Analyzing Supply Chain Resiliency
In Part 1 of this series on supply chain resiliency, I made the case that despite the current cache and buzz-worthiness of the term ‘resiliency’, there is value that can be found by analyzing the resiliency of the network. But the question is, how do you analyze resiliency within a supply chain network? In this post, I will outline strategies for proactive analytics to better understand the effects of a disruption and designing solutions to mitigate them.
There are a number of different algorithms that attempt to quantify the critical sites or lanes within a supply chain. For example, some identify sites with the most connections in the supply chain while others calculate the number of steps that each supplier has to take to bring product to market. While these network-theory based algorithms can be helpful, they don’t do a good job of quantifying the compounding effects that are possible within a supply chain or the balancing effects of inventory management. For example, which is more important, a site with numerous interactions with other sites or a single supplier that supplies a critical part?
Leveraging Scenario Analysis
A more balanced approach to quantifying resiliency uses the latest in supply chain modeling and analytics. The first step in analyzing resiliency is utilizing model scenarios that quantify the cost of alternative network designs that can increase the ability for a company to weather changes. Examples include maintaining higher levels of inventory, multiple suppliers for raw materials or components, multiple ports for imports or adding excess capacity to process lines. In many cases these will result in higher annual operating costs, increased supply chain complexity or lower working capital. However, these costs are offset by ‘benefits’ during times of disruption and crisis. So, which of these scenarios or at which of the facilities are investments worthwhile to have the best opportunity for reducing risk?
To understand which investments will have the best payback, the right metrics need to be chosen for the analysis. When thinking about resiliency, on-time delivery, missed orders, and lost profit are all metrics that are commonly used. Intrinsic to each of these metrics is the element of time. While optimization models can take into account time through use of multiple periods and incorporating production and transportation times, simulating a supply chain can incorporate a much finer set of details that more closely mimics the day-to-day operations of a company. Simulation models utilize a fixed network and a set of business rules to determine how a network performs under a set of circumstances. Using simulation, variability in the lead time of suppliers, delivery of shipments and even the loss of a site can be evaluated.
The Resiliency Index
Simulation models can take time to setup and produce huge amounts of data, including time series inventory positions for each facility, summaries of each shipment’s order and delivery date, and production times for each product. This amount of data can be overwhelming when looking at it for numerous facilities and numerous scenarios. One way to summarize each facility’s performance for each of the scenarios is to calculate the resilience index (as proposed by A.P. Barroso in 2015). The resilience index is the area of the resilience triangle (see diagram below) utilizing the metric of choice for performance (i.e. on-time orders, profit, etc).
For example, when tracking on-time orders, the baseline scenario assumes that all or nearly all orders will be fulfilled on-time under normal operating conditions. During a disruption, inventory will be used up and then orders will be missed. Each site will have a specific recovery time that may be significantly affected by other sites in the network. Using simulation this index can be calculated for each site. Summarizing the resiliency index for each facility under the different scenarios can give visibility into which sites are most vulnerable to disruptions at other sites or which resiliency measures will likely have the best return on invested capital. Similarly, another approach is to utilize a simplified linear model of the supply chain and knock out each site in succession to see which will have the largest effect on cost, revenue or profit. LLamasoft is currently testing tools to evaluate resiliency including the ability to easily simulate a network optimization solution and the ability to check off each site in the network.
The goal of evaluating alternative network designs that incorporate resiliency is to ensure that the network can respond robustly under a diverse set of real world conditions. Companies still need a system to monitor potential disruptions and react quickly and effectively when they do occur. So while there may not be a perfectly resilient network that never misses an order and returns a profit to investors, there are proactive ways to ensure that a supply chain can minimize the effects of a disruption and quickly put in place solutions when they do occur. Analytics and simulation can help bridge the larger strategic objectives with the ground-level decisions that make or break a strategy.
To learn more about supply chain resiliency, click to read our bulletin, “Using Modeling and Analytics to Design a Resilient Supply Chain.”