Case study

Finding the Best Locations for 10,000 New LPG Centers

Part 1 of how we powered one of India's fastest-executed schemes


As part of the Ujjwala Yojana scheme, the Ministry of Petroleum and Natural Gas wanted to open 10,000 new LPG distribution centers. They partnered with us to find locations with the greatest impact and reach people who needed clean cooking fuel the most.


Ministry of Petroleum and Natural Gas

Bharat Gas

Indian Oil

Hindustan Petroleum







new distribution centers

6.5 mil.

data points merged


geocoordinates collected


centers commissioned

The Background

Bringing clean cooking fuel to 80 million women below the poverty line

In May 2016, the Ministry of Petroleum and Natural Gas launched Pradhan Mantri Ujjwala Yojana, its flagship scheme to ensure access to clean cooking fuel. The scheme aims to provide free LPG (liquified petroleum gas) connections to 80 million women under the poverty line by 2020.

The Ministry of Petroleum and Natural Gas partnered with us to use data to carry out the scheme effectively. We focused on three aspects of the scheme; first, we tackled the problem of where to open new LPG centers.

Learn more

This case study was featured at TEDx Gateway, one of Asia's largest TEDx events

The Problem

Opening 10,000 new LPG distribution centers

In India, women usually cook on a chulha (clay cook stove) because the nearest LPG center is just too far away. To solve this, the Ministry of Petroleum and Natural Gas partnered with India's 3 oil marketing companies to open 10,000 new LPG distribution centers.

The Ministry's goal was to reach the people who need connections most, rather than opening centers in India's wealthy or urban areas. How could they quickly figure out which locations would reach the most people in need and have the greatest impact?

The Solution

Using data to maximize each center's impact

The Ministry of Petroleum and Natural Gas partnered with us to plan for 10,000 new LPG distribution centers. We used 6.4 million diverse data points to figure out where to open up each new LPG center so it would have the greatest impact on improving access to clean cooking fuel.

How we brought this dream project to life

1. Merging sales data
1. Merging sales data

We merged sales data for each of India's 17,000 existing distributor centers.

2. Sourcing internal data
2. Sourcing internal data

We layered external data with data from our internal repository of 600 external data sources.

3. Collecting geo-locations
3. Collecting geo-locations

17,000 distributors downloaded our app to submit the geocoordinates of their LPG centers.

4. Data merging
4. Data merging

Since all the data came from different systems, it had to be processed, matched, and geocoded.

5. Data analysis
5. Data analysis

Our data scientists created an algorithm to cluster villages and rank potential LPG locations.

6. Data visualization
6. Data visualization

Officials used our interactive dashboard to find the best location for every new LPG center.

If we plan and implement a scheme based on mere guesswork and estimates, then certain gaps will show in our decision-making. With the use of numbers, data and technology, we can make important decisions in a matter of minutes. Data is going to become a primary medium for governance in India.

Dharmendra Pradhan
Dharmendra Pradhan

Cabinet Minister for Petroleum & Natural Gas

Government of India

Merging sales data from 3 oil marketing companies

The first step in getting all the data we needed was to understand what existed currently. We extracted, merged, and geocoded data about sales, number of consumers and supply, and more from the scattered internal data systems of India's 3 oil companies.

Sourcing external data

Since sales data only cares about profitability, we knew we had to add external data so we could place centers in the places with the greatest impact. We used public data to map population, affluence, infrastructure, LPG penetration, and a host of other data points.

Collecting 17,000 geocoordinates in 1 month

When we thought we had all the data we needed, we realized we didn’t actually know the location of the existing LPG centers. Within a month, 17,000 distributors downloaded our app to submit their location coordinates from remote corners of the country.

Learn more about Collect
1. View existing distribution centers
2. Identify villages without nearby LPG centers
3. Highlight villages with the greatest market potential
4. Find the best location for each center

Finding the best location for each new LPG center

  • Our platform mapped the 17,000 existing distribution centers as the base for our analysis.

  • Next, we identified villages that were more than 10 km from the nearest LPG center. These were the villages to target.

  • We then used market-level data to identify which of the underserved villages had enough market potential to support a new LPG center.

  • With all this data, we finally located the best location for a new LPG center in every district, based on where it will have the greatest impact.

At Atlan, we're opening up internal tools, like data catalogs, that helped us power massive data projects around the world.

Visit Atlan What Is a Data Catalog?