Across 161 production shifts recorded in December 2025, the fleet-wide mean OEE is 47.4% — well below the industry benchmark of 65% for pharmaceutical blister/sachet packaging. This means that on average, equipment is delivering productive output during less than half of its scheduled time. The three OEE pillars tell a clear story: Availability averages 80.2% (acceptable), Quality averages 94.4% (strong), but Performance averages 62.8% — the critical gap that is suppressing overall effectiveness across every machine in the fleet. In practical terms, the machines are running when scheduled and producing good product, but they are running far slower than their rated speeds, leaving enormous untapped capacity on the table every shift.

Breaking OEE down by machine reveals significant variation across the fleet. Hoong-A achieves the highest mean OEE at 49.5%, driven by a strong Availability of 86.4%, though its Performance at 65.3% still leaves room for improvement. CP4 C is close behind at 49.4% with the best Quality score in the fleet at 92.8%. At the other end, CP4 B records the lowest mean OEE at 46.6%, largely because its Performance factor drops to 61.3% — machines are running but not at rated speed. IMA C80B, the high-speed blister line running Glumet DC at 320 BPM rated speed, records mean OEE of 46.2% with its Performance at only 66.7% despite processing the largest batch volumes in the fleet, suggesting the speed gap represents thousands of blisters of lost output per shift.

To understand where time is actually being lost, the OEE waterfall translates the three component gaps into percentage points of lost theoretical output. Starting from 100% theoretical capacity: Availability losses (unplanned downtime, breakdowns, changeovers) consume approximately 19.8 percentage points, reducing usable time to 80.2%. Performance losses then consume a further 27.4 percentage points of the remaining time — this is the largest single loss category, meaning machines are running at 62.8% of their rated speed on average. Finally, Quality losses take a further 5.4 percentage points, arriving at the fleet OEE of 47.4%. The data confirms that closing the speed gap is the highest-leverage intervention available to the production team.

Examining the recorded downtime reasons across all shifts, the data reveals that Material Top-up (T) is the single most frequent planned downtime activity, appearing in 138 of 161 shifts with a mean duration of 91 minutes per occurrence. This is followed by MC Cleaning/Preparation (mean 71 min, 127 shifts) and Rest breaks (R) which are scheduled but consistently consume 60–90 minutes. Breakdown time (B) appears in 45 shifts with a mean of 42 minutes — relatively infrequent but concentrated on specific machines, particularly IMA C80B where breakdown-induced restarts result in compounded performance losses as the machine ramps back to speed. Support Other Room (SP) and Waiting for QC/PM account for significant unplanned losses on CP4 A and CP4 C, reflecting cross-functional workflow bottlenecks rather than equipment issues.

Performance factor is calculated as Actual Speed ÷ Rated Speed, capturing how close machines run to their nameplate capacity. Analysing by product family reveals that Glumet DC Tablet 500mg (IMA C80B at 320 BPM rated) achieves mean Performance of only 66.7%, meaning the line runs on average at approximately 213 BPM rather than 320 BPM — a gap of 107 blisters per minute. Over a 10-hour operating shift, this gap represents roughly 64,200 missed blisters per shift. Covinace 4mg (Hoong-A at 80 BPM rated) similarly runs at 65.3% Performance (≈52 BPM actual), while Simvastatin 40mg is the best performer at 69.1% average Performance. Products with frequent batch changeovers — particularly Cetirizine 10mg and Loratadine 10mg — show Performance dips concentrated in the first hour of each batch as the machine reaches operating temperature and speed.

Comparing Morning and Night shifts across the full dataset reveals a consistent pattern: Night shifts achieve mean OEE of 48.3% versus 46.5% for Morning shifts. This 1.8 percentage-point gap is statistically consistent across most machines. The likely drivers are that Night shifts avoid the downtime overhead of shift handovers, supervisor sign-offs, and QC result waiting that cluster in the morning period. Availability is 1.6 points higher on Night shifts (80.9% vs 79.3%) while Performance is 0.9 points higher (63.2% vs 62.3%). Notably, Quality is essentially identical between shifts (94.3% Night vs 94.5% Morning), confirming that reject rates are product- and process-driven rather than operator-dependent. The practical implication is that Morning shift start-up losses — the first 30–45 minutes of each shift — represent a disproportionate drag on Availability that Night shifts avoid by running through.

The analysis points to concrete, high-leverage interventions that do not require capital investment — they require operational discipline and process standardisation. The four actions below are sequenced by expected impact and implementation speed, drawing directly from the loss patterns identified in this dataset.