Design for the Supply Chain EQUATIONS for Online Educational Course http://design4X.com Instructor: Dr. Mark Martin 1 © design4X 2005 © design4X http://design4X.com 1 Agenda » Introduction • • Design for the Supply Chain Pre-Test » Step 1: Know Your SC • • • • Think About It What is a Supply Chain? Knowing your SC Apply It » Step 2: The Costs of SC • • • • • • • • • Commonality Postponement / LPD Shipping Logistics Apply It » Summary • • • Example - DVD-RW dfSC Assessment Think About It Costs and the Supply Chain Types of Inventory Calculating SS Inventory (optional) Apply It » Step 3: Designing for the Supply Chain • • Think About It The Five dfSC Strategies 32 © design4X http://design4X.com 32 What does inventory cost? » Annual inventory costs = (20% to 30%) x (Cost of goods of the inventory) » Scenario: Macro Calculation • • • • • • Sales: $1 billion (company or division) Net profits: $60 million (6%) Average Inventory: $100 million (10% of sales) Cost of that inventory per year: 30% ($30 million) Super engineer arrives: Reduces inventory by 20% Savings: up to $6 million Company profits just increased by 10%! © design4X 37 http://design4X.com Many companies often assign an annual inventory cost factor of between 20% 30%. This is an estimate of all the costs associated with inventory - opportunity, storage, material handling, insurance, theft, and obsolescence. So let’s run a quick calculation, imagine this: You’re division does $1 billion in sales. A good net profit for a company this size might be $60 million (or 6% of sales). Let’s say the company has $100M dollars of inventory, and assuming a 30% annual inventory cost factor, it’s costing the company $30 million per year. Now, if you can reduce inventory by 20% (which is not undoable), then you’ve saved $6 million. That’s 10% of the company’s net profit. Imagine being able to increase your company’s net margins by 10%. If you’re a public company, this could add up to 10% to the stock price. 37 What does inventory cost? » Micro Calculation • How much does inventory cost per unit? • For a final assembly or component in incoming inventory - Measure inventory in “Weeks of Supply” - Inventory Cost per Unit (ICU) Inventory Cost per Unit (ICU) = X * c * ( h / 52) X = average # of weeks of inventory kept in stock c = cost of goods per unit h = annual holding cost % (typically taken at somewhere between 20% - 30%) - Example: X = 3 weeks of inventory c = $100 h = 30% ICU = (3 weeks of supply) * $100 * (30% / 52) = $1.73 38 © design4X http://design4X.com We just finished a “macro” calculation of how much inventory can cost the company. Now let’s do a micro-calculation and calculate how much inventory can cost for a particular component or product – anything from a printer, a car, a power supply, or even small components such as a pump or a resistor. Let’s calculate our costs in terms of weeks of supply of a product. In other words, let’s find out how much it costs to keep X weeks of supply of something in stock. One week of supply of the product is defined as the average weekly sales of the product. So if you sell 1000 units a week on average, and you keep 2000 units of inventory in stock on average, you have two weeks of supply. This micro-calculation is simple, multiply the number of weeks of inventory, X, by the cost of goods of the product or component, c. You then multiply this number by the annual holding cost percentage, as mentioned before, many companies use between 20%-30%. This is divided by 52 to get it in terms of weekly percentage costs. This is how much your company is paying per unit in inventory costs. As you can imagine, this can add up quickly. Imagine you had a product, such as a printer, that costs $100 and an annual holding cost of 30%. If average sales are 1000 units / week, and the company has 3000 units in inventory (in FGI, in-transit, and in distribution centers. That’s 3 weeks of inventory of just the product. Your cost per unit is 3 weeks of supply x $100 x 30% annual holding cost / 52 = $1.73 per unit! That’s just the cost of holding the final goods inventory! Now, imagine that you’re also keeping 3 weeks of inventory of the power supply for that product on hand. If that power supply costs $10, then that is costing you an additional $0.17 per unit, and the same goes for all the other components. You can see that if you could cut that inventory in half, you would start to see some significant savings. This is where this course can help. By designing for the supply chain, you can reduce inventory and save money. 38 Agenda » Introduction • • Design for the Supply Chain Pre-Test » Step 1: Know Your SC • • • • Think About It What is a Supply Chain? Knowing your SC Apply It » Step 2: The Costs of SC • • • • • • • • • Commonality Postponement / LPD Shipping Logistics Apply It » Summary • • • Example - DVD-RW dfSC Assessment Think About It Costs and the Supply Chain Types of Inventory Calculating SS Inventory (optional) Apply It » Step 3: Designing for the Supply Chain • • Think About It The Five dfSC Strategies 48 © design4X http://design4X.com 48 Types of Inventory » Pipeline inventory • Mean lead time: L • Mean demand: µ Pipeline inventory Punits = L ⋅ µ 52 © design4X http://design4X.com The concept of pipeline inventory is simple. When you order incoming components from your supplier, they ship these parts to you. These parts are “in the pipeline” and if you own them during this time, they are part of your pipeline inventory. The projected amount of this inventory is easy to calculate. It’s simply the average demand of your product that you’ll be ordering from your supplier, multiplied by the lead time to ship to your factory. So, if it takes two weeks to ship, and your average weekly demand is 2000 components/week, you would expect that at any moment in time, you have 4000 components in the pipeline. Of course, this number might vary based on how often the company places an order, what the minimum order quantities are for your supplier, etc. This pipeline inventory calculation works on the other end of the supply chain as well. If you’re shipping 2000 units / week to your customers or retailers, and it takes two weeks to ship. There are 4000 units in the pipeline. Of course, if your customer pays for them once they ship, then in essence that pipeline inventory is now owned by them, so it doesn’t figure into your cost calculations. 52 Types of Inventory » Pipeline inventory » Cycle stock inventory for: • Incoming • FGI Cycle stock inventory CS = 1 2 µ f µ = average demand (during time period) f = delivery frequency (deliveries during time period) 53 © design4X http://design4X.com The next inventory is called cycle stock. It varies with how frequently you receive incoming inventory, and, how often you deliver FGI to your customers. Here’s the concept, if you get weekly shipments from your supplier, then they are typically delivering the amount of product you expect you will need for that next week. So if you expect to build 2000 units, you’ll order 2000 components. When the components arrive at the beginning of the week, assuming you build on a steady schedule, the cycle stock for that component will decrease linearly until you use them all up at the end of the week. So what’s the average level of cycle stock inventory you have? Well, in our example you start out with 2000 units, and end up with zero units. Since the decrease is linear, then the average amount you have on hand is ½ of what you started out with, or 1000 units. Now what happens if you decide to have the components delivered twice a week as needed, then you’ll have 1000 components delivered each time. You’ll start out with 1000, and end up with 0, and your average stock level would be ½ of that, or 500 units. If you decide to have delivery five times per week, that would be a delivery of 400 units each time, and your average inventory would be 200. Do you see the pattern? To get the average amount of cycle stock your company would keep, take the average demand during a time period, which is ½ mu, and divide that by the delivery frequency during that time period. The same is true on the other end of the supply chain, if you deliver to your customers once a week, you are building up that inventory over that week and will have an average cycle stock inventory of ½ mu. If you deliver more frequently, you’ll reduce the amount of inventory. That’s part of what JIT manufacturing is doing, it’s delivering inventory more frequently, and thus reducing your overall inventory levels. Then why not just deliver more frequently (say once a day, or even more frequently). Some manufacturers do this. In automotive, some components are delivered, from suppliers, every four hours. Of course, the reason this isn’t done by everyone is that more frequent shipments can cost more. There are more transactions occuring, the logistics are more complex, shipping costs can be more expensive, and, if problems occur – such as truck breakdowns, union strikes, or major catostrophes, natural or man-made, you risk not getting parts in time. These are some of the trade-offs that must be considered when determining delivery frequency. 53 Agenda » Introduction • • Design for the Supply Chain Pre-Test » Step 1: Know Your SC • • • • Think About It What is a Supply Chain? Knowing your SC Apply It » Step 2: The Costs of SC • • • • • • • • • Commonality Postponement / LPD Shipping Logistics Apply It » Summary • • • Example - DVD-RW dfSC Assessment Think About It Costs and the Supply Chain Types of Inventory Calculating SS Inventory (optional) Apply It » Step 3: Designing for the Supply Chain • • Think About It The Five dfSC Strategies 55 © design4X http://design4X.com 55 Calculating Safety Stock Inventory » Pipeline inventory Safety stock inventory » Cycle stock inventory SS units = k ⋅ σ 2 ⋅ (L + R ) σ pooled = ∑σ 2 i i » Safety stock inventory • Variation in demand • Variation in lead time Safety stock factor: k Standard deviation demand: σ Mean lead time: L Review period: R Note: all time units must be the same 58 © design4X http://design4X.com So, let’s show you how to calculate safety stock inventory that your company should keep. Here’s the math. And now on to the next section! Just kidding, we’ll step through this to help you understand the details. 58 Safety Stock Factor Safety stock inventory SS units = k ⋅ σ = k ⋅ σ 2 Safety stock factor: k Std dev demand: σ 0.5 Cum 99% = 98% = 95% = 90% = 84% = 50% = 0.45 0.4 0.35 0.3 “k” 2.326 2.054 1.645 1.282 0.994 0.000 1 0.9 0.8 0.7 0.6 0.25 0.5 “k” is the safety stock factor 0.2 0.4 0.15 0.3 0.1 0.2 0.05 0.1 0 0 -3 -2 -1 0 1 2 3 Safety Stock = 2.326 x 200 units = 466 59 © design4X http://design4X.com To understand our safety stock equation, let’s start off with a statistical curve indicating the average demand for a product (let’s say for a weekly time period), and the variation in that demand. We’re assuming you have some basic statistics under your belt, so we won’t go into the details here of this Gaussian distribution. The average weekly demand in this graph is indicated with the red line. This being a Gaussian distribution, that means that if you kept this amount of stock on hand, then 50% of the weeks you would have enough stock to meet demand, and for 50% of the weeks, the demand would exceed this amount and you would “stock-out”. Now, if you want to satisfy more than 50% of the demand, you’ll need to keep more stock on hand. Any amount you keep above the average demand is called safety stock. Let’s say you want to have enough stock on hand to meet demand for at least 95% of the weeks. Or for 99% of the weeks. How do you calculate that? It’s simple because we’ve assumed that weekly demand is a Gaussian distribution. In that case, all you need to do is multiply the weekly standard deviation of demand by a statistical factor called “k”, and that’s the amount of extra stock, above the average demand, that you keep on hand. This k factor, or what we’ll call the “safety stock” factor is shown in this table. So, to meet 95% of the demand, you would multiply the standard deviation by 1.645 to get the amount of safety stock needed. To meet 99% of the demand, you multiply by 2.326. These numbers are statistical constants and are based upon statistical theory. So, imagine you had a weekly demand of 1000 units, with a standard deviation of say 200 units. If you want a 99% service level, then you would keep 2.326 x 200 units, or 466 units of safety stock on hand. This is in addition to the 1000 units you’ll need to meet the average expected demand. 59 Leadtime and Review Period » But what about those lead times! » What do you do? • Have enough safety stock on hand to cover you during that lead time • How to calculate? Adding together standard deviations σ = σ 12 + σ 12 + σ 12 + σ 12 since all the sigmas are the same σ = LT • σ 1 2 L T = leadtime for the part σ = ( LT + R )σ 1 2 » Review period » SAFETY STOCK R = review period Safety stock inventory SSunits = k ⋅ (LT + R )⋅ σ 2 Safety stock inventory SSunits = 2.326 ⋅ (4 + 2) ⋅ 2002 © design4X = 1140 60 http://design4X.com So in our example we want to keep 466 units of safety stock on hand, at all times, to ensure that during 99% of the weeks we will be able to cover all demand. If one week we dip into the safety stock by a hundred units, then next time we order, we need to order a hundred extra units above our weekly average demand order to bring that safety stock back up to 466. Ahh, but what happens if the standard lead time is longer, say, 4 weeks? For the next 4 weeks, until you get that safety stock back up to 466 you have less than 99% chance of being able to meet the weekly demand. In fact, with 366 units on hand, that corresponds to a safety stock factor of 366/200 = 1.83, which just puts us slightly above a 95% service level.So what are we to do? Well, we can account for that longer lead time in our calculations. If it takes 4 weeks to get parts in, we want to calculate the standard deviation of demand during that time. We can do this by adding together the weekly deviations. Now here’s where the statistics training comes in, when you add standard deviations together, you don’t simply add them linearly. You first add their variances, which is the standard deviation squared, then you take the square root of that. Now, since the standard dev are the same for each week, this simplifies to the number of weeks of lead time, designated as L-T, times the sigma squared, square root of all that. And one other thing, what if you only check on inventory, or order every couple of weeks, then there might be another couple of weeks before you get those parts on order. That would be a total of 6 weeks for our example. To account for that, we include this review period in the std. dev calculation. In our equation, we designate it as R. Of course, if the review period is continuous and you order as soon as your safety stock is tapped into, then R=0. As an example of all this, imagine the you have a four week leadtime for the part and a two week review period. The standard deviation of demand over that 6 weeks, is the square root of six times the standard deviation of the weekly demand. For our previous part, the standard deviation was 200 units over a week, for six weeks, it is 6 x 200 squared, square root, which gives a standard deviation of 490. If you multiply this by the safety stock factor for 99% service levels, which is 2.326 you require a safety stock of 1140 – quite a jump over the 466 units you required with just a one week LT, and no review period. 60 Reducing Safety Stock Inventory – Rough Cut Analysis » Scenario: 9 power supplies for product line » Strategy: Create a common PS (go from nine to 1 PS), you could reduce inventory by 66%! » In fact, the formula for this calculation, with the stated assumptions, is % inventory reduction 1 1 % reduction = 1 − = 1− = 66% n 9 n= original # of unique parts reduced to 1 common part Assumes: Sales volumes are approximately equal for each part; Demands are independent and uncorrelated; Leadtimes are known © design4X 61 http://design4X.com So, this is all interesting (well, sort of), but what does this matter to you? Well, as mentioned before, safety stock inventory costs the company money, and if you can reduce the standard deviation of demand, you can reduce inventory. Creating commonality does this. Now, we’re going to “boil” down the previous equations to show you how commonality can impact inventory. Imagine you had nine different types of power supplies you were using in your products. Assuming that the sales volumes were about the same, as well as the variations in demand, and assuming these variations are independent, if you could create a common PS (go from nine power supplies to one), you could reduce inventory by 66%! That means if you originally needed to keep 1000 units of safety stock inventory for the nine different power supplies, you’d only need 340 units of safety stock inventory for the one common power supply. The formula for this calculation is pretty simple. To calculate the % reduction in SS inventory you take 1 – 1 / sqrt n, where n is the number of original components. That’s it. 61 Reducing Safety Stock Inventory – Rough Cut Analysis » Start off with n components % inventory reduction Design down to f components » The lower you can make f, the larger the reduction in inventory. f n % Reduction % reduction = 1 − f Assumes: Sales volumes are approximately equal for each part; Demands are independent and uncorrelated; Leadtimes are known 62 © design4X http://design4X.com Of course, going from numerous components down to 1 is often difficult However if we start off with n components and design down to f components, instead of one, again with all the previous assumptions, the basic equation is: 1 – the square root of the new number of components f, divided by the square root of the original number of components n. The lower you can make f, the larger the reduction in inventory. 62 Reducing Inventory » Pipeline inventory Pipeline inventory Punits = L ⋅ µ » Cycle stock inventory Cycle stock inventory CS = 1 µ 2 » Safety stock inventory • Variation in demand • Variation in lead time Safety stock inventory SSunits = k ⋅ (LT + R ) ⋅ σ 2 63 © design4X http://design4X.com We’ve now shown you how to determine the amount of inventory that a company might expect to have on hand—pipeline, cycle stock, and safety stock inventories. Obviously, your company will differ based on the policies you use and a host of other factors. However, what you see from these equations is how you, as an engineer can start to reduce inventory – especially in terms of safety stock. If you can reduce the leadtime of the product, then you can reduce the company’s inventory requirements. And, as we just showed you, by reducing the standard deviation of the demand through commonality, you can also reduce safety stock inventory. Ahh, but how to do that? In our section on commonality, we’ll show you strategies on how you can reduce this standard deviation of demand. 63 Agenda » Introduction • • Design for the Supply Chain Pre-Test » Step 1: Know Your SC • • • • Think About It What is a Supply Chain? Knowing your SC Apply It » Step 2: The Costs of SC • • • • • • • • • Commonality Postponement / LPD Shipping Logistics Apply It » Summary • • • Example - DVD-RW dfSC Assessment Think About It Costs and the Supply Chain Types of Inventory Calculating SS Inventory (optional) Apply It » Step 3: Designing for the Supply Chain • • Think About It The Five dfSC Strategies 111 © design4X http://design4X.com 111 Computer Example – dfSC Strategies 2) Commonality • Cost of inventory - Assume you have 4 weeks supply of product ICU (Inventory Cost per Unit) = 4 * $200 * 30% 52 = $4.62 • Going to one common manual and including SW reduces variety from 16 to 2 % inventory reduction = 1 − f n = 1 − 2 16 = 65% • Thus, inventory costs are reduced by $4.62 x 65% per unit $4.62 * 65% = $3.00 That’s a $3 reduction in inventory costs! 116 © design4X http://design4X.com Let’s first calculate the cost of the inventory per unit if you stay with the 16 varieties. For this example, we’ll assume the company typically carries 4 weeks of supply of the DVD’s in FGI in order to meet demand. The inventory cost per unit is 4 weeks of supply times the $200 cost of goods, times an estimated 30% annual holding cost, divided by 52. That equates to a $4.62 cost per unit for holding that amount of inventory for 16 varieties. Now imagine we choose to go with one common manual and include the Photo archiving software in all the configurations. This drops the number of stock keeping units from 16 to 2. And using our equation for inventory reductions this would reduce the necessary finished goods inventory by 65%. Thus, by making these changes we could reduce the inventory cost per unit by $3. 116 Computer Example – dfSC Strategies 2) Commonality • Further savings available by going to a universal power supply – Reduces to 1 variety % inventory reduction = 1 − 1 16 = 75% $4.62 * 75% = $3.46 • More savings from: - Reduced complexity - Reduced documentation - Volume discounts 117 © design4X http://design4X.com Now let’s assume that you go to a universal power supply reducing the variety all the way to one common product. That’s a 75% reduction in the safety stock inventory costs – for a potential savings of $3.46. And there will also be savings from reduced complexity and documentation, and increased volume discounts. 117 Computer Example – dfSC Strategies 3) Postponement / Late Point Differentiation • Without postponement - Keeping 16 varieties in stock means lots of inventory • Implement process postponement - Keep two varieties on hand (120 VAC, 230 VAC), add the appropriate manual and software at the US and European DC’s – Slightly more complicated distribution system – Some extra labor – Still have to have manuals in stock » But can reduce stock of manuals, since manuals can be printed locally, aren’t shipping paper overseas! Safety stock inventory SSunits = k ⋅ (LT + R )⋅ σ 2 © design4X 118 http://design4X.com Now if you can’t reduce the number of varieties of final products through commonality, there are other ways to reduce the amount of inventory you keep on hand. One way is through postponement Imagine that the company doesn’t want to create a common manual. Partly because it’s environmentally unfriendly due to the extra size of the manuals required, and also because the manual becomes too unwieldy for the customer. How can we reduce inventory if we keep the different manuals? The answer is through postponement. Rather than having 16 different varieties shipped from the factory and trying to forecast demand for each of the 16 varieties that far in advance, you can postpone the insertion of the manuals until later in the process – then you ship just two varieties from the China factory, a 120 VAC version to the America’s distribution center, and a 230 VAC version to the European distribution center. The language-specific manuals are then inserted at the distribution centers. This allows the company to pool the variation in the demand and decrease the safety stock required. Obviously, adding this new operation at your distribution centers (even if you already have such centers) makes the distribution system more complicated and adds extra labor – so you’ll need to weigh the savings against these costs. 118