It is often difficult to explain how difficult it is to come up with an optimal solution to a vehicle routing problem.
I am convinced that part of the problem is that it is so trivial to come up with feasible solutions to a vehicle routing problem: just send out a truck to make every stop or group stops and send a truck to make multiple stops. And, an experienced dispatcher can often quickly improve upon a given schedule by looking at the routes. And, what we've found in practice, the constraints that would make the problem harder (like deliver time windows) are often simply ignored.
However, being about to find a solution to the problem (and maybe even an infeasible one), is not good enough. We've found that using advanced optimization can lead to routes that shave 5-15% of costs (on routes that are already deemed pretty good) and meet all the constraints.
What makes this interesting, is that the optimization is actually very difficult. Mike Trick's recent blog post highlights some of the difficulty with routing by writing about a Traveling Salesman Problem (TSP) a politician might face in Iowa trying to visit all 99 county seats in the shortest amount of drive time. Mike Trick quotes Bill Cook:
Leaving Des Moines, we have 98 choices for our second stop, then, for
each of these, 97 choices for the third stop, then 96 choices for the
fourth stop, and so on until we hit our last county and drive back to
Des Moines. Multiplying all of these choices gives the number of
possible routes through Iowa. And it is a big number, roughly 9 followed
by 153 zeros. No computer in the world could look through such an
enormous collection of tours one by one.
For vehicle routing (with multiple vehicles, capacities, time windows, and so on), the story gets more complicated. In a 2009 paper by Gilbert Laporte titled "Fifty Years of Vehicle Routing" discusses the progress that has been made in this field. He mentions that the vehicle routing problem "is considerably more difficult to solve than the TSP." Over fifty years of research has led to significant advances in different approaches and algorithms. But, he mentions that the field still has a long way to go to solve larger (and more realistically sized) problems found in industry.
This mirrors our experience with the ILOG optimization for vehicle routing-- these problems can be mathematically challenging. To solve these problems in practice requires an optimization-based approach. We have been working on this since the mid-90's. Some readers may remember the ILOG routing engine called Dispatcher or
the tool Transport PowerOps. This routing engine has found its way into
to the product called Transportation Analyst. And, the underlying
optimization engines are now part of the CPLEX toolkit.
At Smarter Supply Chains – Atlanta Regional Conference, Ronan O'Donovan, product manager with IBM ILOG Supply Chain Applications, spoke about "smarter" inventory and product flow optimization.
Key findings: The challenges around inventory management and product flow are not getting any less. Retail and distribution clients, for example, are requiring a product that optimizes both safety stocks and the flow of products to minimize total supply chain costs.
To juggle all the factors impacting inventory and product flow, companies must make inventory and product flow optimization a formal business process within organizations, front and center and part of the regular monthly planning process, e.g., Sales & Operations Planning (S&OP) or Sales, Inventory and Operations Planning (SI&OP). "Making up the numbers" is no longer acceptable as a planning method.
In addition, processes are moving from "strategic" to "tactical," meaning that planning is becoming an increasingly frequent exercise as companies look to be more responsive to the evolving environment. Companies need to understand, at any given moment, what's the best plan for me right now, given the current set of constraints we're facing.
Ronan discussed typical user workflows and roles for Inventory Optimization, breaking down the separate approaches used by typical Super Users (collecting/managing data), Business Analysts (running what-if scenarios, setting new inventory targets) and Manager Reviewers (reviewing exceptions/alerts, approving/overriding recommendations, tracking planned vs. actual and other KPIs, and reporting). The key to a successful process is not necessarily how each of these classes of users performs their job, but how these users interact through a consistent, closely aligned process.
Where organizations put optimization at the heart of the process, the technology tends to follow to enable that process. Note that this process will evolve and mature over time within a company, as the company learns which factors to incorporate and how best to incorporate them. Ultimately, optimization will never be an "easy button" that will make everything work automagically, but it's a necessary and critical process.
Supply Chain authority Andrew Reese is Editor of Supply & Demand Chain Executive. He has been invited by IBM PR to attend this show as a blogger and speaker. Like all other speakers, Andrew will receive all speaker benefits including travel and board.
As previously noted, inventory optimization is a core capability for a firm that wants to perform at the highest possible level. Inventory is a lifeline of a firm. Having inventory in the right place allows you to meet demand and allows your business to operate smoothly. Done right, an inventory optimization processes can reduce inventory by 10-30% while improving service levels. For a large firm, this can translate in over $100M in inventory reduction. This is a one-time increase in cash flow as well as the on-going opportunity cost of this capital.
The technology to do this, provided by a tool like IBM ILOG Inventory and Product Flow Analyst, allows you determine the correct inventory strategies and levels. It does this based on data you are likely already collecting-- demand, demand forecast error, supplier lead times, supplier variability, order frequency, minimum order sizes, flows through the supply chain, and other factors. You have to get this piece right to get the buy-in of the people in the organization. When people see that the technology suggests strategies and inventory levels that allow customer demand to be met while maintaining service, they are more comfortable accepting the outputs.
But, to make the technology (and savings) stick within the organization, you need a strong process as well. We have found that it is important that the technology fit within the organization. This means that the inventory planners see information that is appropriate to them while the business analysts and managers may see a different view. When we implement an inventory optimization solution across the enterprise, we make sure the workflows are correct for each user, we make sure the solution fits within an overall larger closed loop planning cycle (which may span the year), we make sure the process is dynamic to allow you to adjust as the market changes, and we have even developed inventory optimization "playbooks" that help customers through the entire planning cycle.
Previously, Indeval, like most CSDs around the world, operated a
settlement system that required banks to hold liquidity of billions of
dollars while securities were being settled. Linking the delivery of
securities to their corresponding payment requires depositors to have
adequate financial resources available to settle their trades. Financial
institutions may have to borrow if they do not have sufficient funds to
settle stock and debt trades.
At the heart of
Indeval's Dali securities settlement system, the IBM ILOG CPLEX
Optimizers match thousands of transactions simultaneously, so that only
net amounts of securities and cash need to be transferred among the
participating financial institutions. This tremendously reduces the
amount of cash and securities the institutions need to have on hand to
settle the transactions.
Indeval won the 2010 Edelman Award from INFORMS (Institute for Operations Research and the Management Sciences) for this work.
Many S&OP processes stop with the sales planning. When you extend your S&OP process to include optimization through a tool like LogicNet Plus XE, you can reap many benefits.
The picture to the left is shows some highlights from an S&OP case from a beer company. On a monthly basis, as part of the S&OP process, the operations of this beer company are modeled and optimized in LogicNet Plus XE. The details below the map provide information on how much beer, by product by month, is shipped to each of warehouses and how much product is produced in each month. The graph shows the inventory build up in the supply chain by warehouses.
Besides providing insight into the operational cost and capacity implications of a given sales plan, the results also show the impact on total revenue, operational gross profit, and total pre-build inventory.
These results can also provide insight into the sales plan. For example, it can determine the cost and feasibility of promotions, it can suggest additional demand shaping activities by looking at under-utilized capacity.
In this case, the firm analyzed different demand plans around the implications of a marketing program to grow demand in a region, the implications of a promotion for a specialty product, and the implications of a price increase.
Bajaj, Head – Solutions and SCM Strategy at AFL’s Logistics Division
said: “AFL chose IBM’s iLOG software for several reasons, chief among
them – robustness, post-purchase support and a clever user interface
that makes data modelling both simple and secure. This unique
combination of AFL’s domain expertise in logistics and iLOG’s superior
design and modelling capabilities, have helped us unlock significant
value in our clients’ supply chain metrics.”
At this point in the holiday season, most supply chain professionals in the retail space are probably doing a lot more reacting than planning. Keeping the shelves stocked is likely priority #1,2, and 3, while things like cost and efficiency are farther down the list (if even on the list). With that said, I apologize for not writing this post 6 months ago when you may have been able to act, but perhaps this can be helpful for next year.
The holiday season provides many interesting optimization problems for those working in the retail supply chain. For many retailers, SKUs have short lifecycles and high variability. With these characteristics, it makes sense to try to centrally manage inventory in a small number of distribution centers. This allows them to risk pool the demand variability, keeping inventory and working capital as low as possible. During the non-peak season with lower sales volumes, regular (say weekly) replenishment can be planned in order to achieve transportation efficiency. Unfortunately, when the peak holiday season arrives, this strategy may not be ideal. Stores have limited ability to store product and sales velocity is high. With the goal of keeping the shelves full, they need frequent replenishment during the holiday season. If the retailer manages only a few DCs, there will be stores who are quite far away from the DCs, making it impossible (or very expensive through use of air freight) to replenish the store as frequently as necessary to prevent stockouts. Therefore, its quite possible a different distribution strategy may be called for during the holiday season. It may make sense to add seasonal distribution facilities to the network to get more product nearer to the stores. This type of strategy involves many decisions which all can be helped by using optimization:
How many seasonal distribution points do I need?
Where should they be located?
Which stores are assigned to which DC?
Which SKUs do I deploy to the seasonal DCs and which do I carry only at my regular DCs?
How often to I replenish my seasonal DCs and what type of transportation mode do I use?
How do I deliver to stores from the seasonal DCs (mode, route, etc)?
A good strategy ahead of the holiday season is likely to result in less headaches and less costs when the season actually arrives.
At a the recent Manufacturing Leadership Summit, Toshiba gave the "Manufacturing Leadership 100 Award Winner Case Study Presentation." The talk was titled: "Using SCM to Create Competitive Advantage."
In the talk, they mentioned that they used IBM ILOG LogicNet Plus XE to model different options. Among the improvements that they saw, the cost per shipment decreased by more than 12%. In addition, obsolescence reduced by more than 40% and missed service calls reduced by more than 16%. This all helped lead to double digit growth in customer use of Toshiba Supply Chain Services.
As we've discussed before in this space, retailers are facing growing pressure from the rise of a smarter consumer. To keep up, a retailer has to do business in a smarter way. At IBM, we are calling this Smarter Commerce.
One area where retailers have invested is in the ability to deal with customers seamlessly across multiple channels. That is, a customer must have the same experience whether shopping in your store or on your website and should be able to move between the two with ease. IBM's Sterling Order Management is one way that retailers make this happen. Sterling Order Management allows you to orchestrate cross channel selling and order fulfillment. That is, it allows you to order on-line and pick up in the store.
With this increase in flexibility, it creates the need for additional inventory optimization capability at the store. For example, besides the foot traffic, a store also needs to have the inventory for customers who order on-line and pick up in the stores. Since this can potentially lead to additional sales when the customer is in the store, this may create the need for more inventory (and hopefully more sales).
In addition, some retailers are looking to ship on-line orders from the store. The idea is that you can allocate orders to the appropriate store and then ship it to customers the next day using standard ground service. And, the second benefit is that is could be beneficial to have this extra inventory in the stores. That is, if the store has an uptick in demand, the store will be in a better position to capture the extra sale. Of course, these benefits must be traded-off against the cost of fulfilling in the stores, and the stores must have the space for this.
To make this work is to pick the stores where you will ship on-line orders (you might not want to pick every store) and then put in place a good inventory optimization strategy. These stores will need extra inventory buffers to handle the increase in volume. The distribution centers that support these stores will need a new inventory strategy. And, the retailer will need to review the inventory strategy on a more regular basis. Click here for an article on the importance of a good inventory optimization strategy.
As an added benefit, as your inventory strategy is fed into Order Management, you can build smarter rules for how to replenish when you are out of stock at one location. You don't want to take inventory from a secondary location today, only to run of inventory at that location tomorrow.
Today, we announced the release of a new version of our multi-echelon inventory optimization product, Inventory and Product Flow Analyst (IPFA) v7.6
We have put a lot of investment in IPFA so that it fits within the workflow of an organization. That is, it is important to be able to answer strategic inventory optimization questions, but it is even more important to maintain the right inventory targets. You maintain these targets with strong work flow capabilities.
By maintaining the right inventory targets at all levels in the supply chain, you can avoid lost sales, minimize late shipments, minimize expediting, and keep inventory to a minimum. And, since IPFA directly accounts for demand forecast error and supply variability, you get value even when your forecasting is not very robust. For more details on the strategic importance of inventory optimization, click here.
Highlights of this new release include:
Simulation capability: Inventory planners can now analyze inventory and service level performance against simulated customer demands.
More powerful Multi-Echelon optimization: Inventory and
supply chain planners can now optimize total inventory costs based on
variable committed service times or variable upstream service levels.
Forecast error and supplier performance analysis: Inventory
and supply chain planners can now analyze forecast error and take into
account forecast bias in the inventory optimization process. They can
also measure supplier performance with an automated calculation of lead
time variability and receipt period.
Ease-of-use enhancements: Enable inventory planners to perform their tasks more effectively.
The IBM Smarter Supply Chain Management briefing is coming to Toronto on March 4th, 2010! Save the date!
Join us on March 4 at the Intercontinental Toronto Centre's Caledon
Room, to learn how your business can take advantage of these proven
supply chain strategies and tactics to help improve bottom line profits:
1)Optimizing S&OP Processes - Generate real value to the business - Create significant improvement in KPIs 2) Leveraging Network Design to Create Lean Supply Chains - Help reduce costs and better manage risks - Develop efficient production and scheduling systems as more products are moved to the same plant 3) Improving Operational Efficiency - Optimize the production process to improve manufacturing efficiency and reduce inventory - Use a dynamic safety stock approach to improve supply chain flexibility
This one-day event for business executives and supply chain
practitioners will focus on both the strategic and tactical decisions
that are impacted by the current business environment. Thought leaders
from industry and academia will discuss the challenges and
circumstances that are driving the need for business solutions that
optimize overall supply chain costs, free up working capital, and
improve service levels.
According to an article in Businessweek, "Companies from Tiffany & Co. to Home Depot Inc. are restocking
shelves in a move that will boost economic growth and may keep the
recovery on track through 2010."
The questions many retailers are asking is "what should I restock with?"
Today, we recorded an educational webinar with SC Digest on shelf space optimization-- a great way to answer the question about how to stock your stores. The objective is to keep customers happy at every single store with the right products and, in turn, drive up revenues and profits.
SC Digest will keep the Webinar available for 12 months for viewing. You just need to register to watch it. Click here for the link.
Next Session: December 11, 2013 at 07:30 Pacific, 10:30 Eastern, 16:30 CET
Speakers: Mary Fenelon, Development Manager, CPLEX Optimization Studio, IBM
Session Topic: What's New in CPLEX Optimization Studio 12.6
A new release of CPLEX Optimization Studio will be generally available on December 6. In this presentation, you will learn about the new features and performance enhancements, including:
Improved solution times on difficult mixed-integer problems
Improved solution times on scheduling problems
A new algorithm to provide a global optimum for problems with non-convex quadratic objectives
A new distributed parallel algorithm that harnesses compute clusters to solve mixed-integer problems
Constraints to better model ordering relationships between operations and to easily specify highly combinatorial relationships
Reorganization of CPLEX Optimizer parameters into a functional hierarchy
New code assist functionalities in the IDE that help in correctly writing constraints and related structures
An LP-format viewer in the IDE to aid in model debugging
IBM Decision Optimization Virtual User Group Meetings provide you with updated and useful information about our products and how you can get the best value for your organization from our optimization technology and solutions.
These meetings are being held electronically with presentations by our product management and development subject matter experts. The sessions will run for approximately one hour, with about 45 minutes of presentation material and 15 minutes for open questions and answers. Our goal is to provide a forum for you to:
make smarter decisions
get your questions answered
elicit your feedback on various topics.
We look forward to your participation.Please contact Kitte Knight email@example.com an invitation to this Webinar.
ILOG Supply Chain applications for network design, inventory optimization, and transportation optimization
ILOG Optimization solutions for unique scheduling and planning applications
Sterling TMS for SaaS for transportation management
Sterling Order Management and Supply Chain Visibility for distributed order management and visibility. This, for example, enables retailers to better handle their multiple channels and gain control of their supply chain
Emptoris for strategic supply and contract management
DemandTec for trade and promotion management and optimization
Cognos and SPSS for supply chain descriptive and predictive analytics. For example, Cognos and SPSS have been deployed for value added S&OP.
On April 26th, Sterling Commerce, an IBM Company is sponsoring an Aberdeen Group webinar on Inbound Transportation Management: The Convergence of Planning, Optimization, and Dynamic Execution. Register today!
Inbound Transportation Focus –
60% of companies from our July study have realized the significance of gaining
more visibility and control of inbound transportation.
·WalMart’s Inbound supply chain collaboration
initiative is one additional factor driving change- (Kelly Abney, WalMart VP Corporate
Transportation presented a keynotes on this initiative during Aberdeen’s 2010 and 2011 Supply Chain
·Gaining visibility and control of inbound
transportation management is prerequisite and central to…
the success of the WalMart
initiative and is also prerequisite and
extends to companies of all
sizes, scale, and industry segments. It is crucial to the success of any
multi-party supply chain management transformation.
·Organizations lacking inbound control are blind to
shipment details within the transportation pipeline.
how leading companies (those with highest percentages of Inbound Freight under
control) are focusing on new Inbound Technologies to address challenges and take
action. They are:
·2.8 to 5.8 –times as likely to leverage enhanced
–leading companies are converging planning processes with robust automation to
support on-the-fly decision-making
·1.2 to 2.5 –times as likely to automate dynamic optimization
and near real-time visibility with optimization of execution within their processes.
and as we look for insights around best practices and review specific examples
of strategic technology advancement around inbound collaboration.
About Our Speakers:
Bob Heaney, Senior Research
Analyst Supply Chain Management, The Aberdeen Group
Bob Heaney is a seasoned professional with over 25 years of distinguished
leadership experience in research, analysis, and advisory roles in Supply Chain
Engineering. Bob’s coverage area within Aberdeen
includes various elements of Supply Chain Execution (Transportation Management,
Warehouse Management, Distributed Order Management and Supply Chain
Wharton, Manager of Product Marketing, Sterling Commerce, an
As a software industry veteran, Peter has worked for a number of software
vendors of supply chain solutions, including managing the carrier relationships
with UPS, FedEx, DHL and USPS while Director of Strategic Alliances for Kewill.
Peter started his career in development, designing distribution and inventory
systems in England
working for Imperial Tobacco. He attended the University of East Anglia,
where he achieved a joint honors degree in Environmental Sciences and Computing.