Practical PDCA Cycle Case Studies: A Complete Path from Workshop Quality Issues to Systematic Improvement

1. Introduction: A Reality Often Overlooked by Many Enterprises

The PDCA cycle (Plan-Do-Check-Act) is one of the most fundamental and powerful tools in quality management. Proposed by Dr. Deming, it has been embraced as the core engine of continuous improvement by global leaders such as Toyota, Motorola, and General Electric.

However, during my eight years of providing quality consulting to over thirty manufacturing companies, I have discovered a concerning reality: more than 70% of companies claim to use PDCA, but fewer than 15% can truly close the loop.

The three most common forms of "pseudo-PDCA" are:

• Skipping data in the P phase—making countermeasures based on experience or intuition without quantifying the current situation or setting SMART goals;
• Lack of discipline in the D phase—implementing countermeasures without clear timelines, responsible persons, or process checks;
• C and A phases being superficial—making improvements without verifying their effectiveness, and even when verified, not standardizing the successful practices, leading to the recurrence of problems within three months.

This article presents three real improvement cases from different industries, detailing the systematic path of PDCA from "theoretical discussion" to "on-site implementation." Each case is analyzed through the four stages of Plan→Do→Check→Act, helping readers understand the true application of PDCA.

2. Case Study One: PDCA Improvement for Welding Spatter in an Automotive Parts Company

2.1 Background: A Recurring Old Problem

In March 2024, I participated as a quality consultant in an improvement project at an automotive parts company in East China (referred to as "Huari Precision Engineering," a pseudonym). The company supplies chassis welding assemblies to a domestic joint venture brand, with an annual production capacity of about 600,000 sets.

The problem originated from a CO₂ gas shielded welding production line. Since its commissioning, the line has consistently experienced excessive welding spatter. The spatter adheres to the surface of the workpieces and the fixture positioning blocks, leading to three consequences:

• Increased grinding time for subsequent processes (an additional 3-5 minutes per piece);
• Reduced fixture positioning accuracy due to spatter accumulation, causing welding dimensional deviations;
• Monthly rework and scrap losses due to spatter amounting to approximately 87,000 yuan.

More frustratingly, this was the third "rectification" of the line within two years. The previous two improvements were led by process engineers, who adjusted the welding current and changed the brand of the shielding gas. Each improvement temporarily reduced the spatter by 20%-30%, but the problem returned to its original state within two to three months.

The first step of PDCA—facing the facts: On my first visit to the site, I asked the team to ignore historical reports and directly take photos of the current spatter condition on the production line, and trace the spatter defect data from the past three months. The results showed that the spatter problem had never been truly quantified. The previous two "improvements" left no quantifiable data—no baseline data before improvement, no verification data after improvement, and even the conclusion that "spatter was reduced" was based on feelings.

This is a typical example of "pseudo-improvement": actions were taken, but the loop was not closed.

2.2 Plan (Plan) Phase—Nailing Down the Problem with Data

Step 1: Define the problem clearly (Problem Definition).

We convened a cross-departmental team, including welding engineers, shift leaders, quality inspectors, and equipment maintenance personnel. I spent a whole day observing the site and analyzing data, converting the vague feeling of "excessive spatter" into measurable indicators:

Step 2: Investigate the current situation and analyze the causes.

I led the team in using a fishbone diagram to analyze potential causes from the five dimensions of people, machines, materials, methods, and environment, listing 23 possible causes. We then verified each cause using the "on-site, on-object" principle:

• We reviewed 300 sets of welding parameter records and found significant differences in current and voltage between different shifts.
• We measured the actual output current with a clamp meter and discovered that the actual current of one welding machine was 18% higher than the panel display value.
• We inspected three batches of welding wire and found variations in wire feeding stability between batches.
• We observed the operators' techniques and found significant differences between three experienced employees and two new hires.

Through these verifications, we narrowed down the 23 causes to 4 key root causes:

1. Lack of standardized welding parameters—different shifts and operators used different combinations of current and voltage, with no unified WPS (Welding Procedure Specification);
2. Inadequate welding machine calibration—two out of three welding machines had actual outputs deviating from the display values by more than 10%, and the equipment department only calibrated them every six months;
3. Improper storage of welding wire—welding wire was left exposed after opening, leading to moisture absorption and unstable wire feeding;
4. Inconsistent operator techniques—the angle and oscillation amplitude of the welding torch varied among operators.

Step 3: Develop a countermeasure plan.

For each root cause, we formulated specific countermeasures and used the 5W1H format to clearly define the responsible person, timeline, and acceptance criteria:

We also set quantifiable targets: within one week of implementation, the spatter particle density should decrease from 47 particles/piece to ≤15 particles/piece, and stabilize at ≤10 particles/piece within one month.

2.3 Do (Do) Phase—Implementing the Plan Every Day

Week 1 (April 1-7): Basic Rectification.

• April 1: Shift Leader Li used a night shift to clean and set up a dedicated storage cabinet for welding wire. The procurement department urgently purchased an industrial dehumidifier, which was installed the next day. The humidity in the cabinet dropped from 65%RH to 38%RH.
• April 3: Equipment Manager Wang contacted an external calibration agency to urgently calibrate the three welding machines. The results showed that the actual current of Machine 1 was 18% higher than the display value, and Machine 3 was 12% lower. After calibration, the deviation of all welding machines was controlled within 3%.
• According to the new plan, the calibration cycle for welding machines was changed from every six months to once a month, and calibration record cards were posted next to each machine. Operators had to confirm that the calibration label was within its validity period before each shift.

Week 2 (April 8-14): Parameter Standardization.

• Welding Engineer Zhang reviewed the recommended parameter ranges provided by the welding wire supplier and, based on on-site trial welding results, developed three standard WPSs for 3mm, 5mm, and 8mm plate thickness combinations.
• Each WPS specified the upper and lower limits for seven parameters: welding current, arc voltage, welding speed, shielding gas flow, welding wire extension length, etc.
• On April 10, Zhang organized WPS training for all welding operators and posted WPS cards next to each welding machine.

Week 3 (April 15-21): Standardizing Techniques and Practical Assessments.

• Zhang recorded a standard operation video, breaking down the key actions of the welding torch angle (75°±5°), oscillation amplitude (no more than three times the diameter of the welding wire), and welding speed (matched to the plate thickness).
• Each of the 5 operators underwent practical assessments. Two new employees and one experienced employee did not meet the standards and were required to undergo three days of intensive training.

By April 21, all countermeasures were fully implemented.

2.4 Check (Check) Phase—Verifying the Effect with Data

First week (April 22-28) data tracking:

I required that in the first week after implementation, 10 pieces (60 pieces/day) be sampled and inspected per shift, recording the number of spatter particles. The results were as follows:

The data clearly showed two trends: first, the spatter particle count steadily decreased, reaching the target (≤10 particles/piece) by the seventh day; second, the improvement was gradual rather than sudden—operators' proficiency with the standardized WPS and SOP continued to enhance the effect.

One month later (May 22):

• The spatter particle density stabilized between 8-12 particles/piece, with daily fluctuations mainly due to differences in plate thickness.
• Grinding time decreased to 0.8-1.2 minutes/piece, a reduction of about 75% compared to before the improvement.
• The fixture cleaning frequency reduced from 5-6 times/shift to 1-2 times/shift.
• Monthly rework and scrap losses due to spatter decreased from 87,000 yuan to 21,000 yuan.

A key discovery: During the Check phase, we also identified a problem not anticipated in the Plan phase—new WPS parameters resulted in a slightly lower penetration depth for 8mm thick plates than required by the customer. This presented a new challenge for the Act phase.

2.5 Act (Act) Phase—Standardization and Continuous Improvement

The Act phase is the easiest to skip but the most critical step in PDCA. We did three things:

First, standardize successful practices.

• Incorporate WPS into the company's process standard document library and obtain approval from the technical department.
• Write the welding machine calibration process into the "Equipment Management Procedure," making the calibration cycle and record requirements institutional.
• Write the welding wire storage standards into the "Material Management Procedure," and require the procurement department to confirm the supplier's packaging and transportation conditions when new welding wire arrives.

Second, address new problems identified in the Check phase.

For the issue of insufficient penetration depth in 8mm thick plates, we initiated a second round of PDCA (the core of PDCA is "cyclical"—the end of one PDCA is the beginning of the next). Zhang readjusted the WPS parameters for 8mm thick plates, verified them with a small set of test pieces, and only included them in the official documents after confirming the penetration depth met the requirements.

Third, establish a continuous monitoring mechanism.

• Before each shift, operators must sign the WPS confirmation form to verify the parameter settings for the shift.
• Process engineers randomly check spatter data once a week and enter it into the SPC system.
• A monthly improvement results review meeting is held to track the long-term trends of the indicators.

Solidifying the results: As of December 2024, the spatter particle density on the production line had remained stable at ≤10 particles/piece for eight consecutive months, saving an annual amount of approximately 720,000 yuan. More importantly, the team established a habit of "speaking with data and improving step by step," subsequently initiating four more self-improvement projects.

3. Case Study Two: Multiple Rounds of PDCA for SMT Placement Offset in an Electronics Manufacturing Company

3.1 Background: A Troublesome SMT Machine for Three Months

In July 2024, while providing quality system guidance to an electronics contract manufacturing company in the Pearl River Delta (referred to as "Chengda Electronics," a pseudonym), I encountered a typical "chronic issue"—the placement offset defect rate on an SMT production line was as high as 1.2%, far exceeding the industry benchmark (less than 0.3%).

This issue had persisted for three months. Process engineers had successively adjusted the placement pressure, changed the solder paste brand, and optimized the reflow soldering temperature curve—each adjustment temporarily reduced the defect rate, but it rebounded to over 1% within a week.

I recommended that the team use the PDCA method to systematically solve the problem rather than continue with "trial-and-error improvements."

3.2 First Round of PDCA: Identifying the True Dimensions of the Problem

Plan:

Initially, the team believed the main cause of the offset was the insufficient precision of the SMT machine. However, when I asked them to stratify the data, a surprising discovery was made—stratified analysis showed:

• The SMT machine had 12 nozzles, and the defect rate for Nozzle 3 was 4.7%, while the rates for other nozzles were only 0.3%-0.6%.
• Further analysis revealed that Nozzle 3 was responsible for placing 0201-sized components, while other nozzles handled 0402 and larger components.

Root cause analysis pointed to: insufficient vacuum pressure in Nozzle 3. Measurement results showed that the vacuum pressure in Nozzle 3 was 22% lower than the standard value, causing small components to easily shift during placement.

Do: Replace the vacuum seal ring of Nozzle 3 and recalibrate the vacuum pressure.

Check: After the replacement, the defect rate for Nozzle 3 decreased from 4.7% to 0.5%. However, the overall defect rate for the line only dropped from 1.2% to 0.8%, still far from the target of 0.3%.

Act: The problem with Nozzle 3 was resolved, but the overall defect rate still did not meet the target, indicating that other root causes had not been identified. The team entered the second round of PDCA.

3.3 Second Round of PDCA: Uncovering Systemic Factors at the PCB Level

Plan:

During the Check phase of the first round of PDCA, the team discovered an abnormal pattern—the defect rate for placement offset was significantly higher during Monday morning shifts and Friday evening shifts compared to other times. This clue pointed to environmental conditions and operational consistency.

Further investigation revealed two issues:

• The workshop air conditioning system was turned off during weekends, causing the temperature to reach 31°C and humidity to reach 75%RH on Monday mornings, far exceeding the recommended temperature range of 23±3°C and humidity range of 40-60%RH provided by the solder paste supplier.
• Friday evening shifts were handled by substitute operators, who performed material changes improperly, causing the component reels in the feeders to loosen and affecting the pick-up accuracy.

Do:

• Modify the workshop air conditioning management system to ensure that the temperature and humidity remain within process requirements during weekends.
• Develop a standard material change operation video and train all operators, including substitutes.
• Add a material change confirmation procedure to each SMT machine, requiring operators to scan and confirm after changing materials.

Check:

Two weeks after the implementation, the overall placement offset defect rate for the line dropped to 0.35%. The peak defect rates during Monday morning and Friday evening shifts were eliminated, and data across all shifts became more consistent.

Act:

• Incorporate temperature and humidity management requirements into the "SMT Workshop Environment Management Regulations."
• Include material change operations in the monthly skills assessment for operators.
• Establish an abnormal data warning mechanism—when the defect rate in a particular shift exceeds 1.5 times the average, the system automatically sends a warning notification.

3.4 Third Round of PDCA: From Single-Point Improvement to Systemic Prevention

After solving the above two issues, the team did not stop. In the Act phase, they reflected on a deeper question: "Why did it take three months to truly solve this problem?"

The answer pointed to a systemic issue: an incomplete problem classification response mechanism. Initially, the placement offset was considered a "minor issue" and handled by process engineers without escalating to a cross-departmental team or initiating a systematic PDCA process. By the time the problem worsened and caught the attention of management, significant waste had already been generated.

To address this, the team established a problem level matrix:

After implementing this mechanism, the problem response speed in the SMT workshop significantly improved: the average resolution time for issues below L3 was reduced from 23 days to 6 days.

4. Case Study Three: PDCA in Action for OEE Improvement in a Food Company

4.1 Background: An Invisible Efficiency Loss

At the beginning of 2025, I guided a continuous improvement project at a food processing company (referred to as "Weiyuan Food," a pseudonym). The company's filling production line had a long-term Overall Equipment Effectiveness (OEE) of 62%-65%, far below the industry benchmark of 85%.

Interestingly, management believed that the low OEE was due to "aging equipment" (which had been in operation for 8 years) and planned to invest 6 million yuan to replace the equipment. I suggested: first, use PDCA to identify the true sources of efficiency loss before making investment decisions.

4.2 Plan: Exposing the Truth with Data

I led the team in a detailed time study of the filling line over two weeks, recording every minute of each 8-hour shift. The final data shocked everyone:

Key discovery: equipment failure accounted for only 3.8%, while "waiting for materials" (9.8%) and "speed loss" (10.6%) were the largest sources of loss. Replacing the equipment would not solve these problems.

Further root cause analysis revealed:

1. Waiting for materials—the sterilization process in the upstream operation had unstable batch cycles, often causing the filling line to wait for materials.
2. Speed loss—operators reduced the filling speed to "ensure no quality issues."
3. Long changeover times—the CIP (Clean-In-Place) procedure during product changeovers lacked standard times, leading to loose operator schedules.

4.3 Do: Simultaneous Efforts in Three Directions

The team formed three PDCA groups to work in parallel:

First Group (Material Flow Improvement):

• Set up a 30-minute buffer tank between the upstream sterilization process and the filling line.
• Develop production rhythm standards for the upstream process to align with the downstream process.

Second Group (Speed Optimization):

• Use the DOE method to find the relationship curve between filling speed and filling accuracy, proving that a 15% speed increase would not affect filling accuracy.
• Develop standard filling speeds and monitor the deviation between actual and standard speeds daily.

Third Group (Standardized Changeover):

• Record each step of the CIP cleaning procedure in a video and establish standard times.
• Develop a changeover checklist to compress the preparation, execution, and verification times by 40%.

4.4 Check: Data Verification

One month after implementation:

More importantly: Based on these data, management canceled the 6 million yuan plan to replace the equipment and instead invested 800,000 yuan in upgrading the existing equipment with automation, achieving a return on investment in less than 8 months.

4.5 Act: From "Project" to "System"

This PDCA improvement not only solved the OEE issue but also led to the establishment of three systems:

1. Daily OEE monitoring mechanism—record and update OEE data and the distribution of the six major losses on the management board after each shift.
2. Weekly improvement meeting system—use OEE data as input to identify the largest loss source for the week and initiate a rapid PDCA.
3. PDCA analysis required before investment decisions—any equipment investment exceeding 500,000 yuan must first undergo PDCA analysis to confirm the improvement potential of existing equipment.

5. Seven Key Principles for Successful PDCA Implementation

Through the above three cases, I have distilled seven key principles for successfully implementing PDCA in practice:

Principle 1: Time investment in the P phase determines 80% of the success rate. Most PDCA failures are due to a hasty Plan phase—problems are not quantified, root causes are not verified, and countermeasures lack clear acceptance criteria. In the improvement projects I have guided, teams that spent sufficient time in the Plan phase (usually more than 40% of the entire PDCA cycle) had a success rate of over 90%; conversely, the success rate was less than 30%.

Principle 2: Let data, not feelings, define the problem. "Excessive spatter," "low OEE," "high defect rate"—these are feelings, not data. A good problem definition must include: current value, target value, measurement method, and data source.

Principle 3: Use stratification to find the "structural" aspects of the problem. In Case Study Two, if the team had not stratified the nozzles, they might never have found the true root cause. Stratification is the core method for exhausting root causes and avoiding "treating symptoms rather than the root cause."

Principle 4: Countermeasures must be verifiable. Each countermeasure must have clear acceptance criteria and verification methods. Do not say "strengthen training"—say "all operators pass the practical assessment, with a pass rate of ≥90 points."

Principle 5: The Check phase also requires data, and it must be from the same source as the P phase. Many teams put in a lot of effort to collect data in the P phase but only use "it looks good" to replace data verification in the C phase. This is fatal.

Principle 6: The Act phase must "land"—either standardize or initiate the next PDCA. If improvement results are not standardized, the problem will recur; if new problems discovered during improvement are not followed up, the improvement will be incomplete. PDCA is called a "cycle"—there is no end, only continuous improvement.

Principle 7: PDCA is not a tool, but a management culture. The best companies I have seen are not those that use PDCA in the most "fancy" way, but those that have made PDCA a "muscle memory"—reviewing data at daily meetings, using PDCA to debrief at weekly meetings, and ensuring every problem has a closed loop. When PDCA becomes the underlying operating system of daily work, continuous improvement no longer requires additional push.

6. Conclusion: From "Knowing How to Do" to "Getting It Done"

The PDCA cycle may seem simple—four letters, eight words. However, as demonstrated by the three cases in this article, the gap between "knowing how to do" and "getting it done" is a comprehensive mindset and organizational capability.

The team at Huari Precision Engineering learned to "speak with data" in their first PDCA practice and subsequently initiated more improvements; the engineers at Chengda Electronics shifted from "single-point trial and error" to "systematic analysis," establishing a problem level matrix; the management at Weiyuan Food avoided a 6 million yuan ineffective investment through PDCA. These are not exceptions but natural outcomes of serious PDCA execution.

For every quality manager, my advice is: do not aim to get all PDCA steps right in one go, but ensure that each time you are closer to closing the loop. True continuous improvement does not come from a perfect PDCA but from the accumulation of numerous PDCA cycles—each cycle builds on the previous one, improving a little bit each time. Over the years, this will lead to a qualitative leap.

A poorly executed PDCA is not a problem with PDCA itself, but with not closing the loop. Once the loop is closed, the results will naturally follow.

Knowledge code: 5.2.1

Version: v20260726

Author: Quality Think Tank Quality Think Tank is dedicated to providing systematic knowledge, methodologies, and practical tools for quality management professionals, helping companies continuously improve their quality capabilities.