Included study
Machine Learning, MRP, and Lean Tools as an Internal Innovation Strategy to Improve Efficiency in a Bottling Company
1Bibliographic record
- Authors
- Haydee Manuelo, Hellyn Ruiz, Martin Saenz-Moron, Anita Straujuma
- Year
- 2025
- Venue
- LACCEI International Multiconference on Entrepreneurship, Innovation and Regional Development (LEIRD)
- Country
- Peru
- Segment
- Soft drinks, bottled water and juices
- Process family
- Filling and packaging line
- Study type
- Before/after intervention
- Extraction scope
- Full text
- Size
- 1 plants · 1 lines · — machines
- Observation period
- 84 days
- Data collection
- manual_logs
- Intervention
Show the setting as described by the study
This format accounts for 94.48% of the company's total production volume, representing a critical point of intervention in its production management.
2Risk of bias
Scored with the rubric of the protocol: 2 low concern, 1 some concerns, 0 high concern.
Total score 11Moderate risk
| Item | Score |
|---|---|
| Is the observed line, machine or plant described well enough to classify its segment, process and equipment?Product, process and equipment are all described. | 2 / 2 |
| Is the observation period long enough and reported?Period reported and at least 30 days of production (or at least 100 recorded events). | 2 / 2 |
| How were loss data collected?Manual logs or maintenance records with unknown completeness. | 1 / 2 |
| Are OEE, loading time and the loss categories defined explicitly?A standard definition is cited but its application is not shown. | 1 / 2 |
| Does the loss breakdown account for the total loss it claims to explain?Only the top causes are reported and the unexplained remainder can be computed. | 1 / 2 |
| Is there evidence of selective reporting of favourable results?Intervention study in which baseline and post-intervention periods differ in length or method. | 1 / 2 |
| Are the reported numbers internally consistent?Small inconsistencies (2 to 10 percentage points) or values that cannot be cross-checked. | 1 / 2 |
| Could conflicts of interest or the source of the data bias the result?Academic or independent study, or conflicts declared and unrelated to the result. | 2 / 2 |
3Extracted data
Each datum with its classification and the verbatim quote that states it.
| Reported cause | Loss category | Metric | Value | Extraction |
|---|---|---|---|---|
—sealing machineShow verbatim quote (Section V.B.3)After the implementation of autonomous maintenance, this value increased to 41.13 hours, representing an absolute improvement of 17.65 hours. | Equipment failure (breakdowns) | Time between events | 41.1 h | Adjudicated |
—sealing machineShow verbatim quote (Section V.B.3)At the beginning of the study, the recorded MTBF was 23.48 hours. | Equipment failure (breakdowns) | Time between events | 23.5 h | Adjudicated |
—Show verbatim quote (Conclusions)The application of the proposed management model increased production efficiency from 75% to 86%, evidencing an 11% improvement that contributed to reducing bottlenecks and optimizing installed capacity. | none | OEE | 86.0 % | Both extractors agreed |
—Show verbatim quote (Section III / Table I)Currently, the average efficiency of the production process is 75% (Table 1) | none | OEE | 75.0 % | Both extractors agreed |
—injection molding machineShow verbatim quote (Table VII)Availability (%) 89% 96% 97.17% | none | Availability | 97.2 % | Both extractors agreed |
—injection molding machineShow verbatim quote (Table VII)Availability (%) 89% 96% 97.17% | none | Availability | 89.0 % | Both extractors agreed |
lack of inputShow verbatim quote (Section III / Figure 1)Figure 1 shows that stoppages due to lack of input represent the main cause of loss, accounting for 49% of the total, followed by defects in the finished product and mechanical failures in the blow molding machines, which together amount to 83% of the total impact. | Organizational or external stops (material or personnel shortage, utilities outage) | Share of lost time | 49.0 % | Single extractor |
—Show verbatim quote (Section III / Figure 1)Figure 1 shows that stoppages due to lack of input represent the main cause of loss, accounting for 49% of the total, followed by defects in the finished product and mechanical failures in the blow molding machines, which together amount to 83% of the total impact. | Not classifiable from the report | Share of lost time | 83.0 % | Single extractor |
reworkShow verbatim quote (Section V.B.2 / Table VI)In the initial process diagnosis, an inefficiency level of 19% was identified, reflecting high variability, rework, and bottlenecks. | Process defects (scrap and rework) | Defect rate | 19.0 % | Single extractor |
reworkShow verbatim quote (Section V.B.2 / Table VI)With the validation of the pilot test, this value was reduced to 13%, representing a relative improvement of 32% compared to the initial condition. | Process defects (scrap and rework) | Defect rate | 13.0 % | Single extractor |
—injection molding machineShow verbatim quote (Section V.B.3)During the development of the pilot test, a machine availability of 97.71% was achieved, exceeding the proposed target by 1.171%, as indicated in [7]. | none | Availability | 97.7 % | Single extractor |
—blow molding machinesShow verbatim quote (Conclusions)Autonomous maintenance raised the availability of blow moulding machines to 97.17% and improved the MTBF of the injection machine by 17.65 hours | none | Availability | 97.2 % | Single extractor |
—Show verbatim quote (Table VI, STANDARDIZED WORK METRIC RESULTS)Standardized Work | Rework Rate (%) | 19% | 12% | 13% | Process defects (scrap and rework) | Defect rate | 19.0 % | Single extractor |
lack of inputShow verbatim quote (Section V.B.1, Validation of MRP with Machine Learning)The average stockout rate in 2024 was 30.47%, significantly affecting production due to supply shortages. | Organizational or external stops (material or personnel shortage, utilities outage) | Share of lost time | 30.5 % | Single extractor |
—Show verbatim quote (Table VI, STANDARDIZED WORK METRIC RESULTS)Standardized Work | Rework Rate (%) | 19% | 12% | 13% | Process defects (scrap and rework) | Defect rate | 13.0 % | Single extractor |
lack of inputShow verbatim quote (Section V.B.1, Validation of MRP with Machine Learning)With the validation of the MRP during the first quarter of 2025, the indicator decreased to 20.60% | Organizational or external stops (material or personnel shortage, utilities outage) | Share of lost time | 20.6 % | Single extractor |
Observation units
- Sealing machine · Capping and seaming · After intervention
- Sealing machine · Capping and seaming · Baseline
- Whole plant · Whole line or unspecified equipment · After intervention
- Whole plant · Whole line or unspecified equipment · Baseline
- Injection molding machine · Forming, moulding and depositing · After intervention
- Injection molding machine · Forming, moulding and depositing · Baseline
- 8.5-liter bottling line · Whole line or unspecified equipment · Baseline
- 8.5-liter bottling line · Whole line or unspecified equipment · After intervention
- Blow molding machines · Forming, moulding and depositing · After intervention

