Monitoring A Large-Scale Construction Project Using Land Use Change and Object Detection Algorithms (Part 2)

How geospatial analytics can help you see changes that were unseen in the past with Facility Intelligence


In the previous article (part 1 of this series), we introduced the concept of facility monitoring. We described how geospatial analytics could be applied to Cheniere’s Sabine Pass, an LNG export terminal located on Louisiana’s Gulf Coast. We ran Orbital Insight’s Land Use algorithm, which uses satellite imagery to detect critical changes to the liquid natural gas (LNG) Terminal during a specific time range. In this article, we expand the initial analysis to include vehicle traffic algorithms and demonstrate the multisource capabilities of our Geospatial Intelligence platform. By blending the results from Land Use Change and Vehicle Traffic algorithms, we highlight the impact of events such as the Covid-19 pandemic and show how local changes impact the broad LNG supply chain. While we have used an LNG facility in this case, the same analysis can be applied to manufacturing, distribution, or any construction project.

Vehicle Traffic Algorithm

The results of the land use algorithms show two critical differences between 2019 to 2022. First, there were significant changes to the site in the northeast corner. These changes are reflected in the land use map and the associated satellite image. Second, the location of Sabine Train 6 is immediately clear in the results. There is also activity on the site's southeast corner, showing Cheniere may be constructing a new berth for ships. Perhaps the new berth will accommodate the new production capacity brought on by the new train.

With the location of the new liquefaction train identified, it’s now possible to monitor construction progress over time and to determine whether or not that progress has influenced changes in production and activity at a micro level. Micro AOIs can help identify granular changes. Figure 4 shows micro AOIs across the LNG terminal. The location of each is not arbitrary. The land use algorithms not only gave insights about land changes but also provided a snapshot of where plant employees were parking. By drawing AOIs around each parking lot, users can run algorithms that pull data from connected vehicles over the same time frame. One key benefit to each of these algorithms is that they can be run simultaneously on


more than one AOI. In addition to the AOIs for each parking lot, the macro AOI drawn over the entire plant can be used as a control. The control will show vehicle traffic at a macro level, while each smaller AOI for the parking lots will reveal changes to an individual lot. This can help determine if the activity for each lot level influences activity at the entire facility. Due to the construction location, there will likely be an uptick in vehicle traffic in the two northernmost lots. Results of the vehicle traffic algorithm are shown in Figures 5, 6, and 7.

Figure 4 Drawing micro AO Is for each parking lot

Figure 4: Drawing micro AOIs for each parking lot | Micro AOIs drawn inside of one large Macro AOI representing the entire site. Each parking lot should have unique results that vary depending on the date. Weekends will likely show little to no traffic at all, while weekdays will show high traffic.

Figure 5 Vehicle Traffic at Sabine Pass 2018 2022

Figure 5: Vehicle Traffic at Sabine Pass, 2018 - 2022 | Notice two spikes in traffic before and after the start of construction in June of 2019. Then, from April 2020 to Spring 2021, there is a drastic increase in production as construction nears completion. Foot traffic remains high but relatively constant through 2022.

Figure 6 Vehicle Traffic at Employee Parking lot 5 2018 2022

Figure 6: Vehicle Traffic at Employee Parking Lot 5, 2018 - 2022 | This AOI elicits really interesting results when compared to the AOI representing the entire facility, specifically in the numbers. On average, a little over 20 vehicles are registered daily in Employee Parking 5. At the site level, a little over 50 vehicles are registered daily. We now know that vehicle traffic in Employee Parking 5 consistently accounts for around 40% of vehicle traffic site-wide. We can also notice that there is no traffic at all prior to the start of construction. Peak traffic site-wide matches peak traffic for this particular AOI.

Figure 7 Vehicle Traffic at Parking Lot 4 2018 2022

Figure 7: Vehicle Traffic at Parking Lot 4, 2018 - 2022 | While traffic at Employee Parking 4 is less prominent than 5, it still paints a picture of active vs. inactive periods. Additionally, the peaks at all three AOIs (Spring - Summer of 2021) match up.

The results above tell users many things. At a micro level, they show which lots were used for workers during the construction phase. They also show when production was high or low from counts over time. It is evident that the construction of the liquefaction train was very productive for much of 2021. The drop in late 2021 to early 2022 can likely be attributed to construction nearing completion. As a result, fewer workers needed to get the job done. At a macro level, one can infer that activity in Employee Parking 5 influenced activity over the entire site. While these changes weren’t drastic, the peak in production between all AOIs in 2021 aligns. We can also combine the results of land use with the results of vehicle traffic, as seen in Figure 8. Because of the terminal’s enormous size in reference to the smaller construction area, there will likely not be many fluctuations in the buildings and roads plot. 

Another useful measure of production is analyzing these same results on a weekend and holiday basis. In Figure 9, car counts are plotted with both weekends (green) and holidays (red) highlighted throughout construction. Workers were more productive on weekends and holidays shortly after construction began and when construction neared its peak. Figures 10, 11, and 12 show detailed views of the same graph at different times. Figure 10 details project initiation. Figure 11 details the onslaught of coronavirus and its impact on productivity. Figure 12 shows peak productivity and project completion.

Figure 8 Combining Results from Land Use Vehicle Traffic

Figure 8: Combining Results from Land Use and Vehicle Traffic | The graph shows changes to the device (car) counts, buildings, and roads over time. Because Liquefaction Train 6 is just a fraction of the size of the entire site, we won’t see much flux in buildings and roads. It is likely that building counts declined over time because of the decreased need for construction resources like planning sheds and staging pavilions.

Figure 9 Car Counts During Construction Weekends Holidays

Figure 9: Car Counts During Construction - Weekends & Holidays Highlighted | The graph details car counts over time with both weekends and holidays plotted. Interesting spots are the project started in June 2019, decreased activity due to COVID, and peak construction in the summer of 2021.

Figure 10 Start of Construction

Figure 10: Start of Construction | Following the project start, car counts increase from mid-40 to just under 100.

Figure 11 Coronavirus

Figure 11: Coronavirus | While COVID-19 slightly impacted activity from March to May of 2020, car counts quickly rebounded in June.

Figure 12 Increased Productivity and Project Completion

Figure 12: Increased Productivity and Project Completion | As construction neared completion, one might notice that more cars were reported on both weekends and holidays (specifically Halloween and Thanksgiving) in late 2020-early 2022. One might assume that productivity significantly increased during this time frame.

Summary and Conclusion

Orbital’s Land Use and Vehicle traffic algorithms are a few examples of tools that can be used to monitor facilities like Sabine Pass. Running and analyzing the results of these algorithms is just the first step to understanding how small-scale changes affect the entire supply chain. Because we know that Sabine Pass is the leading producer of LNG globally, we also know that their added capacity will only make them more prominent. Will more natural gas from domestic sources flow through the terminal and be shipped to foreign ports? Will the facility now export to a greater variety of vendors? Or will the volume of LNG exports directly from Sabine increase? Perhaps both. Will the global increase in gas prices due to inflation and war be limiting factors to this supply? These are all questions we will answer using Orbital Insight's platform. Stay tuned for our next article!

More to explore...

  • Facility Intelligence offers comprehensive visibility at any industrial facility to identify change, mitigate risk and gain operational insights
  • Download the eBook, 7 Keys to Your Supply Chain Success, and see how customers like Unilever, Celanese, BP, and others, are using geospatial analytics to gain Supply Chain Intelligence