Vacuum Infusion vs RTM: Pressure, Fiber Content, Cycle Time and Cost

1. Vacuum infusion vs RTM: which process should you use?

Both processes start from dry reinforcement and finish with a liquid resin that wets it in the mold, which is why AVK counts infusion inside its RTM segment [S42]. The difference is the boundary. In vacuum infusion the dry stack sits on a one-sided rigid mold under a sealed bag; resin and fiber compaction are both driven by pressures of 101.3 kPa or less [S8]. In RTM the preform is closed in a matched tool and resin is pumped in: about 10 to 20 bar for standard low-pressure RTM, up to 150 bar in the mixing head and 30 to 120 bar inside the mold for high-pressure RTM [S10]. Light RTM sits between them, with a semi-rigid counter-mold clamped by flange vacuum and injection held below 1 bar [S9].

That boundary sets everything a buyer cares about. Infusion makes one tool face, accepts enormous parts on cheap molds and is slow: the main molding cycle for a 100-meter-class offshore blade at Sinoma was cut to 48 hours in 2024 [S23]. RTM gives two tool faces and thickness set by the cavity, with standard cycles of 30 to 60 minutes [S10]; HP-RTM brings that to a realistic 3 to 7 minutes and has moved annual volumes from 1,000 to 4,000 parts a year two decades ago to 12,000 to 50,000 now [S10]. As a rule of thumb, infuse anything longer than a few meters or made in tens per year, use light RTM or RTM for hundreds to a few thousand parts with two good faces, and reserve HP-RTM for automotive volumes with a press budget to match.

2. How resin flows: Darcy's law and infusion fill time

Resin flow through a fiber bed follows Darcy's law: the flow rate rises with permeability and pressure gradient and falls with viscosity [S1]. For a one-dimensional front, integrating it gives a fill time t = φμL² / (2KΔP), where φ is porosity, or one minus fiber volume fraction, μ is viscosity, L is flow length, K is permeability and ΔP is pressure difference; the permeability benchmarks measure K exactly this way, by fitting the square of front position against time [S1]. The square on L is the key: doubling the flow length quadruples the fill time, while doubling the pressure only halves it.

Realistic inputs are now published. An international benchmark measured a carbon twill at about 1.4 × 10⁻¹⁰ m² in the principal direction at 45% fiber volume, with ±20% scatter between laboratories [S1]. For a 295 g/m² E-glass twill, benchmark-based values fall from 0.666 × 10⁻¹⁰ m² at 46.2% fiber volume to 0.434 × 10⁻¹⁰ at 50.3% and 0.285 × 10⁻¹⁰ at 53.4%, with transverse permeability five to six times lower [S2]. A DNV GL-approved infusion epoxy, SikaBiresin CR83, mixes at 155 to 170 mPa·s depending on hardener, with pot lives from 60 to 300 minutes [S3].

Putting those numbers into the formula shows why infusion relies on flow media. With 50.3% fiber volume glass, K = 0.434 × 10⁻¹⁰ m² and μ = 0.155 Pa·s, resin pushed through the laminate itself by a full atmosphere would take about 36 minutes to travel 0.5 m, 2.4 hours to travel 1 m and 9.7 hours to travel 2 m [estimate: Figure 2 inputs]. The fast hardener's 60-minute pot life is gone before the 1 m mark. Infusion beats this by spreading resin across the top in a high-permeability mesh or grooved core, the principle of the SCRIMP patent filed in 1989 [S14], so that the critical flow path becomes the laminate thickness rather than the part length. RTM beats it with pressure: at 10 bar the same 1 m path fills in about 15 minutes, and at 20 bar in about 7.4 minutes [estimate].

Two practical consequences follow. First, a larger vacuum pump cannot raise ΔP above one atmosphere; only shorter flow paths, more permeable media, lower viscosity or warmer resin speed up infusion. Second, permeability changes with compaction and with how fabric drapes into corners, so flat-panel fill trials do not prove a shaped part. Race-tracking along edges, where permeability is locally much higher, is the classic way a gate plan that works on paper leaves dry spots in practice [S33].

3. Fiber volume fraction and void content by process

Pressure is what consolidates a laminate, so infusion sits below closed-mold and pultruded parts on fiber content. In a peer-reviewed ORNL and Sandia cost study, infused glass spar-cap laminate runs at a fiber volume fraction of 0.57, or 73% by mass, against 0.68 for a pultruded glass plank, and pultrusion gives a 17% higher design strength at constant fiber volume [S12]. An IACMI and ORNL analysis put carbon spar caps at 50% fiber volume by vacuum-assisted RTM against 58% pultruded [S13]. That gap is why most new blade spar caps moved to pultruded planks while the shells stayed infused.

Infusion is still a high-quality process when it is controlled. NASA's work on high-temperature VARTM reports epoxy laminates approaching 60% fiber volume and void contents below 3% on biaxial carbon and below 2% on uniweave, against the less-than-2% voids desired for aerospace, measured by acid digestion to ASTM D3171 and micrographs [S8]. The original SCRIMP patent reports panels averaging 26.1% resin by weight [S14]. For HP-RTM, fiber weights up to 75% have been claimed, while structural reaction injection molding typically stays at up to 30% by weight [S10]; light RTM parts commonly carry about 30% glass by weight [S9].

Voids are the number to put in the contract. In wind-blade laminates, each 1% increase in interlaminar voids is reported to reduce mechanical properties by about 7% [S33]. A 2024 DTU study of vacuum infusion at inlet pressures of 283, 533 and 1,013 mbar found the slowest, lowest-pressure infusions produced the fewest voids [S34], a reminder that filling fast is not the same as filling well.

4. Cycle time and production rate

Cycle time spans three orders of magnitude across this family. At the slow end, Sinoma Blade cut the main molding cycle of its SI122 100-meter-class offshore blade to 48 hours in 2024, and reported 44% shorter production cycles and 76% more output per mold across its plants that year [S23]. Hours-long cures are normal for infusion epoxies: CR83 is specified at 8 hours at 70 °C [S3].

Closed molds are faster but not always fast. Light RTM can cycle in as little as 20 minutes [S16]; a vendor puts unheated light RTM at 4 to 6 parts per shift against 10 to 12 for unheated RTM, and up to 20 with heated molds [S9]. Conventional RTM runs 30 to 60 minutes, falling to 5 minutes only for very small parts [S10]; a US molder made 45,000 RTM fan housings a year at 30 to 45 minute cycles by running many composite tools in parallel [S17]. HP-RTM parts are realistically produced in 3 to 7 minutes [S10], and Fraunhofer ICT notes that 50,000 to 80,000 parts per line per year require a cycle of 3 minutes or less [S11].

For capacity planning, multiply tools rather than speed. A single heated light RTM mold at 20 parts per shift delivers about 5,000 parts per year on one shift over 250 working days [estimate: 20 × 250]; an HP-RTM line at 3 minutes and the same 250 days of 24 hours delivers about 120,000 shots, which is why the Fraunhofer figure assumes real-world downtime [estimate: 250 × 24 × 20]. Ask suppliers for demonstrated cycle and uptime on a comparable part, not the resin datasheet's gel time.

5. Process variants: light RTM, HP-RTM, compression RTM, SCRIMP and VARTM

SCRIMP and VARTM are the infusion branch. William Seemann's SCRIMP patent, filed in March 1989 and granted in February 1990, put a resin distribution medium on the laminate so that resin spreads over the surface and then soaks through the thickness [S14]; NASA describes VARTM as resin injection and fiber compaction both achieved at 101.3 kPa or less [S8]. Wind blades are made by vacuum-assisted resin transfer molding in this sense [S12].

Light RTM keeps the low pressure but adds a second face. The semi-rigid counter-mold is clamped by a flange channel held at at least 23 in Hg, the cavity is held at about 15 in Hg, and injection stays below 1 bar, so the molds are simpler and cheaper because they do not have to contain high cavity pressure [S9] [S16]. Conventional RTM adds an injection machine and clamping for 10 to 20 bar [S10].

HP-RTM and compression RTM are the automotive branch. HP-RTM meters reactive epoxy or polyurethane through a high-pressure mixing head and injects it into a preform in a hot steel tool [S10]. In compression RTM the tool is left slightly open during injection and then closed: a Fraunhofer pressure-controlled variant with a fixed 0.3 mm gap held cavity pressure at a 30 bar limit and needed 75% less press force than injection RTM [S11], and a GM and ESI simulation found unrestricted injection peaking at 237 bar while a fast resin cut cost per part from USD 160 to USD 60 [S15]. Wet compression molding, used with HP-RTM in the BMW 7 Series Carbon Core, skips the preform and presses resin-wetted fabric stacks [S20].

6. Infusion and RTM resins: viscosity, pot life and cure

Infusion needs low viscosity and long pot life; HP-RTM needs low viscosity and very fast cure. HP-RTM practitioners recommend 50 to 200 cP at injection [S10]. Hexion's EPIKOTE MGS RIMR 135 infusion resin is 700 to 1,100 mPa·s alone and is mixed 100:30 with hardeners of 10 to 80 mPa·s; heating takes the mix down to about 150 mPa·s, with pot lives from about 0.5 to 4 hours depending on hardener [S4]. SikaBiresin CR83 mixes at 155 to 170 mPa·s, gives 60 to 300 minutes of pot life, cures 8 hours at 70 °C and reaches a Tg of 80 to 84 °C [S3]. Vinyl ester for RTM, such as DERAKANE 8084, runs at 360 mPa·s with a Tg of 115 °C [S5].

At the fast end, Dow launched VORAFORCE 5300 for HP-RTM at an extremely low 15 mPa·s with cure under 90 seconds and about 60 seconds possible [S6]; Olin now sells it as LITESTONE VF4530 and VF4535, alongside LITESTONE infusion systems with a Tg around 80 °C [S7]. BMW's i3 used Huntsman Araldite LY 3585 with hardener XB 3458, curable in as little as five minutes at 100 °C [S19].

Two listed Chinese formulators show the scale of the domestic resin base. Shanghai Swancor Advanced Materials, which trades as Shangwei, makes infusion, hand lay-up and mold resins, adhesives and spar-cap pultrusion resin; it reported revenue of RMB 1.797 billion in 2025, up 20.3%, and acknowledges a gap to Olin globally [S28]. Kangda New Materials sold nearly 90,000 t of wind epoxy structural adhesive, infusion resin and related products in 2024 [S29]. Zhuzhou Times New Material's average purchase price for wind-blade raw materials was RMB 26,981/t in 2024, RMB 30,379/t in 2025 and RMB 32,171/t for January to September 2026 [S25] [S26] [S27]. That is a blended materials basket, not an epoxy price, but it shows the 2025–2026 cost direction.

7. Tooling cost and the volume break-even between infusion and RTM

Tooling is where the processes really separate, and published numbers are scarce enough that every one counts. For light RTM, one US vendor estimates a car-hood mold at about USD 140 to 210 per square foot, excluding the master pattern [S9], or roughly USD 1,500 to 2,260 per square meter [estimate: × 10.764]. Composite RTM mold skins last 2,000 to 6,000 cycles and cost about one quarter to one half of steel or nickel-shell RTM tooling, while nickel shell or steel is chosen for higher volumes of 50,000 and more [S17]. Sinoma depreciates its wind-blade molds by usage over 400 pieces [S47], which gives a useful benchmark for infusion mold life in blade service.

HP-RTM adds a press. BMW's Leipzig i3 line used seven 3,000-ton Schuler presses, each with two KraussMaffei injection units [S19]; the UK National Composites Centre installed a 36,000 kN Schuler press with a 3.6 × 2.4 m bed for HP-RTM development [S21]; and China's PGTEX ordered a 3,600-ton Schuler press with Hennecke dosing for HP-RTM and wet compression molding of electric-SUV body parts [S22]. HP-RTM still cut BMW's manufacturing cost for CFRP body components by about 50% compared with its earlier roof production [S10].

There is no peer-reviewed infusion-versus-RTM break-even volume, because the answer depends on part size, faces and labor. What the sources do support is a set of volume bands. Infusion and composite light RTM tools fit units to a few thousand parts, consistent with blade molds rated for about 400 pieces [S47] and composite RTM skins for 2,000 to 6,000 cycles [S17]. Metal RTM tools start to make sense toward 50,000 parts [S17], and HP-RTM targets 12,000 to 50,000 parts a year per line [S10]. To find your own break-even, ask each bidder for tool cost, tool life in cycles, cycle time and labor minutes per part on the same drawing; break-even quantity is the tooling difference divided by the per-part cost difference.

8. Wind blades: the largest infusion application and China's blade makers

Wind blades are by far the largest use of vacuum infusion. The world installed 117 GW of wind in 2024, 109 GW onshore and 8 GW offshore, and China added 79.8 GW of it [S31]. A peer-reviewed bill-of-materials study found that a single 20 t blade for a 4 MW onshore turbine uses 13.4 t of glass fabric and 7.2 t of epoxy, and a single 55 t blade for a 12 MW offshore turbine 37.4 t of glass fabric and 20.0 t of epoxy, with about 10% material loss [S32]. For a three-blade rotor that is about 9.3 to 10.0 t of glass fabric and 5.0 to 5.4 t of epoxy per megawatt [estimate: 3 × 37.373 ÷ 12 and 3 × 13.386 ÷ 4; 3 × 19.986 ÷ 12 and 3 × 7.155 ÷ 4].

Chinese blade makers dominate the volume. Sinoma Wind Power Blade, a subsidiary of Sinoma Science and Technology, sold 24 GW of blades in 2024 and 36.2 GW in 2025, up 51%, with blade revenue of RMB 12.59 billion and net profit attributable of RMB 0.62 billion; it describes its market share as first in the world, runs 15 Chinese plants and one in Brazil, and offers blades from 31 to 147 m for 6 to 20 MW-plus turbines [S23] [S24]. Zhuzhou Times New Material sold 21.09 GW in 2024 for RMB 8.20 billion, 31.79 GW in 2025 for RMB 10.27 billion, and 22.21 GW in the first nine months of 2026 for RMB 7.38 billion [S25] [S26] [S27]. Both list glass and carbon fiber, balsa, PET or PVC core and epoxy as the main materials [S24].

Outside China the picture is thinner. TPI Composites, the US independent blade maker, filed for Chapter 11 in August 2025 with up to USD 27.5 million of new debtor-in-possession money [S44] and emerged under new ownership in July 2026 [S45]. Siemens Gamesa still molds 75 m offshore blades in one piece at Hull, 450 a year at capacity [S43]. For buyers of infused parts other than blades, the blade industry matters because it sets the price and availability of multiaxial glass fabric, infusion epoxy, core and consumables in China.

9. Automotive HP-RTM: what BMW proved and where China stands

BMW's i3 remains the reference case for HP-RTM in series production. Each Life Module was built from 39 stacked fabric kits, formed into 49 preforms and molded into 16 HP-RTM parts; resin was injected at more than 40 bar and cured in under 10 minutes, in seven 3,000-ton Schuler presses, while about 100 cars a day left the Leipzig plant [S19]. KraussMaffei supplied 20 HP-RTM machines for BMW's Leipzig and Landshut plants [S18]. The later 7 Series Carbon Core combined HP-RTM, wet compression and other methods to make 16 CFRP parts, about 4,000 parts a day [S20].

China's HP-RTM base is smaller and newer. PGTEX, better known as a glass-fabric weaver, ordered a 3,600-ton Schuler press with Hennecke dosing for HP-RTM and wet compression molding of electric-SUV body parts [S22]. The same company appears in the EU fabric anti-dumping regulation at 37.6% [S48], which illustrates how Chinese fabric makers are integrating forward into parts. Buyers evaluating Chinese HP-RTM capacity should ask for the press tonnage, bed size, dosing equipment make, preform cell and demonstrated cycle on a comparable part.

10. Standards, Chinese supply base and trade measures

Laminate acceptance uses common test standards on both sides. Void content is measured to ASTM D2734-23 [S39] or, for carbon laminates in China, GB/T 3365-2008, which covers void content and fiber volume of carbon-fiber-reinforced plastics and, by reference, glass and aramid [S36]. Resin content of glass laminates by burn-off is GB/T 2577-2005 [S37], and density is GB/T 1463-2005, to be replaced by a 2026 edition from 1 December 2026 [S38]. Blades are designed and certified to IEC 61400-5:2020 [S40], DNV-ST-0376, edition April 2024 [S41], and in China GB/T 25383-2025, in force since 1 December 2025 [S35].

The Chinese supply base for infusion is deep. China produced 8.43 Mt of glass fiber in 2025, and China Jushi alone sold 3.20 Mt of roving and products [S30]; Taishan Fiberglass has more than 1.7 Mt of capacity including high-modulus yarn and multiaxial warp-knitted fabrics for blades [S24]. Swancor and Kangda supply epoxy infusion resins and adhesives [S28] [S29]. AVK puts European RTM and infusion output at 109 kt in 2025, down 3.5% [S42], while applying the per-megawatt ratios in Section 8 to China's 79.8 GW of 2024 wind additions implies roughly 0.74 to 0.80 Mt of glass fabric for blades alone [estimate: 79,824 MW × 9.3–10.0 t/MW], an upper bound because pultruded spar caps replace part of the fabric.

Trade measures bite hardest on fabric. The EU's June 2026 expiry review kept anti-dumping duties on woven and stitched glass-fiber fabrics from China at 69.0% for Jushi, Hengshi and Taishan and for all other companies, 37.6% for PGTEX and Chongqing Tenways, and 34.0% for Jiangsu Jiuding, covering CN 7019 61 to 7019 66 and extending to imports consigned from Morocco and Turkey and to offshore installations [S48]. In the United States most woven and stitched glass fabric lines in HTS 7019.61 to 7019.66 are on Section 301 List 3 at an additional 25% [S46]. Finished infused parts and blades fall under different headings, so check the classification of the finished article before assuming the fabric duty applies.

11. Infusion and RTM defects: dry spots, race-tracking and voids

Dry spots are the signature defect of liquid molding. In blades they come from race-tracking, where resin runs ahead along a high-permeability edge or gap and encircles an unfilled region, or from bottlenecking at thick sections [S33]. The same study measured 12 to 20% thickness variation across an infused blade section and found that fiber waves cut failure stress to about 74 to 77% of a flat laminate [S33]. RTM adds tool-specific risks: preform edges that do not fit the cavity race resin around the perimeter, and vents that are not at the last point to fill trap air.

Inspection has to match the risk. Visual inspection catches surface dry spots and wrinkles; ultrasonic testing, X-ray and acoustic emission find internal defects [S33]; acid digestion, burn-off and micrographs measure fiber and void content on witness panels or cut-outs [S8] [S37]. Set void and fiber limits in the purchase specification, for example voids below 2% where structural performance matters, the level NASA treats as desirable for aerospace [S8], and require the supplier to show where the witness coupons are taken.

12. RFQ checklist for infused and RTM parts

An RFQ for liquid-molded parts should fix the laminate and the evidence, and let the supplier propose the gate plan. Ask for process data that can be checked on the first article, and for the same documents from every bidder so that infusion and RTM offers can be compared on cost per accepted part.

• 1. Process and tool: infusion, light RTM, RTM or HP-RTM; number of finished faces; tool material (GFRP, aluminum, nickel shell, steel); tool life in cycles; tool ownership.
• 2. Laminate: fabric types and areal weights, ply schedule, core type and grooving, target fiber volume fraction (for example 50–57% for infused glass) with tolerance.
• 3. Resin: system name and hardener, mixed viscosity at process temperature (for example 150–200 mPa·s), pot life versus planned fill time, cure cycle and Tg, DNV or other approval if required.
• 4. Fill plan: inlet and vent locations, flow media, planned fill time, vacuum level and leak-rate check (for infusion), injection pressure and clamping (for RTM).
• 5. Quality limits: voids below an agreed level (for example 2%) to ASTM D2734 or GB/T 3365; fiber or resin content by burn-off (GB/T 2577) or acid digestion (ASTM D3171); thickness tolerance; no dry spots.
• 6. NDT: ultrasonic or tap testing coverage, witness-panel locations, Tg on witness coupons.
• 7. Capacity: demonstrated cycle time and molds available for the part family; for HP-RTM, press tonnage, bed size and dosing equipment.
• 8. Standards: IEC 61400-5, DNV-ST-0376 or GB/T 25383-2025 for blade work; customer specification otherwise.
• 9. Trade: HS code for the finished part, EU anti-dumping exposure if fabric is imported separately (34.0–69.0% on Chinese glass fabrics), US Section 301 status.
• 10. First article: full dimensional report, sectioned coupon, fill video or log, batch records for resin and fabric.

Frequently asked questions