A systematic approach to fast and economical problem resolution.
It is a recurrent issue well-known in food processing companies: sometimes, a product that has been successfully manufactured over a long period of time suddenly exhibits a quality problem, although nothing has been intentionally changed (Figure 1). The problem may be discovered by routine quality control, routine sensory analysis, or by customer complaints. Examples include variations of volume, shape, pore structure, color, flavor and mouthfeel, in particular.
In such cases, it is vital for the company to be able to react quickly, as the persistence of the issue may cause critical losses of revenues and reputation.
ttz Bremerhaven is often consulted in these issues and has developed over time a systematic approach that can serve as a guideline when such situations are encountered. In this article, we first describe this overall approach and then illustrate its application with a practical case study from a German artisan bakery.
Figure 1: Same recipe, same process, different results – Standardizing the quality of baked goods is difficult. When the reason for a quality deviation is not obvious, a systematic analysis of the root causes needs to be performed
We are not saying that a single approach is appropriate for solving any kind of quality issues, nor that all measures described below need to be applied in all cases. On the contrary, given the need to solve the problem quickly and the always limited amount of resources, it is important to focus on a few selected measures that promise the best result. Therefore, the approach described below should be understood as a toolbox from which single components can be chosen and combined on a case-by-case basis.
Set up a problem-resolution task force
Coping with the provisory quality concern, identifying its cause and implementing a durable solution will require concerted action from all concerned functions of the company. Typically, production, quality management, technical services, sourcing, marketing and sales as well as public relations may need to be involved. It is advisable to set up a task force and to define one person responsible for coordinating the actions. Information needs to be communicated transparently and without delay within the task force.
Describe the problem well
The first step towards a solution is to make an accurate assessment of the issue. This implies addressing the following points in particular:
+ Which quality properties are of concern? Are these properties measurable and has reliable data been collected in the company on these properties (Figure 2)?
+ In the case of a specification being agreed with the customer, is the product clearly out of specification? Or is the product of slightly inferior quality but, depending on subjective appreciation, still suitable for the market?
+ Which products are concerned? In the case that the problem occurs with several products, identifying the common points between these products will give indications of the possible cause.
+ When did the problem occur? Was everything fine up until a certain date, and has it occurred all the time ever since? If so, was the deviation slow before that date, even though the values were still in the safe range? Or is the problem occurring only from time to time, in a non-systematic way?
Figure 2. An instrumental measurement of relevant product properties (here: dimensions, shape and specific volume via VolScan Profiler and pore structure via C-Cell) is helpful to clearly define the quality targets, identify deviations at an early stage and describe quality concerns precisely
To describe the problem accurately, it is important to gather both relevant data and reports from production and quality control staff. Personnel in daily contact with the product will have developed a high level of personal expertise over time and may be able to contribute very valuable information
Decide how to deal with the quality problem as long as it remains unsolved
It must predominantly be decided whether the product(s) can still be put on the market or not. This may imply negotiation with customers, for instance, to agree on a modified specification for a limited amount of time. If the decision is made to stop deliveries, customers need to be informed appropriately and, if necessary, the product recalled.
Find out what has changed upfront
Even though nothing has been changed intentionally, the fact that a quality concern occurs implies that something has changed. There are basically two possibilities: either one of the raw materials has modified properties, or something has changed in the process.
On the raw materials side, analytical values need to be checked, showing possible deviations. Even if the values are still within the specified range, an unusual change of values may cause unexpected effects. Hence, in most cases, specifications are agreed upon without testing the full range of variations. It must also be kept in mind that specifications and the corresponding analytical certificates only imperfectly describe the quality of the raw materials. Some relevant properties may just not be covered by the specification and not measured. It can be meaningful to talk with suppliers about whether any changes have taken place in their facilities.
On the process side, possible deviations of process parameters need to be checked, as well as a change in personnel behavior (new staff or taking on incorrect practices). Modified environmental conditions, especially the temperature of production and storage areas, may also cause modified product properties.
The assessment of changes having occurred in raw materials and in processing can be performed from two points of view: on the one hand, based on an existing representation of the factors that may have an influence on the concerned product property. On the other hand, all directions should be evaluated, with the idea that any change of any kind may have an impact on the product. The first approach has the advantage of being more focused, while the second is often helpful in determining causal relationships that have not been explored thus far.
If the proposed investigations have allowed identifying a change that clearly appears to be responsible for the problem, the next step will be to try to remedy this change, which will solve the issue. Nevertheless, it may happen that it is not possible to remedy the change. For example, raw material properties may have changed as a consequence of climatic events, and it might not be possible to source the raw material at its original quality anymore.
In addition, it may happen that the analysis of changes has revealed only several possibilities but no single clear cause (keeping in mind that the cause may be multifactorial). In the worst-case scenario, the analysis of changes may have revealed no possible cause at all.
Reproduce the problem on a pilot plant scale and test potential mitigation approaches
In all of these cases, it may be worth going to the lab to develop a better understanding of the effect of different factors on product quality and test mitigation strategies. Testing different scenarios on a pilot plant basis is much faster and cheaper than testing on the industrial production scale.
Different hypotheses may be tested alone and in combination, both recipe and process-wise.
The solution may be either to change one single cause back to its original state, or to redevelop the product. The latter applies especially if it is not possible to restore the original conditions.
Develop a deeper understanding of causal relationships
Pilot plant trials may be performed on a trial-and-error basis, which in some cases can be the fastest way to solve the issue.
However, with a view to preventing the reoccurrence of similar quality problems, it is of fundamental importance to develop a deep understanding of the factors affecting product quality. This involves in particular the following actions:
+ Be aware of the current state of relevant research, and follow new scientific publications
+ Use diverse analytical techniques to make a comprehensive description of relationships between raw material quality, process conditions and product properties (Figure 3). This may involve physical, chemical, microbiological and microscopic characterization as well as sensory assessment.
+ Implement data management systems for systematic documentation of analytical values and process conditions. Make correlation analysis on broad data sets to understand the effect.
+ Make written documentation of the representations of causal relationships that are present within the company. This allows knowledge to be captured, especially in the case of staff fluctuations. In addition, these representations need to be challenged and improved over time in the light of new observations made.
Figure 3. For understanding complex causal relationships, it is valuable to characterize the product matrix with a variety of analytical techniques. Confocal microscopy can be one of them
Case study: Off-flavors in pretzel buns
To illustrate the approach described above, we are presenting a project on which we worked a few months ago. A German artisan bakery faced a quality concern with its pretzel buns (Figure 4). The production process involved kneading, weighing and forming dough pieces, freezing the dough pieces, and frozen storage for a duration of 1 to 4 weeks, followed by dipping them in an alkaline bath, thawing, proofing and baking.
The company observed that if the dough pieces were stored for more than two weeks in the frozen chamber, an off-flavor occurred in the final products, which led to complaints. The reason for the quality problem was unclear, which motivated the company to ask ttz Bremerhaven for support.
At first glance, the problem seemed to be quite unusual and surprising, as the frozen products were perfectly fine after one week of frozen storage, and quality changes are not expected to take place within a few weeks in a frozen product.
Figure 4: Off-flavor check of pretzel buns at ttz Bremerhaven
Given the relatively small size and the limited resources of the company, it was of utmost importance to choose a pragmatic approach and to limit the investigations to a reasonable amount. Our first step towards identifying the cause of the problem was a sensory assessment of the concerned products (raw dough and baked pretzel buns). The bakery had not been able to make a clear description of the nature of the off-flavor, and it was therefore first hypothesized whether the problem may have been caused by molds or by the migration of volatile substances from the frozen storage area to the product.
In ttz’s sensory assessment, both the dough and the products were described as metallic, pungent, and rancid. With this in mind, fat degradation was considered to be the most probable cause of the quality concern. In order to verify this hypothesis, fat was extracted from the final product (both samples with and without off-flavor) and two parameters of fat degradation were analyzed: the amount of free fatty acids and the peroxide value. It appeared that, while the amount of free fatty acids was identical, the peroxide value was dramatically higher in the samples with off-flavor (the values were over 5 times higher). This confirmed the hypothesis of fat degradation as a possible cause of the off-flavor.
Nevertheless, it was still unclear why fat degradation occurred, especially under frozen storage conditions. The fact that the off-flavor was already present in the frozen dough indicated that the degradation happened during storage and not in the subsequent phases of proofing and baking. Looking at the recipe and the process, one possible mechanism was hypothesized, although it has, as far as we know, not been described so far in scientific literature. The recipe of the product contains a bakery improver. After asking the manufacturer, we were informed of the enzymes it contains, one of them being a glucose oxidase. Glucose oxidases catalyze the oxidation of glucose to hydrogen peroxide and D-glucono-δ-lactone. Hydrogen peroxide, in turn, is supposed to react with thiol groups of gluten, favoring the formation of disulfide bridges, hence increasing gluten polymerization and strengthening the rheological properties of the dough.
However, it seems conceivable that hydrogen peroxide may also reduce with fat, causing its degradation. ttz asked an enzyme manufacturer about this scenario, who confirmed that this side effect of glucose oxidases has already been observed previously. What is more, as hydrogen peroxide is known to be highly reactive, it appeared plausible that it could react with fat even at -18°C.
As the next steps, ttz Bremerhaven designed and conducted some baking experiments that would help solve the issue. As it was important to solve the problem quickly, it was decided not to focus only on the main hypothesis but also to test other possibilities. The test involved baking different variations of the original recipe at ttz according to the original process of the bakery, storing the dough pieces in the frozen state and baking the products every week. The variations chosen included: (1) original recipe, (2) no margarine, (3 and 4) margarines from other manufacturers, (5) no bakery improver, (6) alternative bakery improver without glucose oxidase, (7) alternative wheat flour.
The sensory tests of the final products showed that a strong off-flavor was observed after 4 weeks of storage in the products 1, 3, 4, and 7. This showed that the problem could be reproduced under pilot plant conditions, and that using another margarine or another wheat flour does not solve the issue. Product number 2, without margarine, showed a light off-flavor, whereas products 5 and 6, without bakery improver or without a glucose oxidase, showed no off-flavor at all. These findings supported the validity of the formulated hypothesis.
To get more confidence in the results, the peroxide value was measured in the final products. It appeared all the products had an increased peroxide value with the exception of products number 5 and 6, which were the only ones with only glucose oxidase. Interestingly, the highest peroxide value was found in product 2, without margarine, which had only a light off-flavor. It was hypothesized that in the absence of margarine, hydrogen peroxide will react only with the fat coming from wheat flour. As the total amount of fat in the product is lower, the degree of degradation may be higher, leading to higher hydrogen peroxide values. Alternatively, even if the degree of degradation is higher, a lower fat concentration in the product can explain a lower sensory perception of the off-flavor.
Based on the results obtained at ttz Bremerhaven, the bakery switched its recipe to an improver without glucose oxidase, which solved the quality problem. From a scientific point of view, further investigations would have been possible to further understand the exact mechanism of off-flavor formation but from an economical perspective, it was sufficient that the problem was solved.
It must be pointed out that in this particular case, the quality concern had a clear single cause, which could be clearly established and remedied with one single change in recipe. This is not always the case, as sometimes reaching the desired product quality is more a question of reaching a balance point between raw material properties, recipe proportions and process conditions. Small deviations from different sides may over time bring the overall system out of balance, without one single cause being responsible. Those cases may be more difficult and more frustrating to deal with: in most cases, it is possible to restore the targeted product quality by adapting the recipe and/or the process, however, there is no guarantee on how long the solution will last, as it is not possible to predict future variations of raw material properties (which may be due to natural fluctuations).
Maintaining quality – a Sisyphean task?
In conclusion, it can be said that food products are complex systems and that it lies in the nature of things that impromptu quality issues may occur. When this happens, a clear issue management strategy needs to be adopted and executed. In the longer term, such events should be used as a chance to increase the company’s expertise. Documenting product knowledge, systematically collecting product and process data, identifying deviations before they get critical, being aware of the state of scientific research and getting all relevant staff involved are keys to getting product quality under control.