Leveraging AI for Inventory Enhancement r0.pdf

Leveraging AI for Inventory Enhancement

1. SCM SHU INTERNATIONAL CONFERENCE
Leveraging Artificial Intelligent (AI) for
Inventory Enhancement & Cost Efficiency
in Indonesia’s Oil & Gas Industry
Abdullah Choirul Imam
Sr Analyst Strategic Alliance – PT PHE
Fakultas Matematika dan Ilmu Pengetahuan Alam
Program Studi Magister Fisika
Universitas Indonesia (Jakarta)
Marthen Sarungngu
Analyst Procurement Zona 9 – PT PHI
Sekolah InterdisiplinManajemen & Teknologi
Program Doktor Manajemen Teknologi
Institut Teknologi Sepuluh November (Surabaya)

2. SCM SHU INTERNATIONAL CONFERENCE
Background (Stakeholders)
Dirut Pertamina (
Fax No.: 139/C00000/2020-S4 –
“melakukan optimalisasi material
persediaan, kolaborasi demand untuk
mendapatkan economic scale dan
melakukan efisiensi biaya operasi lainnya”
Dirut PHE (
SPRIN No.: Prin-027/PHE0000/2023-S0 -
”Target optimasi material ex-terminasi
USD 69.5mio untuk itu perlu dilakukan
upaya pemanfaatan yang dikontrol
melalui sistem maupun sosialisasi dan
kampanye potensi penggunaannya”

3. SCM SHU INTERNATIONAL CONFERENCE
Introduction
Background
▪Low material availability led to
delay in operations activity,
increase cost to acquire material
in a fastest time
▪Inaccurate demand planning led
to incorrect material order in
terms of quantity and the
product, high inventory cost due
to overstocking material or
spare parts, etc
▪Lack of technology utilization in
Material Inventory Management
led to inefficiencies, increased
costs, and missed opportunities
for optimization
Material Inventory
Management Overview
The Oil & Gas industry is
characterized by its high
operational complexity and the
significant financial implications of
material shortages. This makes
efficient inventory management
crucial for maintaining operational
stability and cost efficiency.
This research investigates the
potential of Artificial Intelligence
(AI) to transform inventory
management practices in
Indonesia’s Oil & Gas sector. By
incorporating AI, the study aims to
enhance material availability,
improve demand forecasting
accuracy, and optimize overall
supply chain performance.

4. SCM SHU INTERNATIONAL CONFERENCE
Research Objectives & Methodology
Enhance Material Availability
Ensure critical materials are
available when needed, reducing
downtime and operational delays.
Objectives
Improve Demand Forecasting
Accuracy
Utilize AI algorithms to forecast
material demand with higher
precision, minimizing overstock
and stockouts.
Optimize Overall Supply Chain
Performance
Streamline inventory processes to
achieve cost savings and efficiency
improvements.
Time Series
Forecasting
ARIMA SARIMA PROPHET
Usefulness
Score
Intermediate Powerful Powerful
StationarityARIMA is most effective with
stationary data, where statistical
properties such as mean, and
variance are constant over time
SARIMA also requires stationary data
but is specifically designed to
handle data with seasonal
components
Does not require stationarity, can
handle non-linear trends
SeasonalityNot inherently designed to handle
seasonal data, though seasonal
components can be manually
adjusted for
Effectively models data with regular
seasonal patterns (e.g., monthly,
quarterly, yearly)
Highly effective with data exhibiting
multiple seasonal patterns (daily,
weekly, yearly)
ComplexitySuitable for simpler time series data
with linear trends or no seasonality
Suitable for data with complex
seasonal patterns and interactions
Handles complex and irregular
patterns, including holidays and
sudden changes
Use CasesEconomicindicators, financial
marketdata, inventorylevels
Retail sales data, tourism data,
energy consumption
Business metrics, web traffic, social
media activity
Strengths▪Good for short-term forecasting.
▪Can handle data with a single
seasonal pattern if properly
adjusted
▪Captures both short-term and
seasonal trends.
▪Good for data with repeating
seasonal cycles
▪Robust to missing data and
outliers.
▪Intuitive interface and easy
customization.
▪Can incorporate external factors
like holidays and special events
Limitations▪Requires data to be pre-
processed to achieve stationarity.
▪Less effective with data that has
multiple or complex seasonal
patterns
▪Can be computationally intensive,
especially with high-frequency
data.
▪Requires careful parameter
tuning for accurate forecasting
▪May not perform as well with very
short time series.
▪Requires careful selection of
model components to avoid
overfitting
Methodology
*) Traditional replenishment methods vs. AI-driven forecasting models

5. SCM SHU INTERNATIONAL CONFERENCE
Key Findings
Cost Savings
The use of AI in inventory management led to
a notable reduction in costs. The inventory-
to-expense ratio decreased from 17.83% to
13.24%
Forecasting Accuracy
AI models, particularly PROPHET and SARIMA,
significantly outperformed traditional
methods in forecasting material demand.
PROPHET achieved less than 1% error for
project BOM materials, while SARIMA showed
an error rate below 0.8% for spare parts.
Material Availability
The Inventory Service Level (SLA) was
maintained at around 90.23%, slightly
decrease from the previous SLA 92.5%,
ensuring a reliable supply of materials while
optimizing inventory levels.
Case Study 1: Project
BOM Materials (PROPHET)
▪Achieved an average error
rate of less than 1%.
▪Demonstrated superior
forecasting accuracy,
reducing instances of
overstock and shortages.
Case Study 2: Spare Parts
(SARIMA)
▪Error rate below 0.8%,
optimizing the balance
between inventory levels
and operational needs.
▪Significant reduction in
inventory holding costs
while maintaining high
service levels.
Overall Impact
AI models provided a more
accurate and responsive
inventory management
approach, aligning material
availability with actual
demand.
Cost Avoidance
6.1 million USD

6. SCM SHU INTERNATIONAL CONFERENCE
Discussion & Conclusion
Metric Before After Improvement
Quality Potential for overstock or non-stock
situations due to periodic review of
reordering parameters
Reduced instances of overstock and
NIS, real-time adjustments possible
More accurate inventory levels,
reduced excess stock
InventoryService
Level (SLA)
92%-93% 90.23% (maintained within acceptable
range)
Balanced cost efficiency and service
level
Cost Inventory to Total Expense Ratio:
17.83%
Inventory to Total Expense Ratio:
13.24%
4.59% reduction, resulting in
significant cost savings.
InventoryCostper
Barrel
USD 3.25 USD 3.08 Lower cost per barrel, improved cost
efficiency
Delivery 30-60 days required for reordering
parameter review
Real-time review of reordering
parameters
Faster response times, improved
supply chain agility
Safety 8,635 transactions per month led to
higher accident potential
5,128 transactions per month led to
reduce accident potential
Improved safety with reduced
material handling
The integration of AI into material inventory management systems in
the Oil & Gas industry has demonstrated significant benefits. These
include improved forecasting accuracy, cost savings, and better
material availability, which collectively enhance operational efficiency.Conclusion

7. SCM SHU INTERNATIONAL CONFERENCE
Future Research
▪Real-Time Data Integration
Integrating real-time data sources such as
IoT sensors and market data can further
enhance the accuracy and responsiveness of
AI models. This integration will enable more
dynamic adjustments to inventory levels,
reducing the risk of stockouts or
overstocking.
▪Human-AI Collaboration
Combining human expertise with AI's data
processing capabilities can improve
decision-making processes. Human intuition
and experience can provide valuable context
that complements AI's analytical strengths.
▪AI-Driven Risk Management
Using AI to proactively manage risks in
inventory and supply chain operations can
mitigate the impact of disruptions. AI can
predict potential risks and recommend
preventive measures, enhancing supply
chain resilience.
We invite stakeholders in the Oil & Gas
industry to explore these AI-driven
approaches and consider their
implementation to achieve greater
efficiency and cost savings in inventory
management.
Call to Action

8. SCM SHU INTERNATIONAL CONFERENCE
TerimaKasih
Ketulusan Untuk Melayani