– Recipe Effect Art – ABC Example Art(Download this article for off-line reading)Much of the mountain of plastic products available in today’s global marketplace could not be manufactured, produced or formulated without the involve¬ment of continuous processing techniques.

Click Image to Enlarge

Much of the mountain of plastic products available in today’s global marketplace could not be manufactured, produced or formulated without the involve¬ment of continuous processing techniques. And central to con¬tinuous processing is the weigh-feeder. Its role in assuring product quality through the processor’s requirements of feeder performance will be probed in this article.

Working Backwards
Long before any plastics process line turns out its first stream of product, researchers have precisely determined the pro¬duct’s desired formulation -- a target which, if followed. will result in a product possessing known properties at a known cost. Along with that target formulation goes another precisely determined set of specifications which reveal the tolerances associated with each of the pro¬duct’s ingredients. Regardless of whether the end product is a pellet, part or profile, tolerances are dictated on a case-by-case basis by the limits beyond which the product exhibits some undesirable property or combination of properties. Totally dependent on the particular application, these properties may focus on the color of a plastic, its processability, durability or any of dozens of other relevant properties.

Recognizing the fact that quality assurance tests and analyses are performed on the end product itself, quality assurance specifications usually spell out ingredient levels and tolerances in terms of weight percentage of the end product tested. However, in order to determine the re¬quired accuracy of each ingredient feeder, we must begin with the end product specifications and work backwards.

Assume that we have 100 lb of perfectly constituted compound on hand. It is comprised of exactly 50 lb of in¬gredient A 40 lb of ingredient B, and 10 lb of ingredient C. Laboratory tests might reveal that the ABC compound still retains its desired properties if at most two pounds of ingredient A is either added or taken away from the 100 lb of compound. The end product tolerance for ingredient A is then ±2 lb for every 100 lb of compound. This can be equivalently expressed as a tolerance of ±2%.

Similarly, tests may indicate that only one pound of ingredient B may be added or subtracted from the 100 lb of compound ABC before the overabundance or deficit becomes objectionable in some way. The end product tolerance for ingredient B would then be ±1 lb or ±1%.

In¬gredient C may be an especially critical ingredient whose presence in 100lb of compound must not vary from spec by more than 0.1 lb either way. Its end product tolerance is a narrow ±0.1%.

These tolerances have great significance to the laboratory technicians who analyze the end compound, but the accuracy of the feeders delivering the ingredients is based on the respective ingredient’s flow rate, and not on the total flow. A flow rate adjustment to the tolerances is required to determine the necessary feeder accuracy.

Assuming for simplicity a total material flow of 100lb/min among the three ingredient feeders, feeder A delivers 50 lb/min, feeder B provides 40 lb/min, and feeder C 10 lb/min. In one minute, therefore, feeder A delivers all of its portion required to make 100 lb of com¬pound ABC. If the end product tolerance for ingredient A is ±2 lb, then feeder A is allowed to provide as little as 48 lb or as much as 52 lb during a minute’s operation. That ±2 lb of leeway now becomes ±4% of ingredient A’s feed rate (±2 lb/50 lb). To achieve a ±2% end product tolerance for an ingredient comprising half the blend, therefore, we need only control the accuracy at the feeder to ±4%.

In a similar fashion feeder B is required to achieve an accuracy of ±2.5% (±l lb/40 lb), and feeder C must perform to an accuracy of ±1% (±0.1 lb/10 lb). This ‘recipe effect’ interplay between end product ingredient tolerance, ingredient proportion and required feeder accuracy can be expressed more formally in equation form:

Required Feeder Accuracy (+%) =
Tolerance in Recipe (+%) / Proportion in Recipe (%)

...or more conveniently (if not more accurately) graphically illustrated below.

Assurance Insurance
Translating ingredient tolerance specifications into feeder accuracy requirements is an important step in at¬taining proper formulations, but an equally important cau¬tion is in order because consistently assuring end product quality is not just a matter of math.

As in any feeding equipment selection process the decision to go with gravimetric or volumetric devices must be made. The fact that accuracy requirements at the feeder are less stringent than the corresponding end pro¬duct tolerances may lure the unwary into choosing the volumetric approach wherever possible. In the strong ma¬jority of cases where feeder accuracy requirements do not preclude a volumetric device, such a choice is still discouraged on several grounds.

First, even though a volumetric feeder may be able to perform acceptably under test conditions, there are far too many variables that come into play in extended use to assure that the same degree of performance will be reliably repeated. Material density may change, altering the volumetric feeder’s calibration to the detriment of per¬formance. The feeder’s metering element may, under cer¬tain circumstances, become caked with material, again altering calibration. Many volumetric feeders also lack sufficiently precise and stable control, and thus are suspect. Short of frequent inspection, testing and calibration, there is simply no way to assure a volumetric feeder’s on-spec perfor¬mance.

And that leads to the second reason to think hard before relying on a volumetric feeder for quality assurance duty: volumetrics are open-loop devices. Because volumetric feeders cannot report on their own perfor¬mance (as gravimetric feeders do), reliance must be placed on the shaky inference that if power is applied to the feeder, it must be working. How well it is working is another question, the answer to which is anybody’s guess -- at least until it is checked out manually.

A third and especially compelling reason to opt for the gravimetric approach involves the special circumstance where one or several ingredients must never fall below some specified minimum proportion in the overall formula¬tion. Even substantial overfeeding may be allowed (usually increasing total blend cost), but any underfeeding will result in product reject. The only practical way to avoid ingredient underfall is to intentionally overfeed to consistently guarantee the ingredient’s re¬quired minimum level presence in the formulation.

Using a volumetric device to do the overfeeding re¬quires that its set flow rate be substantially higher than if a gravimetric feeder were to be used for the same purpose. This is because the variability of a volumetrically controlled flow stream is innately greater than that of a gravimetrical¬ly controlled stream, all other things being equal. As a result. employing a gravimetric feeder allows the flow rate to be safely set closer to the minimal, target requirement. Substantial cost savings often result when the presence of the usually expensive ‘constrained’ ingredient is minimized.

Doing the Detective Work
Rejoining our ABC compound example, let us assume that feeders B and C are slaved to feeder A. Further, assume that feeders A and B are weigh-belts, and feeder C is a loss-in-weight.

To recap, feeder A is set to deliver 50 lb/min; feeder B, 40 lb/min; and feeder C, 10 lb/min as shown below. All feeders are running and appear to be on spec. A laboratory analysis of the compound indicates, though, that ingredient C is running dangerously close to its max¬imum tolerance. If nothing is done there is the risk that in¬gredient C, the critical ingredient. will increase further, cross its tolerance limit, and result in product reject. What should be done? Look at the possibilities.

First, it is important to realize that the laboratory pro¬cedure in this hypothetical example only tests for the con¬tent of critical ingredient C in the end product compound. This testing approach provides inadequate information to directly isolate the problem, but it is most descriptive of such real world cases. Few formulators (regardless of in¬dustry) run quality assurance checks on all or even most ingredients.

As a rule, quality assurance tests are performed on the formulation’s one or two critical ingredients -- ingredients whose cost is high or whose effect in the blend is crucial to the product’s performance. Experience has shown this to be a prudent approach because problems can still be quickly and economically identified, if not im¬mediately isolated. Combined with inferential data and the experience of knowledgeable process personnel, critical ingredient testing can point the way to the quality assurance problem’s source.

On the surface it appears that ingredient C is running high. The flow rate display, though, reports that ingredient C is on the mark at 10 lb/min. C’s setpoint confirms that the compound recipe has been entered correctly at 20% of A’s master feed rate of 50 lb/min. But if the feeder is simply out of calibration, the display won’t reveal this problem so further investigation is necessary. Catch sampling is called for.

One of the most common errors in catch sampling a suspect stream of material is inadequate sample size or duration. Obtaining a single sample is worse than taking no samples at all because it will invariably lead to an endless chain of sampling and recalibrating. Recalling high school statistics reminds us all that a minimum of ten samples is required to fairly confidently determine an average. (A minimum sample size of thirty is needed to determine standard deviation, but here we are only in¬terested in an average.) Even taking ten samples may lead us astray if the samples are taken over too short a duration, are too small to adequately represent the material’s flow, or are obtained without due care.

In our ABC exercise we can assume we have observed proper catch sampling etiquette, and find that feeder C (the loss-in-weight) is delivering its critical ingredient well within acceptable limits. We must conclude that the prob¬lem lies elsewhere. Turning to feeder B and repeating the catch sampling procedure identifies the problem. Feeder B is in an underfeed condition even though controller B’s display and setpoint indicate the feeder was operating on spec. The ABC compound is short on ingredient B, increas¬ing ingredient C’s proportion. The problem has been identified but not isolated.

One of the advantages of today’s digital weigh¬-feeders is that, in addition to substantially better reliability than their predecessors, when malfunctions do occur the feeder can sense the con¬dition, announce it, and even shut itself down if desired. In the modern digital world things either work or they don’t -- there is no ‘in between’. This all-or-nothing characteristic of digital feeding equipment makes our detective work quite a bit easier, in fact. The shift to a slight underfeed condi¬tion for feeder B (rather than a feeder alarm or shutdown) tips us off to the real cause of the problem: something other than ingredient B is applying a load to the feeder’s weighing surface. With a falsely high load signal, feeder B’s controller slows the weigh-belt resulting in the underfeed condition revealed by catch sampling.

Inspecting the weighing zone of feeder B, we look for the phantom load. The two side skirts (whose function it is to prevent material from falling off the side edges of the belt during transit) are adjusted properly, coming close to but not touching the belt. If either skirt had not been ad¬justed properly or fastened securely after adjustment the skirt could have come into contact with the weigh-belt and transferred some of its weight to the weighing system causing a spurious load signal. This, however, was not the case.

The only other possibility, then, is that the belt itself has gained weight. Upon removing and inspecting the belt we see that the problem lies clinging to the underside of the belt. Due to inadequate periodic maintenance the weigh ¬belt has been neglected for some time. and the interior surface of it has become encrusted and impregnated with residual material causing the erroneously high load signal. Cleaning or changing the belt and a simple recalibration solves the problem and compound ABC is on spec again.

Another Case
The case we’ve just solved underscores the need for cleanliness where precision weigh-feeders are concerned, but this next brief and perplexing case focuses on another area of maintenance: proper equipment set-up.

Shortly after the hypothetical ABC feeding system was installed and checked out satisfactorily in pre-service tests, the system was placed into operation. But during the very first day of on-line duty, lab analyses of the compound showed that ingredient C was out of tolerance in the samples tested. The out-of-tolerance condition was not systematic (i.e., not a mere shift) but appeared to be varying widely and randomly around its target value of 10lb/min This time thirty catch samples were taken from feeder C (the number needed to determine standard deviation) and the results confirmed the laboratory’s findings.

Because of the operating personnel’s unfamiliarity with the new equipment the strongest clue to the problem went unnoticed until all the catch samples had been taken. Feeder C’s control panel had been reporting all the while that the feeder was bouncing back and forth bet¬ween volumetric and gravimetric control, spending most of its time under the far less accurate volumetric control mode. Once that condition was noticed the problem was easily narrowed to one of inadequate isolation from vibra¬tion. Loss-in-weight feeders are programmed to recognize when vibration prevents accurate weight sensing. and then shift into volumetric control until the vibration has sub¬sided, at which time they automatically resume gravimetric operation.

Since all loss-in-weight feeders operate by controlling the rate of discharge from the entire weighed feeding system (feeder, hopper and material), a stable and uncon¬taminated weight signal is crucial to accurate operation. Except in the most severe vibration-prone environments where special measures must be taken. adequate isolation from vibration (and other structurally transmitted forces) is easily and inexpensively available. In this case it was found that the flexible connection between the feeder’s hopper and the automatic refill device was improperly adjusted and therefore acted to transmit rather than suppress plant vibrations. The flexible connection was readjusted in minutes and the problem, which did not crop up until other nearby equipment was started, was solved.

An Ounce of Prevention,
a Pound of Assurance

Because feeding and proportioning systems are at the very crux of the formulation of thousands of products, guarding against feeding errors is much the same as assur¬ing product quality. As a consequence, gaining an intimate knowledge of your feeding equipment and how to main¬tain it in top working order is your best insurance against a quality assurance crisis. In fact, infrequently encountered as they are, the detective work cases covered above are the two causes of quality assurance problems most often reported. That is an especially bright sign because both causes are under your full and complete control to pre¬vent or rectify.

For weigh belt feeders, simply keeping the feeding zone and the belt itself clean of build-up will assure that you will seldom if ever be faced with the shift problem described earlier. And, for loss-in-weight feeders, knowing that they require strict isolation from their environment will prepare you to spot and resolve out-of-kilter conditions should they ever arise. Being aware of those two points. and practicing pro¬per sampling techniques will cut your quality assurance problems to the bare bone limit.

With the major advances in weigh-feeder technology, both in hardware and software, system reliability has in¬creased to the point where the user’s operational knowledge and maintenance patterns now have the over¬shadowing effect on the incidence and severity of quality assurance problems.

To further reduce the losses and inef¬ficiencies of off-spec product formulations. strong and con¬tinuing commitments must be made by operating and maintenance personnel to assure that their feeding equip-ment functions in an environment as hospitable to proper operation as possible.

Correct maintenance and a familiarity with the equipment and its principles of operation are the keys to avoiding crises of quality assurance.