This article is an excerpt from the report Geospatial Intelligence: Powering the Next Wave of Supply Chain Performance.
A copy of the full report can be downloaded here.
In Part One of this series, we discussed how geospatial intelligence generates value for supply chains. Here we take a look at what it takes to build out commercial-grade maps and a live global supply chain map.
Beyond Consumer-grade Maps
Most of us have used the navigation tools available on our smartphones or dedicated devices. These are truly amazing tools, some with really well-designed user interfaces and remarkable accuracy. However, they are often not good enough for commercial trucking use or for a fully digital supply chain. The needs of a commercial-grade map vary by use but often require:
- Highly accurate and up-to-date base street maps — Maps must provide an accurate and precise reflection of the current reality on the ground, otherwise, drivers waste fuel and their time due to the extra driving and the need to figure out how to get where they need to go because the recommended route didn’t work. This includes details like precise representation of driveways and entrances, precise geocoding of addresses, temporary closures and detours, legality of left- and right- and U-turns including time-of-day and day-of-week limitations, one-way streets, and so forth.
- Route restrictions — Such as bridge heights, tonnage restrictions by the number of axles, speed limits (both posted and actual/practical limits, for example, lower speeds up or down a steep grade), load-type restrictions (e.g. no hazardous materials on specific roads), vehicle-type restrictions, maximum widths with and without permit, height of low hanging wires and other obstacles, hours of use, road type (asphalt, concrete, composite, gravel, cobblestone, etc.), and road quality. With these attributes, a custom map can be created that works for a specific vehicle and their specific time-of-travel.
- Site-specific maps and details — These are particularly needed for large campuses and properties, where the actual delivery point(s) might be a half mile or more away from the road. There are a few potential elements:
- Geocoding delivery points — addresses are typically geocoded to the curbside, which is adequate for many residential deliveries, but not for a big office campus or plaza. The geocode should point to the actual delivery point. If there are several dispersed pickup and drop-off areas on the property, it helps to have a separate geocode for each (e.g. loading dock, entranceway, etc.).
- Private roads — Include all roads and driveways on the property.
- Route restrictions — Speed limits and other restrictions on roads or driveways on a campus.
- 3D maps of building interiors — 3D is needed for multi-story buildings. Consider a driver that has many different deliveries to make at a site with multiple multi-story buildings. The GIS system now has a new routing problem to solve, beyond the vehicle routing problem. To solve this, the system needs to know the layout of the building(s) on the property, including the different floors and levels. Similarly, building-specific data such as the location of the freight elevator and doorway widths is needed for delivery of larger items such as machinery or furniture.
- Last 100’ details and imagery – lthough street-side images have become common, there can be a need for additional within-property imagery to help guide the delivery person to the correct door or area in a facility. It can also be useful to include special instructions with the images. Some of this information belongs in order-specific details, but if the same location is used over and over
(like “Pathology Lab #3”), then it can be useful not only to define that location on the map, but provide other cues that you are in the right location, such as a picture of it or description of a unique architectural feature to look for (e.g. “oversized double-doors, no glass”).
- Non-address delivery points — Sometimes a delivery needs to be made to a specific spot on a tract of land without a specific address. Examples would include a specific staging area for a construction site,
a particular oil wellhead, a specific field on a farm, a particular high-voltage tower, wind turbine,
or pumping station. The map should include precise locations and temporary roads to use. - Historical traffic data — Historical data about average levels of traffic and travel times, by time of day,
day of week, and time of season, helps route optimization algorithms do a better job. These can be augmented by adjustments for known upcoming events, such as adverse weather, major sporting events, parades, arena concerts, presidential visits, planned protests, and so forth.
There may be other static layers and associated data to help the transportation manager do their job. In addition, the GIS platform must incorporate a rich set of real-time data such as current and predicted traffic and weather/road conditions (fog, freezing/ice, flooding, snow, high winds, etc.). Real-time updates on unplanned events, such as riots, major accidents or spills, lockdowns, and so forth, can add value.
Decentralized Data Input and Quality Control
If you step back and think about the number of data points on a nationwide map, and the ever-changing nature of the physical reality they attempt to represent, it seems a near-impossible task to keep most of that enormous quantity of data up-to-date, complete, and accurate. Thankfully, a good GIS platform enables people already out in the field to continually make corrections, additions, and adjustments … alongside real-time verified information fields. For example, a customer who wants to ensure delivery to a precise location in their facility could be enabled to take pictures, submit floor plans and/or campus/yard schematics, provide written descriptions, and provide directions. Some people are starting to use drones to create highly accurate maps of their property and all its features, which can then be fed into the GIS platform.
Workers can be enabled to suggest corrections to a map or geocode-specific locations on a property. Those suggestions, collected across the network, may be gathered and sent to a central map data quality assurance group for review and approval. The process of generating more accurate geocodes can also be automated, as UPS has done.
More than ten years ago, they started automatically gathering GPS readings for each address – when the package is scanned, when the customer signs, and when the driver leaves the building. Because of GPS inaccuracies, as well as inconsistencies in when the driver pushes the button, they end up with a cluster of readings for each address. UPS developed algorithms that take those clusters of points and find highly accurate coordinates where the actual delivery location is. This has been used to build out their nationwide map of over 200 million precise address delivery locations.
That is just one example. In fact, all manner of data can be collected as workers are out and about doing their jobs. That data can be used to improve the accuracy and richness of maps and geospatial data available to the company, and for other purposes, in some cases enabling better service and creating a non-trivial competitive advantage. For example, a driver who is delivering a package when there is no one there to receive it can take a picture showing where they left it and then record the location using their device’s GPS. This can show the customer exactly where on their property to look.
Crowdsourcing of Service Areas and Rules for Automated Pricing
One solid waste hauler wanted to provide automated pricing for customers on their website. But they had a challenge: each of their ~200 different facilities around the country kept track of their own different service areas, the types of services offered in each area, and pricing for those services. In some service areas, the company might already have a contract with the municipality, whereas in other areas they make contracts directly with the homeowner or construction contractor. The company had no central repository defining the boundaries of each service area, which services were offered in each area, and pricing. Instead, these were managed by individuals at each facility who had the local knowledge in spreadsheets, paper documents, their collective memories (‘Joe knows about these areas’), or some combination of these. That method worked OK when customers called up the office for a quote, but the approach could not provide the automation required for self-service on the web.
So, the firm embarked on a multi-year process to create a centralized database and set of tools. Using a GIS system, they collected data such as the boundaries of each service area, the services available, and pricing. The GIS system provided employees out in the various facilities with an easy visual way to draw a polygon around each service area on a map. It provided clear visibility of streets, municipal boundaries, and property lines with the polygon edges snapped to these boundaries. The system double-checks with the employee to ensure it was what they intended. Preloaded municipality boundaries made it quick to select and define areas based on those. Users could import the data from spreadsheets as well.
They kept adjusting the approach and UI over time, until it was intuitive enough for their non-technical people in the field who did not have GIS mapping expertise to input all the information correctly and ensure business rules were working properly. The system worked with a regular browser, there was no special software to install. It included versioning to allow teams of people to work together, since there might be four or five different people entering the data simultaneously. Newly entered data goes to a quality control person, who is also local at each facility, to review and approve before publishing.
It was a big decision to cede ultimate control to people at the facility level, but was the right approach because the employees who work out in the locations are the ones with the intimate knowledge and experience to get it right. This helped the change management too. Giving the local people control and responsibility for the system helped them to take ownership of the project, rather than view it as ‘those people at corporate HQ’ imposing something on them. It also helped them envision how this was going to simplify their lives and automate the boring parts of their jobs. This is an example of how a complex and sophisticated set of data can be input and maintained by the workforce out in the field responsible for it.
Building Out a Multi-Tier Supply Chain Map
Developments such as globalization, low-cost ocean transport, and technology to integrate and synchronize with trading and business partners’ operations have driven massive shifts from vertical integration to a virtual enterprise model, with firms outsourcing most non-core functions. This has splintered supply chains into ever-increasing numbers of tiers spanning the globe with ever-increasing complexity. As a result, the network of physical plants, warehouses, sales channels and shipping routes that a manufacturer relies on to produce and distribute its wares has grown ever wider, more complex, and opaque.
The 2011 Japanese tsunami and Thailand floods and more recent disruptions brought into sharp relief the challenges for manufacturers not knowing their own multi-tier supply chain, the impact of any given event on production and revenue, and the actions needed to mitigate the effects to ensure business continuity and performance. This is often due to lack of up-to-date information on the most critical components, exactly where they were made, location and status of those plants and warehouses, up-to-date contact information, and other pertinent information.
To get a handle on risks, some companies have been mapping out the critical portions of their multi-tier supply chain and are linking those production locations to their products and BOMs. Most manufacturers focus on downstream (supply side) supply chains, though some also map out their upstream (demand side) distribution chains. Mapping a supply chain to multiple tiers is typically a labor-intensive undertaking, although tools to help automate the process continue to be improved. Tier 1 suppliers can be requested to provide and continually update the lat-long of each of their production facilities, as well as contact information for key roles. This should become a required part of the supplier onboarding process.
In addition, tier 1 suppliers can be requested or required to reach out to key tier 2 suppliers to ask them to submit similar information about their factories, who in turn may be asked to make similar requests of tier 3. The GIS system can be used to keep track of which suppliers have updated their plant locations and which have not, providing a project tool for the mapping team to ensure follow-up and continued progress toward mapping goals.
To be useful, the map must be continually kept up-to-date and accurate. GIS applications can be useful to maintain the quality and currency of this supply chain map data. For example, some manufacturers use satellite imagery to confirm the presence of a physical supplier plant at each specified location.2 Ideally this type of quality checking is done during supplier onboarding and periodically thereafter, especially when there is a trigger event such as a change in the ‘ship from’ address for received materials.
Maintaining a Live Global Supply Chain Operating Picture
Building an accurate map of a company’s supply chain is not a one-and-done activity. Supply chains are ‘living breathing’ entities that change daily. Maintaining an accurate global picture of production, distribution inbound and outbound routes, and suppliers require continuously updating it as changes occur. In addition to factories and distribution centers, a living map should also include the location of vehicles, personnel, inventory, routes, ports, and customer facilities, among many other geographically-specific objects. When the map has an up-to-date picture of where everything is, at the current moment, as well as status and information about the environment around it and global events, it provides the company with a ‘Live Global Supply Chain Operating Picture.’ This is a living digital representative of their supply chain — conceptually, a digital twin of their entire supply chain. Having a Live Global Operating Picture is a powerful competitive tool that can be used for many different purposes, some of which we discuss below.
IoT, Digital Twins, and Digital Supply Chains
Supply chains are increasingly instrumented with IoT sensors such as GPS on vehicles, temperature sensors in containers and pallets, various sensors on engines and equipment, and cameras everywhere. At the same time, many companies are striving to remove paper-based and manual processes from the chain and integrate systems end-to-end. It is a tall order, but much progress has been made.
All of this allows the creation of ‘Digital Twins’ where each asset, piece of equipment, or item has a digital representation of various attributes of that item that is continuously updated. For example, a digital twin of a refrigerated truck could have continually updated location, vehicle speed, engine conditions, temperature at different places within the refrigerated container, and sensor readings for the refrigeration unit (such as oil and refrigerant levels, whether it is running or not, and whether it is overheating).
This digital twin can then be used for things like tracking the condition of a shipment (alerting to any temperature excursions), providing continuous real-time situational awareness, and predicting when maintenance should be done. The GIS map then embodies an accurate reflection of reality on the ground for all assets at any given moment.
In addition to vehicles, digital twins can also exist for facilities, production lines, field service technicians, oil platforms and rigs, all sorts of machinery (e.g. portable generators, mining equipment, drilling rigs, etc.)—virtually any stationary or moving asset, including people, can have a digital twin representation. Digital twins’ location and sensor readings are often combined with data about the environment around them, such as weather and storms, natural disasters, power outages, political unrest, traffic conditions, port congestion, large gatherings of people, conflicts, market conditions … essentially any relevant data. When combined with the digital twin data, these create powerful predictive, real-time, optimizing capabilities for a fully digital supply chain.
Live Operating Picture Tracking a Fleet of Gondolas in California Wine Country
August to early December is harvest time in California’s wine country. During this period, three to four million tons of grapes are harvested and transported for processing at the crushing facilities of the region’s wineries. About a third of those grapes are moved by the logistics division of a single agricultural services company using their fleet of more than 1,200 food-grade gondola trailers. Each trailer has a single large gondola (bin) with a capacity of about 24 tons of grapes. In addition, the company has a large fleet of tankers for carrying wine, spirits, juices, and distilling materials. During the harvest season, there is a fast-paced and enormously complex ‘dance of assets’; dropping off empty bins where they are needed next, picking up full bins as soon as they are ready, bringing them to the crushing facilities, and then bringing the empties back out. Choreographing this dance to minimize driving time/distance and idle assets is extremely difficult, especially without visibility into where the bins are and their status.
Prior to setting up a tracking system, the logistics firm did not know exactly where all its bins and tankers were. In off-harvest season, the bins sit in fields and the company would lose track of where some of them were. It was even harder to keep track of them during the fast-paced harvest season. They outfitted each asset with a GPS tracker and cellular backhaul. Now they see in real-time where all their assets (trailers, bins, tankers) are, whether on the road moving or sitting stationary at a site. They use an esri-based dashboard, built by GeoDecisions. They also keep track of the movement history of each asset. They created geofences around key areas (such as ports, fields, and wineries) to see when assets enter and exit those areas. An alarm can be set to go off if an asset sits at one of those sites too long. This is especially important for the port where they will get charged a detention fee if the trailer sits there too long.
In addition to the visual interface, the system has an API integrating this data into their Oracle TMS, where they can use it for transportation planning, reporting, and other business uses. With this system, asset utilization has improved dramatically. Looking forward, the company plans to start using the same system to track the grapes within the bins as they are moved from the field to the winery, which they expect will add to their already high payback for the system. They have been able to leverage their GIS capabilities not just to track assets, but to improve performance dramatically.
In Part Three of this series, we look at the use of geospatial intelligence in distribution network planning and supply chain risk management.
__________________________________________________________
1 One firm used a mobile device to notate on a floor plan map of their suppliers’ facilities which locations were fully or partially covered by sprinklers. They asked suppliers to increase coverage where needed. They also identified supplier locations that were exposed to flood risks and convinced some of them to build or raise levies protecting their plants. By showing these risk mitigation improvements to their insurance company, they were able to get about a 5% reduction in premiums, a savings of tens of millions of dollars per year. — Return to article text above
2 If the factory location that the supplier provides shows up on the satellite image as an empty field or something obviously not a factory (such as a residence, or farmhouse) then corrections can be made to the plant location data provided. With the right experience, someone can tell not only if a building is likely to be a factory, but whether it is the right type of factory. For example,
a steel rolling mill will look rather different than an apparel cutting and sewing site. — Return to article text above
To view other articles from this issue of the brief, click here.