Match your check valve to your screw design.

Research shows there still isn't a 'best' check valve for all occasions. So the key is to fit the valve to the screw design and materials processed.

Injection screw design has necessarily evolved beyond its roots in extrusion to become a science of its own. One overlooked factor in most models and studies of reciprocating screw design is the effect of the non-return valve on screw performance. Recent evaluations by Dow Plastics have demonstrated that there is, in fact, a critical relationship between the two.

Today there are at least 10 designs of non-return valves available in North America. Dow compared the performance of several valve designs in molding four engineering thermoplastics on a production-size, well-instrumented injection molding machine. The project grew out of screw-design research that revealed a drastic difference in the performance of a screw that was tested with and without a non-return valve. In the first trial, the screw ran on an Egan 2.5-in. extruder at a rate of 200 lb/hr. That rate dropped to 15 lb/hr after the same screw design was fitted with a check valve and run on an injection molding machine with a 2.5-in. barrel.

The key role of the non-return valve in relation to screw performance was again demonstrated in another project involving the complete rebuilding of an injection molding press to achieve a statistically "capable" machine. In this project, a 200-ton, 1974 Cincinnati Milacron press was completely refurbished with new hydraulics, machine controller, barrel, "general-purpose" screw, and "free flow" non-return valve in an effort to obtain better shot-to-shot consistency.

When the rebuilt machine was evaluated with an easy-flowing ABS, outstanding process capability was recorded. Yet performance dropped off when molding polycarbonate and PC blends. To improve melt quality, a low-compression-rate screw was substituted,and repeatability was improved somewhat but still couldn't match that of the ABS. (Compression rate is a measure of how fast the compression ratio is developed in a screw design.) Finally, the existing check-ring valve was replaced with one specifically designed for PC, after which the molding performance matched that achieved with ABS.

These projects served as the impetus for further research into non-return valves. It was evident that the non-return valve should be considered a suspect in virtually every kind of part defect. It was also apparent that performance of any given non-return valve design is polymer dependent.

KEY VALVE DESIGN FEATURES

Having established the importance of the non-return valve to screw performance, Dow's next project goals were to identify key design features of a successful non-return valve by type, and to establish valve design recommendations for Dow resins. Eventually, the project expanded to include an understanding of the performance of the screw/non-return valve system through finite-element flow analysis, as well as a recommendation for a "general-purpose" valve design that would work with all Dow resin families.

Four basic valve types were chosen for evaluation: front-discharge and side-discharge ball check valves; low-flow, medium-flow, and high-flow ring check valves, including three-piece and four-piece designs; a poppet check valve; and a premium-priced proprietary poppet-valve design.

To represent a "real world" production environment, the evaluation was carried out on a 330-ton Mannesmann Demag injection machine with a 70-mm barrel. A production mold using 50% of barrel capacity was chosen along with three screws, including an OEM general-purpose screw and two proprietary mixing screws. Four resin families were run, including ABS, PC/ABS, TPU/ABS, and rigid TPU (R/TPU).

Among the performance criteria for a successful non-return valve: the valve must seat rapidly, not leak, and seal effectively against injection pressure. At the same time, the valve should be cost-effective, not degrade the polymer, and provide a streamlined flow path--i.e., there should be no areas in the valve that would cause material to hang up.

Only the last two of these criteria are directly measurable--everything else had to be inferred. A variation in part weight, for example, could result from either valve leakage or a change in melt temperature, or both; a third measure, cushion after injection, is helpful in determining the actual cause. An effort was made to measure every possible source of variation on the injection machine during the molding cycle.

SCREW/VALVE INTERACTIONS

A design-of-experiments procedure was performed to screen and compare the performance of the various valves. Variables included the three screws, nine non-return valves, four different polymers, screw rpm, and backpressure. Measured responses were melt temperature, cushion, peak injection pressure, injection time, recovery time, torque, part weight, part appearance, injection work (integration of injection pressure with respect to time), and recovery work (integration of screw motor pressure with respect to time). The results are summarized in Table 1, where the "three-sigma" range, "R" (three standard deviations above and below the mean) of variation is expressed as a percentage of the mean value.

One surprising result was that the type of valve design had relatively little effect on the performance. A good basic design and how well a valve seats are more important to performance than the particular type of valve it is--whether it be ball, ring, three-piece, or four-piece. For example, the proprietary poppet valve gave the same part-weight variation as the ball valve. Cushion variation was greater with the ball valve than the ring valve; this is expected because the ball valve must leak in order to seat during injection, driving the cushion back up the screw rather than into the mold. But recovery time for the high-flow ring check valve and front-discharge ball valve was about the same, as was the injection time for the medium-flow and high-flow ring valves and front-discharge ball valve. On this particular machine, injection time variation would indicate valve leakage.

On the other hand, machine set-ups, such as screw speed and backpressure, affect how the valve operates. There is an optimal machine set-up for each valve, pointing to an interaction between valve and screw design. Poor plasticating performance often manifests itself in high melt temperatures, suggesting either viscous heating in the valve or inefficient screw performance.

In order to rank the valves according to performance, the response in each category was ranked from 1 (best) to 7 (worst). In addition, the different response categories were weighted from 1 to 4 according to a judgment of their relative importance--again, with 1 equaling highest importance. Achieving lowest melt temperature was weighted an importance of 4; lowest part-weight variation, 3; lowest recovery-time variation, 2; shortest recovery time, 2; and lowest cushion variation, 1. The response ranking was multiplied by the response weight in each category, and the results for all categories were added up for each valve and material combination. This evaluation enabled Dow to begin to come up with recommendations for different valves.

ADVANTAGES OF RING VALVES

Based on the trials, a four-piece ring valve--consisting of rear seat, retainer, ring, and forward seat--gave the best performance for the materials tested. This valve performed well in terms of seating performance and flow of the material through the valve. One reason is that the ring valve provides plenty of seating area for the resin to push against, resulting in more positive sealing force. A ring valve typically has four times more area exposed to injection pressure than a ball valve.

One drawback of the ball valve--in addition to its small forward surface area--is that it must leak in order for it to seat. Material must actually flow past the ball to suck it into its seat. Ball valves also contain a large mass of steel that must be heated to prevent the material from freezing.

Like the ring valve, the poppet has the advantage of a large surface area. But the holes for the internal flow path in the proprietary fixed poppet valve that was tested were too small, restricting flow.

From these experiments, several conclusions can be generalized:

* Type (ball, ring, poppet) of non-return valve is not as important as basic design features of the device--e.g., does it seat well.

* Non-return valve flow-passage design is important.

* Selection of the "best" non-return valve depends on the material.

* Expensive, proprietary designs aren't always the best choice.

* Non-return valve design is an important consideration in the design of a reciprocating screw.

SEARCH FOR A 'GENERAL-PURPOSE' VALVE

Traditionally, researchers have assumed that the head pressure on an injection screw--i.e., pressure required to force melt through the non-return valve and reciprocate the screw during approximated the force required to reciprocate the screw. In fact, we have shown experimentally and mathematically that the actual discharge pressure--i.e., pressure at the last flight of the screw--before various non-return valves can be 50-100% higher than that required to reciprocate the screw and is due to pressure drop across the valve. This pressure drop can manifest itself in viscous heating of the melt and poor part quality, and explains how the use of the "wrong" non-return valve can ruin a good screw design. Table 2 shows that no one valve tested gave best (i.e., lowest) melt temperature with all materials.

Current ongoing work is focused on combining a mathematical model of the non-return valve with that of the screw. To that end, a "QSO" adjustable-flow poppet valve (supplied by Great Lakes Feed-screws, Tecumseh, Mich.), which allows adjustment of the clearances for each material to be run, was modeled and evaluated with FIDAP two-dimensional finite-element fluid-analysis software from Fluid Dynamics International, Evanston, Ill. The analysis of the non-return valve was combined with that of a Barr ET screw (from Robert Barr Inc., Virginia Beach, Va.) with double-flighted mixing section. The ultimate goal is to find a general-purpose non-return valve and screw combination with the least compromises for all materials. The adjustable-flow popper shows promise in meeting this target.

Until the goal of a general-purpose non-return valve and screw combination is reached, a four-piece ring valve is recommended for all applications. One benefit of the four-piece ring valve is flexibility: inexpensive replacement of one component, the rear seat, allows a valve to be "tuned" to a particular application. Finally, valve selection should take into consideration the screw as well as the materials to be processed, and vendors should be knowledgeable about these requirements.

TABLE 1--TYPICAL EVALUATION RESULTS
Easy-Flowing ABS with color concentrate.
                                                                 Front
                               Medium           High             Discharge
Response                       Compression      Compression      Ball

Peak Injection, psi            1634             1635             1564
Injection Time Ave.            4.58             4.62             4.69
Injection Time 3|Sigma~R       3|Sigma~=6%      3|Sigma~=6%      3|Sigma~=6%
Injection Work 3|Sigma~R       3|Sigma~=0%      3|Sigma~=5%      3|Sigma~=1%
Recovery Time 3|Sigma~R        3|Sigma~=1.6%    3|Sigma~=2.6%    |Sigma~=2.2%
Recovery Work 3|Sigma~R        1300             1391             1396
Cushion 3|Sigma~R              3|Sigma~=2.9%    3|Sigma~=4.0%    3|Sigma~=5.1%
Mass 3|Sigma~R                 3|Sigma~=0.1%    3|Sigma~=0.3%    3|Sigma~=.03%
Melt Temperature, |degrees~F   424              431              430
TABLE 2--VALVE INTERACTIONS WITH SCREW DESIGNS
Best conditions(*) for each valve with easy-flowing ABS.

                                    Front
                           Fixed    Discharge    Low             Medium
Condition (setting)        Poppet   Ball         Compression     Compression

Screw Speed (Step)(**)     32       16           48              32
Backpressure (Step)        6        6            6               32

Worst-case melt temperature for each valve/material set.

                                     Front
                           Fixed     Discharge   Low             Medium
Material                   Poppet    Ball        Compression     Compression

ABS                        492       444         446             430
TPU/ABS                    519       413         413             404
PC/ABS                     515       489         478             493

* Considering part mass variations.
** Control setting on this digital value machine.
TABLE 3--RANKING OF VALVES (Lowest=Best)

Design                      ABS     TPU/ABS     PC/ABS     R-TPU

Low Compression Ring        41      20          23         37
Medium Compression Ring     18      42          23         50
High Compression Ring       39      22          39         28
Front Discharge Ball        24      28          55         45
Fixed Poppet                51      47          52         42