Automotive and defense analytics

A brief description of select completed or ongoing analytics projects in the automotive and defense sector is provided below. If you need further information, please contact us.

Warranty analytics: Exploiting upstream supply chain data for enhanced reliability issue predictions

The automotive industry spends roughly $10–$13 billion per year in the U.S. on warranty claims and up to $40 billion globally, consuming roughly 1–5.2 percent of original equipment manufacturers’ (OEM) product revenue and roughly 0.5–1 percent of suppliers’ product revenue. To help mitigate and better manage these costs, we have developed a statistical methodology to construct models for early detection of reliability problems using information from warranty databases, and most importantly, upstream supply chain.

Upstream supply chain data sources for improved anticipation of reliability and warranty issues

This is contrary to extant methods that are mostly reactive and only rely on data available from the OEMs. In particular, we propose hazard rate models to link upstream supply chain quality/testing information as explanatory covariates for early detection of reliability problems. In doing so, we improve both the accuracy of reliability problem detection as well as the lead-time for detection. The proposed methodology is partially validated using real-world data from a leading global Tier-1 automotive supplier.

Product assortment planning analytics: Balancing market needs and supply chain costs/complexity

A manufacturer’s assortment is the set of products that the company builds and offers to its customers. Assortment planning involves finding an assortment that maximizes company’s profit subject to various constraints such as limited budget to purchase products and limited shelf space to display products. Assortment planning requires a tradeoff between sales revenue and product offering costs for the company. Growing awareness for the costs associated with increasing manufacturing complexity and plant productivity issues under large product configuration assortments is compelling major volume-driven automotive original equipment manufacturers (OEMs) to consider controlling their configuration variety to decrease their operational costs while maintaining their sales and market shares. However, industry is lacking effective decision support tools to manage this process, in particular, for configurable products such as automobiles.

Functionality of assortment planning tools developed for the automotive industry

We are developing state-of-the-art industrial strength decision support and planning tools in collaboration with a major U.S. automotive OEM for product assortment planning. The tools can optimize the product assortment, plan target manufacturing capacities, and integrally optimize the supplier network configuration. The tools are also designed to support sensitivity/scenario analysis, technology selection, regulation compliance (e.g., CAFE requirement implications), and bundling of features/options.

Freight logistics analytics: Dynamic routing of just-in-time cargo under ITS

The development of efficient algorithms for freight routing on time-dependent networks is one of the major challenges under Intelligent Transportation Systems (ITS). Vehicle routing navigation systems, whether built-in or portable, lack the ability to rely on online servers and, given an origin/destination pair and departure time, must compute the route in a stand-alone mode with limited hardware processing/ memory capacity. We are developing computationally efficient yet effective hierarchical search strategies and algorithms that exploit community-structure, detection-driven hierarchical network representations to solve the time-dependent shortest path (TDSP) problem. The algorithms are efficient in terms of finding shortest paths in milliseconds for large-scale road networks while eliminating the need to store preprocessed shortest paths, shortcuts, lower bounds, etc. We conduct this research with support from the U.S. DOT through the local UTCs and in collaboration with ITS data providers (, NAVTEQ) and freight carriers and consignees. Performance of the proposed algorithms using data from Detroit, New York, and San Francisco road networks is very promising. We are currently extending the methods to support hybrid vehicles and pure electric vehicles. Methods are also being developed to manage dynamic routing of cargo on intermodal networks under ITS, in particular, air-road intermodal transportation.

Supply management analytics: Value stream management of commodity supply networks

Commodity business planning groups within automotive companies are lacking effective tools to evaluate both the cost and risk of existing and potential supply chain configurations. In collaboration with a leading U.S. OEM, we have developed such a Supply Network Analysis Package (SNAP) tool. The tool breaks down value chain costs into categories such as shipping, holding, racks, sequencing, and tariff.  It also includes costs associated with risk events like an engineering change or a damaged shipment.  The SNAP toolkit permits a total cost comparison of one value chain against another for effective and proactive decision making.  SNAP toolkit also enhances current value chain mapping processes by utilizing standardized inputs with tabular and graphical outputs. It allows the non-expert user the ability to configure a variety of scenarios to understand the cost impact due to factors including changes in supplier, supplier facility location, mode of transportation, inventory levels, and engineering changes. The tool has been validated and is currently deployed for use by a leading U.S. automotive OEM.

Remanufacturing analytics: Efficient management of supply and demand in closed-loop supply chains

In today’s global economy, firms are seeking any and every opportunity to differentiate from competitors by reducing supply chain costs and adding value to end customers. One increasingly popular option, under growing consumer awareness and increasing legislation, is to reintegrate returned products into the supply chain to achieve economic benefits as well as improve sustainability. An important class of such “reverse” goods flows has to do with remanufacturing (reman), which refers to activities that restore returned products or their major modules to operational condition for using in place of new product or distributing through other channels (e.g., spare parts). While opportunities abound, key complications are: 1) difficulty in timing the launch of reman product (while accounting for uncertainties associated with product life-cycle demand and core supply), 2) difficulty with capacity planning for remanufacturing (while accounting for the fact that volumes can be low and that facilities/lines should target multiple product families for economies of scale), and 3) operational difficulties in maintaining efficiencies in production planning and control of remanufacturing activities. These difficulties are mostly attributable to limited visibility and higher levels of uncertainty in reverse logistics (in comparison with forward logistics). Despite advances in the remanufacturing literature in the last two decades (both in the academic literature and practitioner community), there is no integrated decision support framework that can guide companies to successful launch and execution of remanufacturing operations. This is particularly true for companies that engage in both original equipment (OE) service as well as the independent after-market (IAM) in the automotive industry. In collaboration with a leading Tier-1 global automotive supplier, we have developed a comprehensive decision support framework and necessary models and algorithms for effective remanufacturing in the automotive industry.

Supply chain and logistics network analytics: Inbound and outbound logistics network design and optimization in automotive supply chains

Efficient flow of parts, subassemblies, and finished vehicles are critical for the competitiveness of automotive OEMs and their suppliers. While the just-in-time logistics has been well-known and strived for in the automotive industry, there still exist many challenges and opportunities to improve the logistics costs and service levels. Uncertainty in the retail side as well as fluctuating transportation costs are two examples of the ongoing challenges. In addition, due to the network structure, the automotive industry has an untapped opportunity in terms of collaborative logistics where multiple OEMs and suppliers can consolidate their parts and vehicle shipments to reduce costs and improve deliver lead times. We have developed strategic, tactical and operational models and algorithms that enable the OEMs and suppliers to evaluate collaborative logistics opportunities and determine novel and innovative ways to exploit the opportunities. Case studies using two major OEM’s US finished vehicle logistics network data, we demonstrated that potential savings are in excess of tens of millions of dollars annually by only considering short and medium-term collaboration in the outbound logistics. More savings are realizable with longer term alignment of the network with shared and co-owned facilities. In addition, with reduced dwell times and cost per finished vehicle, we demonstrated that OEMs can reduce the in-transit inventory cost and increase the finished vehicle availability at dealerships and thereby reduce the lost sales due to stock-outs. We are currently investigating ways to extend the developed models and methods for similar gains in the inbound logistics collaboration among Tier 1 suppliers.

Supply chain disruption analytics: Managing supply disruptions through dynamic recovery strategies

With increased globalization and shorter product development cycles, supply chain disruptions have become more evident and wasteful. Many companies facing disruption risks are resorting to such strategies building flexibility and resilience in their supply chains. However, these two strategies often correspond to higher costs and lower service levels. For instance, to hedge against supply disruption, a frequently used strategy is do source a component from at least two suppliers which corresponds to reduced economies of scale in procurement costs. Instead of building costly flexibility and resilience in their supply chains, companies can develop advanced predictive analytics and dynamic resource allocation capabilities to respond to the disruption in a more effective and cost efficient manner. We have developed predictive models and dynamic optimization methods to enable the companies facing supply disruptions quickly mitigate the impact of these disruptions and recover more quickly. While this way of coping with disruptions require contingency plans and visibility solutions, their requirements in terms of redundancy and flexibility are greatly lower thus provide significant cost advantages. We implemented the developed approaches in a case study involving a supply disruption scenario of an automotive OEM which is motivated by a real-world disruption event in 2005. By exploiting the vehicle commonality and components substitutability, we were able to demonstrate that the OEM could have responded the supply shortage with minimal losses and recover profitability quicker.

Product design analytics: Multi-criteria design optimization of configurable products

Design optimization of configurable products with is an important problem in many industries including automotive and defense. The challenge arises because of the presence of multiple conflicting criteria (cost, performance, reliability, etc.) and the high cost of the decisions. Given the complexity of today’s vehicle systems, the number of possible system design alternatives is very large. Further, the complex relationships between the subsystems and options are mathematically difficult to model and thus make the whole system level optimization a very challenging task. We have developed novel models and algorithms to efficiently capture the subsystem level interactions and tackle the computational complexity. The developed optimization methods are able to generate the entire Pareto solution set or an approximation set that is highly representative of the entire Pareto frontier. We have also developed trade space exploration and visualization tools to help the designers to close-in on the final configuration set to be developed or guide the design optimization iterations with more information on the trade-offs. We have implemented the developed models and methods in real-world design problems in the defense industry.