Improving Autoclave Performance
This article by Barry Berenberg was originally published in High Performance Composites magazine and later appeared at www.compositesworld.com.
Autoclave processing is the most common method used for curing thermoset prepregs. The curing of thermoset composites involves both mechanical and chemical processes. Mechanically, pressure is applied to remove trapped air and volatiles, and to consolidate the individual plies and fibers. Chemically, a crosslinking reaction must be initiated and taken to completion to form a rigid matrix. Crosslinking is most commonly initiated through the application of heat, though it also may be initiated by exposure to ultraviolet light, microwaves, or high-energy electrons (e-beam curing). In the autoclave process, high pressure and heat are applied to the part through the autoclave atmosphere, with a vacuum bag used to apply additional pressure and protect the laminate from the autoclave gases. The cure cycle for a specific application is usually determined empirically and, as a result, several cure cycles may be developed for a single material system, to account for differences in laminate thickness or to optimize particular properties in the cured part.
The typical autoclave cure cycle is a two-step process. First, vacuum and pressure are applied while the temperature is ramped up to an intermediate level and held there for a short period of time. The heat reduces the resin viscosity, allowing it to flow and making it easier for trapped air and volatiles to escape. The resin also begins wetting the fibers at this stage. In the second ramp up, the temperature is raised to the final cure temperature and held for a sufficient length of time to complete the cure reaction. During this step, the viscosity continues to drop, but preset temperature ramp rates and hold times then stabilize viscosity at a level that permits adequate consolidation and fiber wetting, while avoiding excessive flow and subsequent resin starvation. These control factors also slow the reaction rate, which prevents excessive heat generation from the exothermic polymerization process.
Controlling the autoclave
A good cure requires uniform pressure distribution. The autoclave is designed to deliver uniform pressure on even the most complex part contours. Even more important, however, is uniform temperature distribution throughout the part. Final laminate properties depend not only on reaching and maintaining the cure temperature, but on the entire thermal history. If adjacent regions in the laminate heat at different rates, the resin properties will differ between those regions, potentially leading to built-in cure stresses, nonuniform consolidation and trapped volatiles. The ultimate result is a laminate with sub-optimal properties.
While the autoclave can almost guarantee uniform pressure, it alone cannot guarantee uniform temperature. Temperature distribution depends not only on the autoclave design, but also on the shape of the part, the thicknesses of the part and the tool, the loading of the autoclave (how many parts are cured at once and how they are placed within the autoclave), the thermal properties of the part and the tool, and the reactive properties of the resin system. Because of these myriad variables, composite manufacturers must translate the recommended prepreg cure cycle into a recipe, or a set of instructions, for each unique combination of autoclave and part batch. To reduce the trial-and-error required to develop these recipes, manufacturers are increasingly turning to control systems that can adapt to changing conditions.
Autoclaves are basically controlled by setting the desired temperature and pressure. Vacuum ports also allow control of vacuum pressure on individual parts. The conditions within the autoclave can be monitored through pressure transducers (PTs), thermocouples (TCs) and vacuum gauges. Open loop systems simply monitor autoclave conditions, verifying that the actual temperature, pressure and vacuum accurately reflects what preset controls intended. A closed loop system, however, can use the monitored readings to evaluate cure progress and, if necessary, adjust the temperature, pressure and/or vacuum while the cure cycle is in progress.
Guarding the cure cycle
Most commercial control systems capable of such real-time modifications today primarily use the TC readings to control the cure cycle. Multiple TCs measure both the autoclave air temperature and the part temperature at critical locations. The air temperature is set to bring the part up to a specified temperature at a specified ramp rate. Most controllers monitor both the highest (lead) temperature and the lowest (lag) temperature on the part and can slow the ramp rate if the difference becomes too great.
Although the Sentinel system, a general-purpose industrial controller from CompuDAS (Shelton, Wash., U.S.A.), has been in use since 1978, it was not applied to autoclaves until a few years ago. Because CompuDAS makes only the controller, the Sentinel can work with any autoclave, and can be retrofitted to existing systems. A standalone system that performs both control and data-acquisition functions, the Sentinel uses a Windows-based personal computer (PC) to set up the recipes and view data after the cure. During the cure cycle, however, it runs without intervention from the PC and stores data on its own internal hard drive. This built-in safety feature enables the cure cycle to continue uninterrupted and protects against data loss in the event that the PC crashes. When the PC comes back online, stored data from the Sentinel can be downloaded for analysis. The system can be monitored from a PC located within a company’s internal network, over the Internet, or through a dialup line. CompuDAS has the ability to dial in to a customer’s system during a cure cycle to troubleshoot problems in real time.
Eric Pierce, chief engineer at CompuDAS, points out the flexibility of the Sentinel. The cure cycle can be controlled by any process variable, he says. Usually customers use the lagging part temperature, but control also can be based on the leading temperature, the average of all temperatures, weighted averages of temperatures, the vessel (air) temperature, or even a pressure or vacuum sensor. The system can adjust the ramp rates so all TCs remain within a specified range of each other. If a TC malfunctions during a cure and gives anomalous readings, the control algorithm automatically ignores that TC.
The standard Sentinel has 30 analog inputs for sensors, but is expandable to 420 inputs. It also has 200 digital input/output (I/O) channels (also expandable), as well as eight analog outputs for additional control devices. Configuration of a new Sentinel system, for either a new or retrofitted autoclave, takes about one day. The Sentinel also provides end-to-end calibration: once the system is up and running, the TCs and other sensors can be calibrated on a channel-by-channel basis.
Texas Composite (Boerne, Texas, U.S.A.), a manufacturer of composite structures, upgraded its autoclaves to the Sentinel system in August 2000. Although a single Sentinel is capable of running multiple autoclaves, Texas Composite decided to give each of its three autoclaves a separate controller to increase reliability. If one controller goes down, they can still run the other autoclaves. Fred Garber, plant manager at Texas Composite, describes the Sentinel data as “incredible.” Data can be viewed and graphed in a Microsoft Excel spreadsheet. Each morning, he reviews the previous day’s runs from his office PC, comparing the recipes against the actual cure cycles. The spreadsheet holds data in three-minute intervals, and includes vacuum, part temperature, air temperature and pressure. He can easily search for problems, such as loss of vacuum or part temperatures outside of the allowable range. Setting up a new recipe takes about 10 minutes. Recipes are stored in the PC and can be selected from a dropdown menu. At startup, the system runs through a standard checklist to verify that all sensors are operating properly. It runs an automated vacuum check by pulling a vacuum on the part and holding it for five minutes. If the vacuum loss is greater than specified in the recipe, a warning is issued and the cure cycle does not start. The system also monitors critical channels, as defined by the operator, during the cure: if a sensor exceeds a specified range, an alarm will be shown on the operator PC and an e-mail will be sent to other users.
Downsizing
Anchor Autoclave Systems (Houston, Texas, U.S.A.) provides the Sentinel controller with most of its custom autoclaves. Anchor evaluated a number of other control systems, but the Sentinel proved to be the most cost-effective, providing a more versatile control system for about the same price as conventional controllers. This is especially important for small autoclaves, where the control system accounts for a larger percentage of the total price. Anchor has seen an abrupt shift in the composite autoclave market, from large systems for aircraft and similar work, to small, full-featured systems for military and research laboratories. These small autoclaves now account for nearly 40 percent of Anchor’s sales, compared to less than 3 percent only two or three years ago, due in large part to “homeland security” initiatives in the U.S. that focus on the development of personnel armor, helmets, gun stocks/barrels and similar small products. Anchor president Al Allen says that the laboratories develop the entire manufacturing process for these products in-house, and then contract out the production. The process development includes very specific cure recipes, so the researchers need autoclaves with full control systems.
Uniform temperatures
Even with modern control systems, maintaining uniform part temperature during cure can still prove difficult. In a conventional autoclave, heat enters the aft end of the autoclave and circulates through the entire length and circumference by means of a series of ducts and fans. Attaining proper airflow is critical to achieving uniform temperature distribution. Design of the overall vessel, closures, ducts and fans becomes especially challenging for large autoclaves. Taricco Corp. (Long Beach, Calif., U.S.A.) specializes in large autoclaves, with inside diameters of 1.8 to 4.6m/6 to 15 ft (one current project has a usable inside diameter of 6.2m/20 ft), and lengths of 46.2m/150 ft or more. Company owner Tari Taricco and his engineering team carefully engineer the duct, fan and motor systems for each new autoclave and have developed specialized door hinges and closures for the larger, heavier doors, to minimize sag that can result in leaks due to door/seal misalignment.
Taricco uses its own TCS Thermal Control System. Running on Windows computers, TCS is fully programmable and has no physical limit on the number or types of sensors it can handle. The system can simultaneously control and monitor temperature, pressure, vacuum, and other parameters. Cures proceed automatically, but TCS monitors the sensors for deviations in cure parameters and sends the operator a message if there is a problem. Users can print out standard or customized reports, and analyze TCS data in an Excel spreadsheet or in its native Microsoft Access database format. A single computer can control one or several autoclaves, and a built-in security system provides secure access over a network.
Aeroform Ltd. (Dorset, U.K.) has come up with a unique autoclave design intended to optimize control over the internal temperature distribution. Instead of a single air source that flows axially, Aeroform installs, along the top or bottom of the autoclave, multiple impellers or fans that circulate the air circumferentially. The patented Circumferential Airflow Technology System (CATS) divides the autoclave into zones, one for each impeller. The zone size and spacing, which is uniform, is determined by a thermodynamic analysis of the autoclave dimensions coupled with the tool design or end use. A temperature profile for each zone is programmed into a single controller. The controller then sets the air temperature and airflow rate for each impeller to achieve the desired profile.
Vertical airflow is inherently more uniform than horizontal flow, but the real advantage of the CATS is localized control. The first commercial autoclave of this design was built for creep forming of the Airbus A380 aluminum wing. This wing is 33m/108 ft long, tapers from 3m/9.8 ft wide at one end to 1m/3.3 ft wide at the other end, and has a variable shape and thickness. An axial flow system could not maintain temperature within the maximum allowable variation of ±2°C/3.6°F. When curing a test skin with 100 thermocouples, though, a CATS autoclave with nine zones allowed a maximum variation of only ±0.8°C/1.4°F.
With composites, the mold is often much more massive than the part being cured. As a result, the mold heats up at a much slower rate than the autoclave atmosphere. For sandwich laminates especially, this means that one face of the composite is much cooler than the other face. The CATS system overcomes this problem by directing air up from the bottom of the autoclave, so it impinges on the tool first and transfers heat at a faster rate than into the less massive skin. Again, this airflow can be independently controlled along the length of the autoclave to account for variations in tool thickness and shape.
Like the Sentinel and TCS, CATS is a closed loop controller and is accessible through a modem. The operator can set a maximum temperature differential between adjacent thermocouples and the system will automatically adjust the ramp rates as needed. The system is designed to sense excessive heating from exothermic reactions and back off the heat rate for that portion of the cure. Optional actuators can vary the shape of the circumferential ducts based on the temperature control feedback, providing greater control over local airflow velocities and directions.
Cure monitoring
When developing cure cycles, prepreg chemists use Fourier-transform infrared spectrometry (FT-IR) to measure the percentage of reacted epoxy groups or other reactants. From the results of these laboratory experiments, they determine a heating profile that, ideally, duplicates this reaction profile. However, when manufacturers control the cure process through a thermal sensor, they control the reaction indirectly. A more direct, and therefore, more reliable approach would measure the cure state itself and use that data to adjust the cure cycle in real time. Less trial and error would be required to develop cure cycles for these systems, and the resulting laminates would have improved properties because of the tighter control over the process.
Once a resin system has been characterized in the laboratory, its cure state at any time can be determined from the complete thermal history. AvPro (Norman, Okla., U.S.A.) has developed a system to model the cure state, based on thermocouple readings during an actual cure cycle. Although the process uses conventional temperature measurements, it uses the actual thermal history to model the reaction process and determine the effects of variations in the cure cycle. Most customers currently use the system to develop new cure cycles or validate legacy cures; the hardware is just becoming available to tie the system into a closed loop controller.
Some methods do allow direct measurement of the cure reaction in real time. Currently, the commercially available methods are dielectric and ultrasonic cure monitoring, but the latter is not typically used in autoclave cures because it requires a closed mold. NETZSCH Instruments (Burlington, Mass., U.S.A.) has developed a wide array of instruments for dielectric monitoring of the cure process. Dielectric sensors measure the ionic conductivity of the resin, and can be embedded either within the tool face or within the part itself. Both types of sensors provide the same accuracy, but the embeddable sensors must be incorporated into the layup for each part and are not reusable. The conductivity is directly related to the viscosity of the resin, so the sensors can clearly detect the minimum viscosity point and the onset of gelation. This data can be used to determine when pressure should be applied, or when a temperature ramp or hold should begin or end.
Dirk Heider from the University of Delaware’s Center for Composite Materials (Newark, Del., U.S.A.) has studied a number of cure monitoring systems. He found that most sensors, including dielectric sensors, can show when the cure reaction is complete, but they cannot give an accurate, quantitative measurement of the percent of cure at any given time. Heider’s studies did find one system that does provide such data: the Time Domain Reflectometry (TDR) from Material Sensing & Instrumentation (Lancaster, Pa., U.S.A.). The TDR system interrogates a small, embedded sensor with a rapid pulse to directly measure high-frequency molecular dynamics of the polymer. TDR was developed under a U.S. Army Small Business Innovation Research (SBIR) contract. In that project, TDR was used in a closed loop control system to apply pressure to a laminate at the optimal time in the cure cycle. The system also can be used to control temperature. The Boeing Co. (Chicago, Ill., U.S.A.) and Bell Helicopter (Fort Worth, Texas, U.S.A.) have been testing the system for in-plant use. The sensors also have the potential to be used for in-service strain and damage monitoring in “smart” parts.
The quality of finished composites depends strongly on the cure cycle. Variations in the cure cycle, sometimes even apparently minor variations, can have a negative impact on laminate properties. Advances in autoclave technology, including modern control systems and new duct and heater configurations, are leading to overall improvements in composite quality. New methods for directly monitoring resin properties during cure hold out the promise of fully closing the control loop, enabling autoclaves to adapt to cure conditions in real time.
