Geospatial Intelligence: Part Three – Distribution Network Planning, Logistics and Route Optimization, and Supply Chain Risk Management

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 Two of this series, we examined what it takes to build out commercial-grade maps and a live global supply chain map. Here in Part Three, we introduce seven applications of geospatial applications and drill down into the first three — distribution network planning, logistics/route optimization, and supply chain risk management.

Geospatial Applications for Supply Chain

Geospatial applications span a rich set of supply chain domains. Here we dive into seven areas in depth.

As indicated below, some of these applications focus mainly on planning and improving, others on monitoring and responding, and some span all four areas.

Distribution Network Planning

A distribution network consists of the facilities (manufacturing plants, distribution centers and cross docks, hubs, terminals, dealerships, retail outlets, and other endpoints of demand), as well as the equipment and workers associated with those facilities, the territories covered by the various assets, and the network of roads and trade lanes connecting those facilities. Network planning is the periodic (or event-triggered) reevaluation of whether the current network is optimal or needs to be modified. Some modifications can be done without a major capital investment or disruption, such as redrawing the boundaries of a territory or adding or redeploying resources (more vehicles and drivers over here, fewer over there). Others are decisions that involve larger capital expenses and longer-term larger impacts, such as opening a new facility or closing an existing facility. Network planning may be triggered by an acquisition, where you are trying to merge two networks and extract synergies by figuring out who does what best and assessing what functions and assets can be consolidated.

Network planning exercises need to take into consideration an incredible variety of data, the bulk of it related to location. Hence a GIS system can play an invaluable role in these projects. There are algorithms that can optimize certain aspects of a network, such as average end-to-end route length, population covered, and so forth. Companies often want to look at impacts across multiple dimensions. GIS systems can provide rich graphical views to help visualize these multi-dimensional impacts.

For example, Figure 1 shows a company’s various distribution centers, with concentric shaded polygons showing the 1-hour, 2-hour, 3-hour, and 4-hour drive time areas around each DC. It also displays all customers’ locations that are outside the 4-hour drive time. Each customer location is color coded and connected by a line to whichever DC has the closest drive time. Based on this analysis, the company might consider adding a DC in southeast Ohio or southwestern Pennsylvania, or consider other strategies to better serve a large number of underserved customers in that region. Different candidate locations could be evaluated using these same tools and views, along with other considerations.

Location Optimization Factors Differ Depending on Objectives

As a company considers expanding its production, distribution, or dealer/retail network, it will look at the market potential: how it can reach the most current and potential customers within the least delivery time. A tier one manufacturer planning out their production and distribution network might look for proximity to their OEM customers’ existing and future planned plants. This may include not only their existing OEM customers, but others they aspire to sell to. Other factors to be considered may include proximity to labor with specific skills, such as high-tech engineers in Silicon Valley or automotive specialists in Detroit. A company may also consider the cost of labor, cost of living, and quality of life, for attracting the right labor force they need, at the right price. The extremely high cost of living in San Francisco, San Jose, and Boston has caused many high-tech companies to open development offices in other locations such as Austin, Raleigh, Columbus, Houston, and Salt Lake City. These have become emerging high-tech hotbeds because of that. Hence location quotient¹ is often an important factor to consider. Awareness of these trends and the growing concentration of labor skill sets in various geography can be a critical factor in location selection.

Logistics-intensive operations will want to consider the surrounding infrastructure, such as the capacity of road and rail systems, average travel time, as well as forward projections of what congestion will be like along various routes and ports over future 5, 10, and 15-year time horizons. This may include not only increases to population and traffic, but the effect of planned infrastructure projects, such as new roads, bridges, and public transportation projects.

Logistics and Route Optimization

Once a map has been populated with accurate, up-to-date information, it can be used for route planning and optimization. This is where fleet owners, whether private or common carrier, optimize the route for a given set of shipments, using a specific set of criteria. These criteria typically include things like distance, driving time, idling time, avoidance of left-hand turns, delivery windows, and so forth. The optimization problem is different for three different scenarios: 1) multi-leg ocean routes, 2) multi-leg ground-only routes (such as for domestic parcel, and 3) single-leg trips, such as from a DC to a series of customer sites.

1. Multi-leg ocean routes center largely around the constraints of booking a slot on a given sailing.
2. Multi-leg ground-only route optimization typically considers three stages A) pickup, B) linehaul, and C) delivery, although more complex routing problems exist, such as multi-mode involving rail. Multi-leg optimization can be considered as a series of single-leg optimizations, but with iterations to optimize the handoffs between the legs.
   • A manufacturer who has a few dozen factories (some domestic and some overseas), thousands of suppliers, and hundreds of customers may ask their suppliers to manage inbound freight to their factories. The manufacturer manages shipments from the factory to their own DCs and from their DCs to customer’s sites. They may use their own fleet domestically, and carriers for overseas ground, ocean, and drayage legs. Many manufacturers are not sophisticated enough to optimize across the network and hence lose opportunities to improve efficiencies, such as reducing empty backhaul.
     A good GIS may help them take steps in that direction, such as inbound-outbound optimization.
     More often it will be used in tandem with a TMS or specialized route optimization software.
   • A parcel carrier typically has many regional centers, a few ground hubs, and a single air hub to serve
     a continent. They will usually take more comprehensive responsibility for optimizing the entire network, though they may also hand off the responsibility for individual routes to independent carriers.
3. Single-leg route optimization varies depending on the use. A retailer shipping from DC to store may focus primarily on truck load optimizing. A parcel carrier typically assigns trucks to a specific service area, and thus tries to optimize the route for all the vehicles in their area.

In addition to having all the right data and layers to perform routing, the GIS platform must support various business rules, such as equipment availability and capabilities, driver capabilities (e.g. special certification or skills, such as crane operation), driver availability, hours of service, customer delivery windows, and company-specific rules (for example, UPS doesn’t allow left turns on their routes). Optimization algorithms may be built into the platform, come from a third party, be built in-house (often using 3rd party libraries), or some combination of these. For example, esri ArcGIS provides built-in optimization, using origin-destination time-and-cost matrices and optimization algorithms to suggest routes. Logistics personnel can then iterate, curate, and share the optimized routes. The routes should be made available to each driver via their navigation device.

Supply Chain Risk Management

Assessing Supply Chain Risks

Once a company has built and maintains a reasonably accurate map of at least some portion of its supply chain, it can start to assess risks. There are many types of risks to consider:

• Natural Disaster Risk — Includes events like hurricanes, earthquakes, tornados, floods, wildfires, ice and snow, volcanos, tsunamis, and rising sea levels. There are maps that show levels of probability of these various events occurring, across the globe.²
• Political Risk — These include instability, civil unrest, strikes, civil war, regime change, nationalization (seizure of private property and assets), cancellation of import/export licenses, criminality and gangs, embargoes, currency or banking crises, and trade wars leading to dramatic changes to duties, tariffs, and quotas.
• Geographic Concentration Risk — If too many of a firm’s factories and their multi-tier suppliers’ factories are located too close together, it creates additional risk, because a single adverse event (such as natural disaster, or political catastrophe) could have a devastating effect.
• Capacity Risk – specific supplier or the industry overall may have insufficient capacity to meet unexpected demand. Industry-wide capacity shortages can lead to dramatic spikes in prices and lead times, and ultimately loss of revenue, profit, and market share.
• Reputational Risk — This includes substandard or illegal labor conditions (including child or slave labor) at the supplier or sub-tier factories, use of conflict minerals, illegal logging, environmental violations, and more. Reputational risk management may include keeping track of suppliers’ certifications, tracing of materials back to their origin, and onsite inspections and audits of conditions and practices.
• Infrastructure Risk — Some areas are prone to failure of electric power, water, communications, transport, or fuel supply.
• Security Risk — Supply chain-related security risks include cargo theft, IP theft, cybersecurity,³ and supplier fraud.
• Supplier Performance and Viability — Includes supplier solvency risks, delivery failures (late, short, damaged, or quality-check-failed deliveries), and unauthorized subcontracting.
• Logistical Risk — There are various categories of logistics risk: 1) delay and disruption — these are extremely common and should be handled in planning and daily execution, 2) hazardous materials — requires special route planning and preparedness, 3) security — theft, kidnapping, and other threats.
• Conflict-free minerals risk mitigation — Legal and ethical standards mandate that companies conduct due diligence to ensure their products are free from conflict minerals, such as tin, tantalum, tungsten, gold,
  or diamonds.

Providers of risk maps and risk-related data feeds may deliver updates in a batch mode or through web-service interfaces. Web services provide a way to ensure that the map stays up-to-date in near real-time for fluid, fast-moving, time-critical situations.

Composite Risk Scores

A composite risk score can be created for each location, part, product, or supplier. For example, a composite supplier risk score might have several components such as production risk (likelihood of disruptions to the plant’s production, which is itself a composite of natural disaster, labor, political and other risks), delivery risk (likelihood of late deliveries, poor quality, inadequate capacity, etc.), viability risk (supplier’s financial health, competitive landscape, etc.), and so forth. This gives risk managers and others a way of seeing at-a-glance the overall risk for a supplier, as well as ranking them so that mitigating steps can be taken for the riskiest suppliers. The same can be done for each site, part, or product.

Risk Mitigation, Contingency Planning

Armed with a more precise understanding of the risks it faces, a company can implement more targeted mitigation strategies, focused on where it will make the biggest difference (i.e. get the most bang for each ‘risk management buck’). Examples of mitigation actions are: seeking alternate sources to reduce supplier/plant concentration risks or natural disaster risks; adjusting inventory safety stocks to reflect risks faced; shifting the frequency and diligence of reputational risk audits to focus on high-risk areas; and heightened monitoring or limiting activities where political risks are high. Companies usually try to integrate risk information into their supplier-sourcing processes. With this highly granular information, suppliers can be given very specific instructions, such as the need to move production to a different location, or should provide an alternate location as a backup and store sufficient safety stock inventory at another location to cover the period of time it will take to switch production locations. Such information and procedures may be incorporated into a company’s and its suppliers’ business continuity plans.

Similarly, a company can develop more specific, less generic contingency and disaster recovery plans that more accurately reflect actual risks faced by each location and their priorities based on revenue or other impacts. This gives you more bang for your risk-mitigation buck. It is important that these plans are practiced and updated on a regular basis. A plan that has never been tried out or has just ‘sat on the shelf’ for years, is almost certain to be highly flawed.

Continuous Situational Awareness During Disruption Response and Recovery

When a disruption or disaster strikes, up-to-date contingency plans help tremendously, but no matter how good you are at planning and executing, ‘stuff happens.’ This is where real-time decision-making comes in — the moment-by-moment judgment calls made by dispatchers, plant workers, first responders, logisticians, and their leaders in the ‘heat of battle.’ Making the right moment-by-moment decisions requires good instincts and intuition, but just as critically, it requires continuous situational awareness — a clear, accurate, current, and full understanding of what is going on, across all locations, so you can figure out the best course of action and continuously manage and optimize the response and recovery.

Operational decisions are made every day. The operational decision-making process, and the enterprise systems that support it, usually suffer from the equivalent of the ‘fog of war’ — having an incomplete, imprecise, or out-of-date picture of what is actually happening on the ground. Traditional enterprise systems are not designed to provide that kind of continuous real-time situational awareness. This is where having good field data collection capabilities, integrated into the GIS platform, is critical. For example, FEMA, the Red Cross, many federal agencies, and the majority of insurance companies use mobile field data collection technology from esri. This enables field workers to report in near real-time on conditions and needs. In the chaos of a major disaster, this helps these organizations get the right resources to the right places, which can literally make the difference between life and death. The same tools can be used to record progress for various activities. For example, if someone in the field has reported several employees are stranded at a plant by rising flood waters, the same system can be used to track their rescue and record when they have been brought to safety and where.

Some utilities are using similar field data collection tools during both normal maintenance and disruptions. They take geocoded pictures of branches that need to be trimmed or downed trees that need to be removed, which are entered into a work order management system. The work orders are now linked with high confidence to the precise location information, saving crews valuable time. These same location data can also now be used to optimize routing for the crew. And, the crew’s current location can be correlated to the expected path of any approaching lighting storms to give the crew a ‘take cover’ warning so they can get down off a highly vulnerable boom ladder or power pole and get to a safer location on time.

IoT and Big Data Analytics in Response

IoT data from sensors are becoming increasingly important in getting a full detailed picture of what is happening before and during a disaster. A good example is the National Water Model that the National Water Center launched in 2016. This uses data from the U.S. Geological Service’s network of nearly 8,000 stream gauges and NOAA’s atmospheric modeling to provide high-resolution forecasts for 2.7 million stream locations nationwide. When Hurricane Harvey hit Texas, these forecasts were combined with live stream gauge data feeds to predict flooding around the Houston area. That was used by emergency workers to determine priorities for evacuation and to know where to erect temporary shelters.

Insurance companies combined the stream gauge data and flood forecasts with information from inbound phone calls, their own field assessments, footage from their own drones, and even social media postings to create a more complete and precise picture of the level and location of damage. In their GIS system, they overlaid their clients’ addresses with this data to anticipate the impact on each property and speed up the insurance provider’s response and the claims process. The same data could be used by manufacturers to determine the likelihood that their plants will be impacted by rising waters. This could make a huge difference in the manufacturer’s response to the event.

IoT data is almost always too voluminous to be directly monitored by people. It typically requires at least some sort of filtering and thresholds for alerts. We are more often seeing analytics used to ingest and interpret the sea of IoT data generated and extract meaning from it. This can be combined with machine learning and/or rules engines to help make predictions and direct attention to the most important events or insights being generated.

In Part Four of this series, we look at the application of geospatial intelligence in asset tracking and asset management.

_______________________________________________

¹ Data is more likely to be available and/or contain more detail for locations where people are inclined to buy insurance coverage. To an extent, these tend to be the same areas where factories are located. — Return to article text above
² In the context of supply chain risk, here we’re referring to the security of your suppliers’ systems. A breach in a trusted supplier’s systems can potentially impact and/or provide access by unauthorized parties to your own systems. — Return to article text above
³ For example, data on which factory each product is being produced in, which DC serves each customer location, which lanes and ports are used by each factory-DC pair, and so forth. — Return to article text above
⁴ Location quotient is a measure of how concentrated an industry, skillset, or demographic is in an area. — Return to article text above

To view other articles from this issue of the brief, click here.