a regeneration-control issue: exothermic carbon-catalyzed side reactions among adsorbed species — such as the oxidative degradation reactions ketones like MEK can undergo on the carbon surface (see §8.6) — release heat directly within the bed itself, and this solvent-chemistry-dependent heat source should be accounted for in bed temperature monitoring and interlock setpoints alongside the more commonly cited regeneration-profile and oxidizing-adsorbate risks. This is not a theoretical concern: documented multi-decade industrial operating experience with mixed MEK/toluene/cyclohexanone recovery has specifically flagged diacetyl and adipic acid formation as a recognized "reactive chemistry" risk in this class of service, tracked in practice via continuous low-level carbon monoxide monitoring (on the order of single-digit ppm) as an early warning of oxidative activity in the bed well before a temperature excursion would otherwise be detected — consistent with, and a real-world confirmation of, the CO/CO₂ trend-monitoring guidance discussed above. The same operating history includes a documented incident combining a carbon-bed fire with adsorber vessel implosion, a failure mode worth naming explicitly because it is easy to overlook: a vessel that has just been subjected to fire, steam, or rapid internal cooling (including emergency water/steam knockdown) can develop a significant internal vacuum as trapped vapor condenses faster than air or inert gas can safely re-enter, and a carbon vessel's shell and internals are not necessarily rated for that reverse (external-over-internal) pressure differential the way they are rated for normal operating overpressure — so post-incident vessel-integrity procedures should explicitly address vacuum relief and controlled re-pressurization, not just fire suppression and cool-down, and this vacuum-relief adequacy is worth confirming with the vessel designer at the specification stage rather than only after an incident.

A point worth making explicit on detection method: a point temperature probe inside a large carbon bed only reports conditions at its own location, and an outlet-stack temperature sensor is inherently lagging — a hotspot's thermal signature has to propagate through, and get diluted by, the full moving SLA volume before it reaches the stack, so temperature-only detection is structurally slow for large-diameter, deep beds. This is the underlying reason gas-composition trend detection (CO/CO₂, discussed above) is the more effective early-warning layer, and CO₂ specifically deserves more weight than it often gets: smoldering carbon oxidation typically yields substantially more CO₂ than CO by mass, and — just as importantly — ambient/process CO₂ background is normally far more stable than CO background (which is more prone to interference from combustion-related trace sources elsewhere in a facility), so a given absolute rise in CO₂ is both larger in magnitude and easier to resolve statistically against a quiet baseline than the same event's CO signature. A CO-only detection strategy therefore risks missing or delaying detection of exactly the low-and-slow oxidation events most likely to precede a hotspot; pairing CO with CO₂ trend detection, referenced to each gas's own site-specific background (as already discussed above), is the more robust design.

The suppression medium itself is a design choice with real downstream consequences, not just an equipment selection. Water introduced into a hot carbon bed does not simply cool it: at the temperatures a genuine hotspot can reach, water can also drive the classic water–gas reaction with incandescent carbon (C + H₂O → CO + H₂, an endothermic reaction that nonetheless generates additional flammable and toxic gas, on top of whatever CO/CO₂ the oxidation itself was already producing, and consumes some of the water's cooling benefit in the process) — a mechanism worth naming specifically rather than describing only as generic "steam generation," since it is the reason a hot-carbon/water encounter is a chemical hazard and not only a thermal or structural one. This is also why overbed — never underbed — deluge geometry (discussed above, per FM Global Data Sheet 7-2) matters as much as the choice of medium itself: underbed water injection risks floating retained solvent upward into the vessel's vapor space exactly when combustible-gas generation from the water–gas reaction is also elevated. A well-engineered water deluge system also needs a correctly sized bed drain (commonly sized well above a bare minimum to tolerate loose carbon fines washing out and partially obstructing the line, with flow-velocity and pressure-drop monitoring to catch progressive blockage before it defeats the drain's purpose) and pressure relief sized to a credible deflagration basis (per NFPA 68) rather than to simple thermal gas expansion alone — because an isolated vessel containing desorbed solvent vapor, entrained fine carbon dust, and any hydrogen generated by the water–gas reaction is a credible dust/vapor deflagration scenario even where normal operating conditions are not expected to reach it, and relief sizing is a defense-in-depth measure precisely for the case where the other safeguards do not all hold. Structurally, deluge engagement is also a real transient loading event, not just a pressure-protection one: activated carbon retains a substantial fraction of its pore volume as water once wetted, and a multi-minute discharge adds further free water plus spray-impact force on top of the bed's normal dry weight — vessel supports and foundations sized only for normal operating weight can be meaningfully undersized for this wetted condition, and this transient wet-load case should be an explicit structural design load, not an afterthought. In cold climates, or for elevated platform/rooftop installations exposed to freezing, wet-pipe deluge piping carries its own freeze risk; a dry, pre-action system with nitrogen-pressurized piping (per NFPA 750) is the standard mitigation where freezing is a credible risk.

Nitrogen injection is a materially different suppression mechanism, worth evaluating directly against water deluge rather than treating as a simple substitute: rather than cooling the hotspot, it works by diluting the sealed vessel's atmosphere below the oxygen concentration needed to sustain combustion (the same Maximum Oxygen Content principle discussed in §13.2), typically injected through bottom-located nozzles into an isolated vessel with the displaced atmosphere vented via the same flue-gas stack path used in normal operation. Because it acts by inerting rather than quenching, its effectiveness depends on delivering enough nitrogen, quickly enough, to bring the full free-gas volume of the isolated vessel below the target oxygen threshold — a flow-rate and response-time sizing exercise specific to the vessel's free-gas volume (total vessel volume less carbon volume, adjusted for carbon pore volume) that should be verified for the specific adsorber geometry rather than assumed from a generic nitrogen supply rating. Its comparative advantage is what it avoids rather than what it actively does: no water–gas reaction chemistry, no wetted-bed structural loading, no freeze-protection burden, and — critically for return-to-service time — no chemically altered, waterlogged carbon bed afterward. Post-incident recovery from a water-deluge event is genuinely costly in its own right: the wetted, chemically altered carbon typically requires drying, extraction, sieving and partial-to-substantial replacement (illustratively on the order of 10–30%+ of the full charge, depending on fire severity) before the adsorber can be requalified and returned to service, and in jurisdictions with strict environmental permitting this SRU downtime can halt the entire upstream solvent-handling process until requalification is complete — a real financial and compliance cost, on top of the direct carbon-replacement expense, that is easy to underweight when comparing suppression options purely on installed cost. Where a reliable nitrogen supply (or on-site generation, see N2G™) is available, this combination of avoided chemical hazard, avoided structural/freeze burden, and materially faster, less destructive post-incident recovery generally makes nitrogen injection the preferable default suppression medium for gas-phase activated-carbon adsorbers specifically — water deluge remains a valid and, in some jurisdictions or insurer frameworks, a specifically referenced or mandated approach, and should be engineered to the overbed-geometry, drainage, structural and freeze-protection standards above wherever it is the medium of record.

Bed overheating shall also be monitored via carbon monoxide (CO) and carbon dioxide (CO₂) trend detection in addition to temperature sensing: because background CO and CO₂ levels are installation-specific, alarm, safety inertization and/or automatic-shutdown setpoints are referenced to that site's own established background level (a percentage above it) rather than to a single fixed ppm figure applied uniformly across different plants.

13.5 Vacuum System Integrity & Operational Reliability

Vacuum-regeneration systems (§9.3) are inherently more sensitive to small air in-leakage than atmospheric or pressurized designs, because even a minor ingress path represents a proportionally larger disturbance to a low-pressure system; robust vacuum integrity testing at commissioning and during scheduled maintenance, together with a clear leak-isolation methodology, materially reduces unplanned downtime. As noted in §9.3, some specialized vacuum-system components can also fall under export-control or dual-use regulatory regimes in certain countries, and procurement teams should confirm classification and any licensing requirements early in project planning to avoid delivery delays.

13.6 Emergency Shutdown & Contamination Risk

SRUs incorporate emergency shutdown (ESD) logic tied to LEL, oxygen, temperature and pressure interlocks, with defined safe-state sequences (e.g., nitrogen purge, isolation valve closure) to bring the system to a safe condition without operator intervention. Contamination risk — cross-contamination between solvent batches in shared equipment, or moisture ingress degrading recovered-solvent quality — is managed through dedicated piping/valve segregation where multiple solvent types are processed and through condensate quality monitoring.

Idle or standby periods — overnight, over a weekend, or during an unplanned production stop — deserve their own procedure, since a loaded carbon bed left unattended without adequate airflow is a recognized cause of spontaneous-heating incidents (see §13.4). Property-risk-engineering guidance (FM Global Data Sheet 7-2) accepts any of four measures to secure a bed during such periods: maintain at least 75% of normal adsorption-mode airflow through the bed, physically remove the carbon from the vessel, keep the bed wet through periodic water or steam spraying, or inert the vessel with nitrogen or CO2 so oxygen content — including in the void spaces within the packed bed, not just the open headspace — does not exceed 1% by volume. A defined idle-period procedure, selecting one of these four options explicitly rather than simply leaving the last adsorption or regeneration state in place, should be part of the same ESD/shutdown documentation referenced above.

13.7 Plant Layout & Utility Placement

Physical layout is a safety decision, not just a space-planning one. A steam or thermal-fluid boiler (see §9.2) is itself a fire and explosion hazard, so its position within the overall plant footprint should limit the consequences of a boiler-related incident rather than maximize piping convenience: locating the boiler centrally, surrounded by adsorber vessels, ductwork and process piping, means a fire or explosion originating at the boiler can affect the entire SRU rather than being contained to one edge of the installation. Layout reviews (often part of a HAZOP or equivalent process hazard analysis) should specifically evaluate utility and ignition-source placement — boiler, electrical switchgear, vacuum pumps — relative to the rest of the plant footprint, not only relative to piping runs and civil works cost.

13.8 Carbon Handling & Storage Safety

Activated carbon itself presents handling and storage hazards distinct from the in-process VOC/LEL/MOC risks discussed above, and these apply during carbon loading, unloading, replacement and reactivation logistics rather than during normal SRU operation. Loose activated carbon dust is combustible and can present a dust-explosion hazard at sufficiently high airborne concentrations, so bulk handling operations (loading, screening, transfer) warrant the same grounding/bonding, ignition-source control and dust-collection discipline applied to other combustible-dust operations. Solvent-laden spent carbon removed from an adsorber is a separate concern: it can self-heat when exposed to open air, so it should not be piled or stored loosely in bulk without a defined cooling/handling procedure, particularly for higher-boiling or oxidation-prone solvents. Finally, both fresh and spent activated carbon can adsorb oxygen from the surrounding air in an enclosed space (vessel interior, storage container, shipping container), so confined-space entry procedures and atmosphere testing before entry apply to carbon handling generally, not only to the closed-loop nitrogen circuit discussed in §13.2.

14. Digitalization & Industry 4.0

Modern SRUs are controlled through PLC logic integrated with plant-level SCADA systems, providing real-time visibility of bed status, temperatures, pressures, oxygen/LEL readings and cycle timing. A concrete example of this in practice is a continuous EMS™ (Emission Monitoring System) at the stack — built on DEC's underlying ATX™ analytical systems platform (Flame Ionization Detectors, Flame Thermal Analyzers or FTIR, depending on the application) — reporting real-time TOC (Total Organic Carbon) results as TVOC: rather than switching beds to regeneration on a fixed timer, EMS™ can trigger regeneration only once the outlet concentration actually approaches the permitted limit (see §8.4), so utility consumption tracks the true incoming solvent load rather than a conservative worst-case schedule — directly reducing regeneration frequency, and therefore energy and nitrogen/steam consumption, during periods of lower-than-design loading. A similar adaptive-control principle applies to airflow: variable-frequency-drive (VFD/inverter) control on the main SLA fans, referenced against an upstream pressure or flow setpoint, allows the fan to draw only the power the actual process demands rather than running continuously at the fixed design flow rate, which is directly relevant wherever the process operates below its rated turn-down (see §7.1). A related, complementary technique in multi-line facilities is linking SRU intake directly to production-line status: automatically isolating or bypassing the ductwork and fan capacity associated with a coating or printing line that is not actively running, rather than continuously processing SLA from idle lines, avoids unnecessary fan power consumption and prevents unneeded dilution of the SLA stream that would otherwise reduce adsorption performance for the lines that are running. Industry 4.0 connectivity enables predictive maintenance — using trending of pressure drop, breakthrough timing and regeneration performance to flag carbon aging or developing mechanical issues before they cause an unplanned stop — and remote diagnostics, allowing vendor engineering teams to support troubleshooting without an immediate site visit. Energy monitoring and performance analytics dashboards track specific energy consumption and recovery efficiency over time, supporting both OPEX management and sustainability reporting. Emerging AI-assisted optimization approaches use historical cycle data to fine-tune regeneration timing and temperature profiles for the specific solvent mix and carbon condition at a given site, improving both energy efficiency and carbon service life.

15. DEC.SRU™ Product Range: SMS™ & CBS™

DEC Solvent Recovery Units (SRUs) are available in both skid modular layout (DEC-SMS™ • Smart Modular Systems, pre-built and pre-tested at our workshops) or in full customizable version (DEC-CBS™ • Custom Built Systems, with on-site erection).

15.1 Redundancy in Modular Skid Architecture

Not all "modular" SRU architectures provide the same level of redundancy, and it is worth understanding the distinction when comparing modular offerings across the market. In an independent-skid architecture, each modular adsorber unit carries its own complete regeneration loop — its own carbon bed(s), regeneration fan, and condenser — so a fault in one skid's regeneration equipment, control system, or a single component (a valve, a limit switch, a vacuum pump) affects only that skid's capacity, while the remaining skids continue operating normally. In a shared-loop architecture, by contrast, multiple adsorber skids share a common regeneration loop and control system; this can reduce equipment count, but it also means a single fault anywhere in that shared regeneration loop or its control system can take the entire multi-skid installation out of service at once, rather than just the affected unit. When evaluating modular SRU proposals, it is worth asking explicitly whether the regeneration loop, fan and condenser are dedicated per skid or shared across skids, since that answer — more than skid count alone — determines how much real redundancy the "modular" configuration actually provides during maintenance or an unplanned fault.

Equipment-level redundancy is only half of the picture: physical site layout matters just as much for a true backup capability. Property-risk-engineering guidance for activated carbon solvent recovery — specifically AXA XL Risk Consulting's PRC.9.6.2.2 — recommends that duplicated adsorption trains be separated by at least 50 ft (15 m), or alternatively by a firewall rated to withstand at least 3 psi (0.2 bar) of overpressure, so that a fire or explosion event involving one train cannot also disable the unit meant to back it up. This is consistent with the same guideline's broader loss-prevention framing for carbon bed adsorption: because a large adsorption system can represent a multi-million-dollar replacement cost with a lead time measured in months, the business-continuity case for physical separation — not only equipment duplication — is a legitimate input to site-layout decisions, alongside the civil-works cost of spreading equipment further apart. Sites evaluating a multi-train SRU installation should confirm this separation requirement (or an equivalent fire-rated barrier) with their own insurer or risk engineer during layout planning, since specific thresholds can vary by policy and jurisdiction.

15.2 Engineering Credentials & Quality Certifications

The engineering content on this page reflects applied experience rather than only theory: DEC's activated-carbon solvent recovery technology lineage traces back to the 1920s [CECA, Carbonisation Et Charbons Actifs, now part of the DEC group], with DEC itself operating since 1946 and now in its third generation of family ownership and management, and design/engineering work is carried out by an in-house, multidisciplinary team of chemical, mechanical, electrical, environmental and civil engineers, supported globally across five continents through a distributed structure of subsidiaries, partnerships, joint ventures, licensees and agencies covering more than 50 countries. Engineering and manufacturing operate under an integrated management system in conformity withISO 9001 (Quality Management), ISO 14001 (Environmental Management) and ISO 45001 (Occupational Health & Safety), hazardous-area equipment follows ATEX/IECEx certification requirements (see §13.1), and pressure-retaining vacuum equipment (RTV™) is designed and manufactured to PED/ASME pressure-equipment codes — atmospheric-pressure RSG™ vessels are generally exempt from this scope. Sector coverage spans the industries detailed in §6 above — including flexible packaging, converting, coating, chemical, petrochemical and pharmaceutical manufacturing — consistent with DEC's broader industries & markets footprint. As of this writing, DEC reports more than 2,500 projects delivered across its installed base; specific installation references by sector are available on request via DEC's references page.

If you are looking for a reliable and efficient Solvent Recovery Unit, DEC.SRUs™ (Solvent Recovery Units) are the right and best option: DEC.SRUs™ are the most advanced and reliable Solvent Recovery Units on the market, with processes, configurations and options to meet the needs of different applications, present and future challenges. Feel free to contact DEC: we can help you assess your needs and recommend the best SRU for your operation.