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 Six of this series, we examined how geospatial intelligence can aid in ensuring safety and security in supply chains. Here in part seven, we look at the role of geospatial intelligence in the service supply chain.
Service Supply Chain
Service Network Planning
Building and growing a network of service centers can be an enormous investment, with many different factors that go into picking the ideal locations. A GIS platform with network planning capabilities (such as esri’s Territory Design) can be very useful in evaluating the various options, looking at the tradeoffs, and deciding where new service centers should be placed. This kind of tool can be used by itself or as a complement to existing algorithmic network optimization tools. A GIS platform can provide a way to visually evaluate the various options prescribed by the optimization algorithms, bringing to light additional data and insights. To support the decision-making process, a variety of information may be brought into the GIS platform such as: locations of existing customers, locations and numbers of potential customers, driving distances, customer types and their equipment needs, potential for servicing another brand’s equipment, and so forth. Armed with this information, the platform can help planners make smarter choices, considering all the factors for these high-cost, high-consequence decisions.
Knowing with Confidence What SLAs You Can Deliver, Growing the Service Business
The same GIS system and sets of data can also be used during the sales and renewal processes to ensure that salespeople and customer reps are only offering Service Level Agreements (SLAs) that are actually achievable. Mapping technology can draw concentric boundaries around each service center based on the typical drive times from the center to each SLA boundary (such as 30 min., 1 hr., 2 hr., and 4 hr. boundaries). The system can quickly show which SLAs are actually achievable for any given customers’ site, so the salesperson knows which SLAs they can reliably promise.
Each time a customer or prospect requests an SLA that could not be achieved within the current network, it can be recorded in the system along with the customer location. That information can be used to show the gaps in the network and identify opportunities to grow the business by adding service centers, using a data-driven approach to sizing the opportunity. In addition, the system can spot all customers that have the potential for a higher SLA than they currently have. When the service rep is next talking to those customers, they can be prompted to inform the customer that a higher-level SLA is available, or a marketing campaign could be run informing those customers.
Positioning and Managing Service Resources
Optimizing Service Parts Inventory
Within an existing service network, decisions need to be made about where to position resources; specifically, how many of what spare parts are needed where, the number of service technicians with specific skills needed in each territory, and the type and quantity of repair equipment needed. There are systems that do advanced service parts inventory optimization for multi-echelon1 networks. This is different than standard inventory optimization algorithms for manufacturing raw materials or for finished goods for sales, because of the difference in the nature and predictability of service parts demand, the need to account for extremely slow-moving parts (sometimes less than one per year for a location), and the multi-echelon characteristics of the network. These systems need geospatial information (often in the form of lead times/transportation times) and an understanding of the location of priority customers in order to optimize spare parts inventory location. The right GIS system can maintain and provide source and customer location information, which can then be fed into the inventory optimization engine.2
Optimization of Jobs, Technicians, Vehicles, and Equipment
As service calls come in, technicians need to be assigned to each job. This is a continuous optimization problem since jobs come in throughout the day and because the various jobs-in-progress or travel time may take longer or shorter than planned. When these deviations from the original plan cross a threshold, it can require a replan of the rest of the day’s jobs and resources. Many organizations handle these deviations and disruptions manually, often with limited visibility into the options available, resulting in missed service windows and dissatisfied customers. As with service parts optimization, there are specialized systems designed for this, or it can be done using optimization algorithms integrated into a GIS platform. These systems are used to assign the technician and equipment with the right capabilities, knowing their current and future locations, to optimize time, cost, miles-driven, priority customers, and multiple other factors according to the business’s specific requirements and objectives. It is not just about optimizing a single technician or truck or service call/job, but rather the overall combination of all jobs, vehicles, equipment, and technicians together. When done well, it typically results in 30%-40% improvements in performance over the dispatcher making all the decisions manually.
Many factors may be incorporated into the optimization, such as the technician’s labor rates (both regular and overtime), run cost of the vehicle (including fuel, insurance, and maintenance), special equipment needs (such as a lift gate needed on the truck for a heavy item), size restrictions per customer (some sites may have width or height limits at their facility), technician skills (matching specialties with the need), relative priority of different customers and jobs, and the promised time window. Any attribute the business has data for can be used in optimizing the sequence and schedule.
From the technician’s perspective, they see their assignments and the sequence of jobs on their mobile device. For each job, specific instructions can be sent on what parts and equipment to bring, where to go next, and specific repair instructions.
Cisco Optimizes Service Resources to Meet SLA Obligations
Cisco Systems is the world’s largest manufacturer of networking infrastructure (e.g. hubs, routers, switches, access points, etc.). Cisco has a network of service centers where it offers maintenance and repair services for the equipment it sells to clients. For many of its clients, if their data and voice network go down, the client’s business goes down (i.e. cannot operate). Thus, Cisco offers 24X7 and 8X5 service, with service level agreements (SLAs) specifying “Response Time Obligations” for how quickly they will get to the customer’s site. Depending on the type of service contract and the severity of the issue, the SLA response time window can be 15 minutes, 30 minutes, 1 hour, 2 hours, or 4 hours.
To consistently meet those SLA targets, Cisco has to have the right resources (equipment, parts, technicians) positioned and available in the right places, at the right time. Cisco uses a GIS system from esri to determine which parts it should stock where, based on what networking equipment is at all the customer sites served by each service center and the expected rate of repair for that equipment. That information is used by the system to prescribe stocking levels and locations for the various parts.
Cisco’s SLA also includes a “Repair Time Obligation.” In case a piece of equipment can’t be repaired within the repair time obligation, Cisco keeps an inventory of spare working units around that they can swap in while the malfunctioning equipment is being repaired. As with spare parts, they use the esri GIS system to calculate which spare equipment inventory they need to keep in stock, based on the types of equipment in the customer base within the service area and the promised SLAs.
While shorter SLAs provide higher revenues for Cisco, they don’t want to sell a service level that is too short, i.e. an SLA window that is physically impossible to deliver for any particular customer’s location. During the sales cycle, the same GIS system is used to determine the best feasible response times for each customer site, based on the driving distance to the service center. This enables them to have an honest conversation with prospects and customers and avoid promising a service time that can’t be delivered. The customer thereby gains confidence that the SLA will be honored. Cisco uses this same system to determine overlap and gaps in their service coverage, so they can add or move service centers to optimize coverage for their customers’ and prospects’ sites.
Optimizing a Traditional Territory Model
Some companies like to use a traditional territory model, where each technician or driver has a fixed territory they are responsible for. This has the advantage that drivers get to know their territory well: the streets, customer locations, and the customers themselves. The technicians can establish a relationship and trust with the customers. For companies that want to use a territory model, they can use ‘soft constraints’ for optimization. In this, drivers and technicians still have their territory, but may also be assigned service jobs that are near but not in their territory, whenever it saves time and money and balances the workloads.
This approach works best when there is a degree of fungibility3 between the different technicians’ skills and the various equipment.
Continuous Optimization of Resources and Jobs Throughout the Day
As mentioned, optimizing service jobs requires continuous optimization. An emergency call may be received in the middle of the day. Delays can happen, due to traffic, weather, or a repair taking longer than planned. The optimization algorithm needs to be able to look at all existing jobs, vehicles, and technicians and reshuffle the plan in the middle of the day, to make best use of available resources, while still honoring all the constraints and business rules. This includes honoring time constraints, like keeping the promised time windows or arriving before the facility closes. Perhaps a technician that is finishing a job early and is not too far away can take over a job for another technician that has been delayed. At a minimum, the customer can be notified, sooner rather than later, that the technician is running behind schedule and a more realistic updated ETA can be provided, or the customer can be given the option to reschedule if desired.
Servicing and Maintaining Assets
A geospatial system can be tremendously helpful for servicing and maintaining assets, such as factory equipment, vehicles, computing, or office equipment. This is especially important for a service technician that travels to different unfamiliar customer locations. It starts with navigation to the customer facility. Once the technician is at the facility, it is critical that they don’t waste time trying to find the right building, room, and machine — and even more critical that they are working on the right machine. By tracking the location of each asset, the technician can see it on a local map and be given instructions on how to go to the fixed asset. The asset may also have a 2D barcode or RFID chip on it, to ensure that the tech is working on the right machine and doesn’t make any data entry errors when reading and typing in its serial number.4 This same auto-ID5 tag can also be used to provide proof-of-inspection, because it could only be read by an inspector who is physically there at the machine.
A geospatial system can also help with predictive maintenance by providing information about the conditions around the machine. For example, an outdoor machine may not have a temperature sensor on it, but with location data, you can correlate the local weather conditions over time, including events like sandstorms, which might drive the system to recommend accelerated replacement of certain parts (e.g. air filters, seals, lubrication fluids, etc.). For more on predictive maintenance, as well as the optimization and scheduling of technicians, vehicles, and equipment, see the section on Service Supply Chain above.
Repairs and Failure Analysis
Geospatial systems can be used to analyze location-related patterns of service repairs and parts consumption. Manufacturers can analyze the data to find out when and where parts are being used more in one area than others. They can bring in additional data to analyze and uncover the patterns and causes. They might discover that their air filters need to be changed more often in places with high pollution or lots of sand storms or that other parts’ life expectancy is impacted by high humidity. Knowing these patterns and causes of consumption can help them more accurately forecast service demand, do a better job of inventory optimization, and potentially make changes to designs.
Companies may modify designs or make special-condition versions to accommodate use in areas based on weather, air and water conditions, chemicals or pollution, driving and/or usage habits, varying road conditions, or other factors that might influence how often different repairs are needed. One automotive manufacturer used a GIS system to discover where brake pads and rotors were failing at higher rates and try to understand why. By analyzing a heatmap of areas of high failure, they realized that certain municipalities or highway departments used a specific type of road salt during the winter, which was causing rapid damage to the pads and rotors. As a result, they started making those parts using a different material that stood up to those chemicals better. In addition, they did a recall of vehicles in those areas to replace the old pads and rotors. Another vehicle manufacturer was getting complaints of significantly higher than usual number of fuel injector failures. When a geospatial hotspot analysis was done, it homed in on specific fueling stations, all supplied by the same refinery. In the end, it was discovered that a disgruntled employee at that refinery had dumped a bunch of sand into a large batch of fuel. The GIS system was instrumental in solving that mystery.
The traditional approach to scheduling maintenance of equipment is based on elapsed time, hours of use, or miles of use. Recommendations for maintenance milestones (such as what should be done at a 5-year/50,000-mile checkup) are based on estimates (with a margin of safety) of when the various parts are likely to fail. Of course, not all parts will fail after the same period of use, and there is no way to tell which specific parts will fail first. So, this approach compels a conservative approach to maintenance timing. Most components are replaced long before they actually need to be replaced while a few will still fail before the recommended replacement period, causing disruption and expensive emergency repairs.
In contrast, predictive maintenance monitors the actual conditions of use and data from sensors on the equipment, measuring things like temperature, vibrations, presence of chemicals and particulates in the lubricating liquids, and so forth. Machine learning can be applied to create a much more accurate picture of actual wear and tear and a precise prediction of when each specific part will fail for each machine. When a replacement is needed for a specific part, the system may also recommend replacing specific other parts while the technician is on site. With this approach, there are fewer scheduled maintenance calls, fewer spare parts used, while simultaneously dramatically reducing the number of failures. According to independent
surveys,6 implementing a functional predictive maintenance program reduces equipment breakdowns, on average, by 70%-75% and downtime by 35%-45%.
Role of Geospatial Data in Predictive Maintenance
Data from sensors is typically only one part of the data needed for good predictive maintenance. The nature and timing of repairs is heavily influenced by many factors, including the environment and how the equipment has been used. For example, a bulldozer used in Phoenix (exposed to sand and heat, but not cold or humidity) will have different types and rates of repairs needed than one being used in Fairbanks (extreme low temperatures) or New Orleans (high humidity). Manufacturers and equipment rental companies7 can put GPS trackers on their equipment and match the equipment’s location with weather data from that same time and location to understand the conditions it has been exposed to.
For vehicles, grade steepness data can be used to predict the expected wear and tear for brakes on trucks frequently traversing mountain passes. For mining and farming equipment, a geospatial system can use data on surface roughness, soil mechanics, geology and other characteristics impacting wear and tear. Preventing the breakdown of critical farming equipment during harvest season can save a lot of valuable crops from wasting in the fields.
Some truck tire manufacturers are offering tires-as-a-service, where the customer doesn’t buy the tire, but pays by the mile. The manufacturer is responsible for maintenance and ensuring that tires are changed before they wear out and become unsafe. They put sensors on the tires and wheels to measure things like hard braking and combine that with weather conditions, road steepness, and road friction (based on road surface type — concrete, asphalt, dirt, cobblestone). The system predicts tread wear and prescribes when to change the tire — not too soon or too late.
When maintenance is needed, one heavy equipment manufacturer is using their GIS platform to determine which dealer is closest to where the equipment is at the time of the scheduled maintenance. They automatically assign the service call to the right dealer. For onsite repairs, the system can also provide the dealers with exact equipment location information, beyond the address, since the construction or mining equipment might be anywhere on a several thousand-acre parcel.
Part Eight of this series looks at the role of geospatial intelligence in enabling sustainable and socially responsible supply chains.
1 Multi-echelon networks for spare parts are distribution and service networks with multiple tiers; from factory, to multiple-levels of DCs, to the repair centers, and ultimately to the customer’s sites. Typically, the fastest moving and/or most critical parts will be held at local service centers, or in some high-impact cases (such as >$1M/hour downtime costs) parts may be stored right there at the customer’s site. Slower moving, less critical, and/or more expensive spare parts will be held at a regional DC. And the most expensive, least critical, and slowest moving parts may be kept at a central DC, the supplier’s DC, or in extreme cases the manufacturer may carry none and wait until needed and built on-demand. — Return to article text above
2 Note: Inventory optimization engines are typically not part of a GIS system, but rather are separate specialized systems that can benefit from receiving the data held in the GIS system. — Return to article text above
3 Fungibility is interchangeability, the ability to swap out (in this case) one technician with another and still get the same results. Highly specialized technicians or specialized equipment are usually not very fungible; they may be the only one that can get that specific job done. — Return to article text above
4 Having an RFID, NFC, or prominent barcode to unambiguously identify the machine can save the technician time as well, especially if the manufacturer’s serial number is in a hard-to-access and/or dimly lit location. — Return to article text above
5 Auto-ID = Automatic Identification, such as barcodes, magstripes, RFID, or NFC. — Return to article text above
6 The surveys with these improvement numbers are cited on page 7 of the EEAP (Energy Engineering Analysis Program) publication Top Operations and Maintenance (O&M) Efficiency Opportunities at DoD/Army Sites. — Return to article text above
7 One heavy equipment rental company has GPS on all their big equipment. They have had incidents where a customer returned a backhoe or other equipment and complained ‘it stopped working.’ When they analyzed the historical GPS data, they saw that the equipment was left in an area that got flooded with several feet of water during a storm, giving them strong evidence of the real cause of the failure. — Return to article text above
To view other articles from this issue of the brief, click here.