Anyone who has conducted factory audits knows a key detail: walk into two factories that both hold certifications and look at their SPC control charts. In one, the data points are scattered across the entire tolerance range, almost filling the space from the lower specification limit to the upper limit. In the other, the data points are tightly clustered, steadily hugging the target value.Both are “qualified.” But if you were the brand owner, which one would you choose?Passing final inspection only means that batch didn’t cross the line; stable data indicates the factory truly has its process under control.
Read Before You Start · Three Key Terms
1.SPC (Statistical Process Control)Like taking a factory’s temperature log—it’s not just asking “Is this batch qualified?” but plotting historical inspection data over time into a curve, observing trends and fluctuations to judge whether the production process is stable and in control.
2.σ (Standard Deviation)An indicator measuring the amplitude of data fluctuation. The smaller the σ, the smaller the variation between batches, and the more stable the process.
3.Cpk (Process Capability Index)Understand it using driving as an analogy: the specification limits are the lane boundaries, σ is the natural left-right sway of your driving, and Cpk is how much margin remains between your sway range and the boundaries. (Higher Cpk means more margin, lower probability of problems.)
I. Specification Limits, Control Limits, and Process Capability—Three Levels That Should Not Be ConfusedLet’s start with two often-confused concepts: specification limits and control limits.Specification limitsare the bottom lines drawn by regulations or standards. For example, a certain type of dry food requires a protein standard of ≥36%. Below this number, the product cannot leave the factory.Control limitsare the production targets set by the factory itself, based on actual production data. A reliable factory’s control limits will always be narrower than the specification limits—it doesn’t produce by scraping the bottom line, but manages to produce stable products well within that line.The most commonly used indicator to measure the relationship between them is the Process Capability Index Cpk:
Process Capability Index Cpk · The Core Indicator Measuring Process Control CapabilityCpk = min [ (Upper Specification Limit − Process Mean) ÷ 3σ, (Process Mean − Lower Specification Limit) ÷ 3σ ]No need to memorize the formula; just remember this meaning: the higher the Cpk, the further the actual process variation is from the specification boundaries, and the lower the probability of problems. The formula uses the actual production mean μ—if the production mean deviates from the target, even if the variation amplitude doesn’t change, Cpk will decrease.
≈ 1.0 or lower: Process variation has already consumed most of the specification tolerance. Any slight disturbance can easily cause problems.
≥ 1.33: Passing the baseline. When the process is centered, process variation occupies about 75% of the specification tolerance. Batch-to-batch variation enters an acceptable range.
≥ 1.67: Has sufficient safety margin—but this is contingent on the measurement method itself being reliable.
Note: Cmk assesses the repeatability precision of equipment under short-term stable conditions; Cpk is the comprehensive capability after factoring in personnel, raw materials, environment, and all other elements. Equipment Cmk is the upper limit of Cpk—if the equipment itself isn’t precise enough, there’s a ceiling to process-level optimization.
Comparison of Actual Protein Control Ranges (Same Specification Requirement: ≥36%)
“Qualified” only means crossing the bottom line. A high Cpk means the factory truly has its process under control.
II. Four Core Indicators: Where Does Stability Go Out of Control?
01 Protein · Is the Protein Content the Same in Every Bag You Buy?Protein fluctuation is the result of three overlapping factors: unstable raw material batches (it’s common for fish meal from the same supplier to differ by several percentage points in crude protein between batches; poorly managed suppliers may have relative differences up to 8–10%), errors in formula execution/weighing, and processing losses.
The practice of high-level factories is: first test the protein content of each incoming raw material batch. If the measured value is, say, 1 point lower than the labeled value, the formula must compensate—what to use and how much is determined by a clear calculation logic. This closed loop of “incoming material inspection → dynamic formula adjustment” is the core of protein stability.
Another easily overlooked point: the precision of the weighing equipment itself. Weighing sensors have inherent percentage-level errors. When multiple ingredients are weighed cumulatively, errors accumulate. Industrial weighing systems like Mettler Toledo have standard calibration procedures, but establishing a regular calibration mechanism and retaining records is essential to ensure the baseline doesn’t drift. Liquid raw materials—fats, meat slurries, water—rely on stable output from flow meters; once a flow meter fluctuates, the actual amount added per batch will show systematic deviation (industrial flow meters from companies like E+H have clear engineering specifications for stability).
Equipment is the foundation. If the foundation is unstable, even the most precise formula adjustments will be negated.
Key Actions: Link measured values of incoming materials to formula parameters, triggering dynamic fine-tuning; regularly calibrate weighing systems and flow meters, retaining records; use SPC to track batch trends of finished product protein.
02 Moisture & Aw · It’s Not Just About “Moisture Below 10%”Many people understand moisture control simply as drying to a certain moisture value. The problem is: Aw (Water Activity) is the real indicator determining microbial risk, and Aw is not a simple function of moisture content.
Even at the same 10% moisture content, the Aw of different formulas can differ by more than 0.05. This 0.05 is significant—below an Aw of 0.60, most microorganisms struggle to grow; above 0.70, the risk of mold and yeast increases noticeably. Between these two numbers, the growth state of microorganisms is completely different.
It gets more complex: as pellets exit the extruder and go through drying, cooling, transfer, and oil coating, moisture migration can occur at each stage. The endpoint moisture value is not equal to the Aw when the consumer opens the bag.
Factories that take this seriously will establish separate moisture-Aw relationship curves for different formulas. The control target at the drying endpoint is not a fixed moisture value, but a predicted Aw target. This requires accumulating substantial historical data and is a direct reflection of systemic capability.
Key Control Points: Segmented drying curve settings + post-cooling pellet temperature uniformity + pre-packaging Aw verification—these three together determine the microbial safety over the shelf life.
03 Fat Oxidation & Acid Value/POV · The Silent Threat on the ShelfWhen consumer feedback says “my dog won’t eat this batch of food,” often the problem lies in the fat, not the formula.
Most factories test the acid value at the time of production but don’t perform shelf-life prediction. The issue is: fat oxidation has an “induction period”—it may look normal at production, but oxidation is an accelerating process, and the state by the end of the shelf life can be completely different.
Another point: acid value mainly reflects hydrolytic rancidity. The “rancid” odor consumers smell often comes more from secondary oxidation products. For high-fat, high-fish-oil products, relying solely on acid value is insufficient; Peroxide Value (POV) should be tested simultaneously, and sensory evaluation should be conducted when necessary.
Tracing back to the source, fat quality control is a chain: testing the freshness of incoming oil/fat raw materials → storage temperature and antioxidant management → precise control of coating application → packaging material barrier properties and residual oxygen in the package. If any link weakens, products with the same acid value at production can end up in completely different oxidation states in the consumer’s hands.
Recommended Actions: For high-fat products, include both acid value and POV in final inspection; establish shelf-life oxidation trend tracking with retained samples to predict end-of-shelf-life quality.
04 Pellet Density & Expansion · It Looks Like Appearance, But It’s About ProcessFor the same food, one batch has loose, light pellets, another has dense, heavy pellets—it’s not just an appearance difference; it’s a sign of process fluctuation.
The bulk density of pellets affects packaging accuracy. If the packaging machine fills by weight, a change in bulk density won’t directly affect net weight proportionally, but it will affect the feed rate and cut-off stability, easily causing weight over/under deviations on high-speed packaging lines.
Controlling expansion is essentially controlling the stability of SME (Specific Mechanical Energy input). SME determines the degree of starch gelatinization, which in turn affects pellet density and structure. Many factors influence SME: screw speed, material fill ratio, feed rate, material moisture, screw torque, barrel pressure, starch characteristics of the formula… Continuously tracking SME as an online indicator is common practice in foreign-owned factories but is not yet widespread domestically.
Troubleshooting Logic: If there is a noticeable change in pellet appearance or hardness between batches, first check SME, then investigate screw configuration, conditioning parameters, feed rate, and material moisture. This is more efficient than directly disassembling equipment to find the cause.
III. Why Are Control Limits Hard to Narrow?
① Equipment Capability Cm/CmkIs the tool itself precise enough?Before analyzing process capability (Cpk), has the equipment capability (Cmk) been assessed? This step is often skipped.
Cmk is the repeatability precision of equipment under short-term stable conditions; Cpk is the comprehensive capability after factoring in personnel, raw materials, environment, and all other elements. Equipment Cmk is the upper limit of Cpk—if the equipment itself lacks precision (Cmk < 1.67), no matter how much you optimize at the process level, it’s difficult to maintain high process capability long-term.
Many factories calculate Cpk and think it looks good—that’s often because the specification tolerance is wide enough, not because the process is truly good.
② Measurement System Analysis (MSA)Is the data you’re seeing real?The fluctuation on a control chart comes partly from the process and partly from the error of the measurement itself. If the repeatability and reproducibility (Gage R&R) of the inspection method accounts for more than 30% of the total variation, many “abnormal signals” on the control chart are actually measurement noise.
In this scenario, adjusting the process based on the control chart is chasing a non-existent problem—the more frequent the adjustments, the more systemic fluctuation is introduced.
For key indicators like moisture, protein, fat, Aw, and acid value, first perform a Measurement System Analysis to confirm the measurement itself is stable and reliable, then talk about SPC. This sequence is often skipped, but it’s the foundation of the entire system.
③ Supplier Quality Assurance (SQA)If the source is unstable, the endpoint won’t be stable.For the same fish meal from the same supplier and of the same type, it’s normal for crude protein to differ by several percentage points between batches. In extreme cases with poorly managed suppliers, the relative difference can approach 8%, and moisture can differ by 2 percentage points.
Relying on stricter final inspection is just setting up a defense at the exit—variation has already entered with the raw materials. The real solution is to push upstream: quantitatively assess the batch consistency of key suppliers, drive supplier improvement, and eliminate the root cause of instability before it even enters the factory.
IV. About SPC: Different Indicators, Different StrategiesSPC isn’t about plotting control charts for all indicators; it’s about designing sampling frequency and response rules based on the nature of each indicator.
Moisture · Bulk Density · Package WeightReal-time indicators that can be measured frequently, suitable for online or near-line SPC, directly reflecting process state.
Protein · FatIndicators with longer testing cycles, more suitable for batch trend charts and monthly process capability analysis.
Acid Value / POVBest viewed in conjunction with retained sample tracking to monitor shelf-life oxidation trends and end-of-shelf-life risks.
Water Activity (Aw)Primarily batch sampling, with increased frequency for high-risk products, establishing correlations with the formula.
SPC done for its own sake just accumulates piles of data no one looks at. Truly effective SPC applies measurement resources to the places that most accurately reflect the process state.
Summary: The Three Levels Affecting CpkA stable Cpk isn’t determined by a single factor; it’s the result of three interlocking levels:
① Measurement System (The Foundation)If measurement results are inaccurate, Cpk is a meaningless number. If Gage R&R exceeds 30% of total variation, most signals on the control chart are measurement noise. Adjusting the process based on these signals will only make it more chaotic. Key instruments like moisture analyzers, NIR, scales, and Aw meters must be calibrated regularly to prevent baseline drift.
② Equipment Capability (The Hardware Ceiling)Cmk is the ceiling for Cpk. Weighing sensors have inherent percentage-level errors; errors accumulate when multiple ingredients are weighed cumulatively. Unstable output from liquid flow meters creates systematic deviations in the actual amount added per batch. If equipment capability is subpar, process-level optimization has no solid ground to stand on.
③ Process Variation (The Process Layer)Both variables in Cpk—the mean μ and the variation σ—are directly influenced by multiple process factors：
| Influencing Factor | Mechanism of Action |
| Raw Material Batch Consistency | Source variation directly amplifies the final σ (standard deviation). |
| Formula Execution Accuracy | Accumulation of measurement errors affects the shift of the mean μ. |
| Incoming Material Inspection + Dynamic Adjustment | Determines whether μ can be maintained near the target value. |
| Extrusion Process (SME) Stability | Fluctuations in screw speed, fill ratio, material moisture, etc., affect pellet density σ. |
| Drying / Cooling Process | Moisture migration pathways affect the final Aw (water activity) and moisture Cpk. |
| Fat Coating & Storage | Batch-to-batch fluctuations in acid value / POV (Peroxide Value) are influenced by the superposition of multiple factors. |
Reliable measurement system → Equipment capability meets standards → Raw material and process variation are controllable. Only when all three layers are in place does Cpk have meaning. If any layer is missing, SPC control charts either generate false signals or simply fail to track the fluctuations.
V. Industry Reality and the Path Forward
The density of certifications is rapidly increasing, but the improvement in certification levels and manufacturing capabilities is not synchronized. Many factories have obtained certifications, yet in process control, they still rely on the experiential judgment of veteran operators rather than data and statistical methods.
This gap will be amplified by market forces. Consumer upgrades, rising demands from brand owners, and intensifying competition among contract manufacturers are turning process stability from a “bonus point” into a “basic requirement.”
Factories that can present Cpk data, demonstrate batch trend tracking, and clearly explain why a specific batch’s indicators deviated will have a true competitive edge in the next round of competition. Those who cannot explain will ultimately have to compete on price alone.
To assess a pet food factory’s manufacturing capability, more telling than its certification certificates are its Cpk value, the appearance of its SPC charts, and its ability to clearly explain why a particular batch’s moisture content was 0.3% higher than the previous one’s.
These are not mere formalities. They are direct proof of whether a factory has a genuine understanding of its own processes.
The moat lies not in celebrity endorsements or fancy packaging, but in the control limits.
Phone:+86 130 7377 0763
Email: enquiry@mufengzhineng.com
Official website: www.mufengzhineng.com
