Apr 14 2020
David Simchi-Levi has been a contributor to supply chain management for more than 20 years and I have two of the books he co-authored on my shelf, Designing and Managing the Supply Chain, a textbook for business and industrial engineering students, and The Logic of Logistics, for readers who care about the math behind the formulas and algorithms. So, when he announced a webinar on supply chain recovery after a pandemic, I signed up to listen. The full webinar is now available on Youtube, including a couple of minutes of dead silence in the middle while he was reading audience questions. The slide set is also available in PDF, and it includes an appendix with descriptions of algorithms not discussed in the webinar.
Michel Baudin‘s comments: As the COVID-19 pandemic is a currently unfolding catastrophe, the webinar devotes a large opening section to admiring the problem. This section is as of April 8, 2020. If the webinar were held, say, in August, 2020, this section would require an update. Simchi-Levi then goes on to describe a Risk Exposure Model that he and his team co-developed with Ford in the early 2010s based on the experience of supply chain recovery after the Fukushima earthquake or the Thailand floods of 2011. It is less connected to the latest news than the introduction.
These events disrupted, in particular, the supply of black paint to Ford, so that its customers could buy a car in any color as long as it was not black. They were major disasters at the time but, in retrospect, look small when compared with COVID-19. Simchi-Levi then uses the Risk Exposure Model as the basis for a series of recommendations to manufacturers.
Does Lean increase supply chain risks?
According to the first slide, supply chain risk increased in the past ten years due to outsourcing/offshoring and lean manufacturing. For a US manufacturer to source parts from China or Vietnam is clearly risky but the lean approach, at least as in TPS, is instead to work in close collaboration with local suppliers as much as possible. It speeds up supply chain recovery. It does not increase risks.
The paradox of stock
If Just-In-Time and low inventories enhanced supply chain risks, then, presumably, Just-In-Case and high inventories would reduce it. You can have full warehouses and still be short of the specific item you need, like Rodney Dangerfield’s wife in the movie “Back-to-School,” with “six closets full of nothing to wear.”
Strategic reserves work for crude oil, as it is one item with many uses. A grocery supermarket, on the other hand, sells 10,000 different items and cannot have “2 months of everything on the shelf.” Neither can a factory assembling 100 different products from thousands of purchased items.
The need for cash in a crisis
When the activity abruptly stops, 2 months’ worth of an item turns into eternity. Then the cash tied up in inventory is unavailable for immediate needs and this can bankrupt the company.
Milk runs and prevention of disruption
Ironically, one of the key tools of lean logistics, the supplier milk run, was invented to prevent supply chain disruption. As part of its “ABC system,” the German military aircraft industry developed the milk run in the run-up to World War II. The purpose was to mitigate the effect of enemy air raids. They organized a network of suppliers within a 50km radius around the assembly plant. Trucks then made the rounds to pick up matching sets of components from multiple suppliers. They then assembled the components into planes that flew off immediately. This was less risky than holding parts in a warehouse at the assembly plant.
To the extent that Lean is based on TPS, the strategy is not “low” inventories. Instead, you keep the amount necessary to support routine operations and buffer against routine fluctuations, like minor stoppages in the suppliers’ lines, icy roads, or traffic jams. You just don’t try to use inventory as a hedge against fires, floods, earthquakes, stock market crashes, or pandemics.
The Risk Exposure Model
It is based on metrics:
- Time-To-Recovery (TTR). The time needed for a supplier to recover full function after a disruption.
- Performance Impact (PI). The impact of a disruption on a performance indicator.
- Time-To-Survive (TTS). How long the supply chain can operate with one node down.
From both the webinar itself and from the algorithms provided with the handout, it appears that the model has a binary view of the supplier: either it operates fully or it doesn’t. A real situation, however, more likely involves a gradual ramp-up, with the supplier being able to provide a growing fraction of the demand until getting back to 100%.
While Performance Impact is only one line item above, there actually is a different one for each metric of performance. Regardless of the metric, the Performance Impacts of multiple suppliers are not additive. The performance impact of two suppliers for the same product being out of commission is the same as for just one. You can only make the product when all the suppliers are working.
The time to survive (TTS) also varies. The available inventory of components will last a certain time if the company keeps operating at the pre-disruption level but longer if it reduces its activity. The COVID-19 pandemic has put most of the manufacturing sector into an induced coma and running out of inventory won’t be an issue until it restarts. The TTS clock will only start running then.
Simchi-Levi’s presentation addresses only the case suppliers being unable to deliver. This was the situation in January and February 2020, with shutdowns at Chinese suppliers of US and European companies. Two months later, the tables are turned, with the Chinese suppliers recovering while their customers shut down, and the performance impact now hits the suppliers.
Simchi-Levi makes five recommendations, using the Risk Exposure Model as a framework. Some of them, however, combine two different steps, which I have separated here into nine instructions with each a single action verb:
- Map your supply chain.
- Estimate time-to-recovery by scenario.
- Estimate demand for each scenario.
- Assess which products and assembly facilities will be impacted by affected suppliers.
- Determine when and for how long you should shut down or reduce manufacturing activities.
- Determine when to ramp up capacity.
- Allocate the available resources to products that allow you to achieve specific objectives during the recovery period.
- Determine when to expedite and for how long.
- Book logistics capacity as soon as possible.
Map your supply chain
A quick poll of webinar attendees revealed that the majority had not mapped their supply chains. For decades, supply chain management experts have been telling companies that they should know their suppliers’ suppliers and their customers’ customers. It is, however, not easy to do. A company has business relationships with its direct customers and its tier 1 suppliers but not beyond. Tier 1 suppliers may not be willing to share information about their own suppliers. In fact, the point of having tiers is to have the customer work with 200 rather than 2,000 suppliers but the corollary of not directly managing these 2,000 suppliers is that you have less information on each.
The webinar cites two technology companies, interos and resilinc that offer automatic multitier supply chain mapping. Perhaps, a feasible alternative to mapping the entire network down to the mining company that extracts materials from the ground is to map the Tier 1 of suppliers and rely on each to map its own. If it cascades down through the lower tiers, the map will exist even if the customer only sees the results as aggregated by the Tier 1 suppliers. Knowing your customers’ customers is a different issue. If you are not selling directly to end-users, the distribution network can screen you from market information that you need to plan and operate.
Estimate time-to-recovery by scenario
The webinar describes three scenarios of worst, best, and most likely evolution of the pandemic. For 200 Tier-1 suppliers and three scenarios, it means estimating 600 parameters. None can be left out. Shortages of nuts, screws, or washers can prevent final assembly as well as shortages of motors, pumps, or microprocessors.
Sometimes, the same number can possibly be used for different suppliers making technically similar components in the same geographical area. In general, unless a supplier has recovered from disasters before, you don’t have much data to go by. All you know is the supplier’s performance in routine operations and the perception of their capabilities by your employees.
Given that suppliers are likely to have made their own recovery plans, the most expedient way to get times to recovery is to ask them but the answers will not be objective. The managers are keenly aware that they are more likely to remain a supplier if they recover in 2 weeks rather than 6 months. As a result, they will quote the shortest recovery compatible with the laws of physics and count on forgiveness when they are late in actual recovery.
Estimate demand for each scenario
While not easy, it is done by the same methods as under normal operations to support Sales & Operations Planning (S&OP).
Assess which products and assembly facilities will be impacted by affected suppliers
In principle, the data needed should be available in the company’s ERP system. Even if it does not support these particular queries, it should be retrievable from the master data. As management may realize as a result of this crisis, however, the dominant ERP software products do not necessarily make it easy. It may take longer, require more skills than expected, and reveal inaccuracies.
Determine when and for how long you should shut down or reduce manufacturing activities
With COVID-19, plants are being shut down to prevent or slow the propagation of the disease, not because of shortages. Restarting will most likely happen before the pandemic has run its course. Manufacturers will need to re-engineer production lines for contagion-proofing. Supply chain recovery issues will only take center stage once these safety-related issues are resolved.
Determine when to ramp up capacity
In the webinar, this is about choosing which products to make in which facilities during recovery. Simchi-Levi formalizes this as a linear programming problem. The goal is either to minimize the Time-To-Recovery or to maximize the Time-To-Survive. The constraints affect the inventory, the demand, the capacity, and the fact that one node in the supply chain is halted. He refers to the products as “vehicles.” It suggests that these formulations were intended for supply chain recovery at Ford in 2011, at the time of the Thailand floods and the Fukushima earthquake.
The COVID-19 pandemic situation, however, is not one where just one node in a supply chain is halted. The formulation would have to be modified for supply chain recovery with multiple nodes halted.
Allocate the available resources to products that allow you to achieve objectives during recovery
This is about executing the plan established in the previous step but isn’t addressed in the webinar. It is a large-scale effort, involving not only company operations but the supply chain as well, which raises many questions on how to lead this effectively.
Determine when to expedite and for how long
“Expediting,” in this context, means using air transport instead of ships. When the Time-To-Recovery and the Time-To-Survive are roughly equal, the supplier is again at full capacity when the customer runs out of inventory. With instant delivery, there is no stockout at assembly; with a 45-day ocean voyage between supplier and customer, there is. To avert it, the customer must plan and budget for the temporary use of faster methods.
If the COVID-19 pandemic receded everywhere at the same time, there would most likely be a surge in demand for air freight. Manufacturing is already restarting in China, while plants in North America and Europe are down. For supply chain recovery in such circumstances, the slow boats from China could deliver the parts in time for the assembly plants to restart.
Book logistics capacity as soon as possible
In 1993, Toyota’s Chicago consolidation center reserved all the trucking available in the area 3 days before the Mississippi flood. When water covered the railroads normally used to ship parts to the NUMMI plant in California, they were ready to switch. They could do it because the logisticians at other companies didn’t. The others were either oblivious to the weather or betting against the flood. If everyone had been as vigilant as Toyota, the surge in demand for trucks would have created a shortage.
With COVID-19, we are in uncharted territory. The methods developed to overcome earlier, smaller disasters are useful starting points but not solutions.