Appunti Advanced supply chain planning

Operations planning – Ground zero

Supply chain

The entire network of firms who interact to turn raw materials into finished goods and services and to deliver them to end customers (Supply Chain Council). Supply Chain Management is the integration of business processes from end user through original suppliers that provides products, services and information that add value to customers.

Why?

Supply chain planning

The objective is connecting SCM practices, industry experience and knowledge about IT tools to cope with advanced Supply Chain Planning tasks, in today's business scenario. This goal should be reached through coordination between capacity planning and operations planning.

What does advanced SC planning mean?

MT Railways manufactures 8 product families, driven by available forecasts over a planning horizon of 10 months. Production costs may experience (small) variations during each month due to expected changes in components' cost and availability. Also, selling prices may vary, and stock-out cannot be ruled out (because having zero stock out is too costly). Production routings are well defined and stable, as well as resources' capacity, which is limited. The trade channel could hold some inventories to cope with capacity limitations.

Setting up production for each family will trigger setup times (reducing available capacity) and extra costs, which have been quantified by manufacturing engineers. Main suppliers have capacity limitations as well; SteelComp, for instance, has shared some available capacity plans, considering an aggregate supply lead time of 1 month and an aggregate coefficient of utilization of 1 supply unit per each product made by MT Railways. Steelcomp is the supplier we consider, with a time-related specification and its inventory, then there is (after the railway) MT Railways, and after there is the trade channel and an inventory before the trade channel.

How much to produce? How much to purchase?

All happens to reach a set goal (i.e., minimizing costs including opportunity or maximizing profits). This looks like a mathematical programming problem. So, we must frame decisional variables; by convention, production is happening within the time period and supply/delivery at the end of the time period. Then we consider relevant data (such as variable production costs) and service variables.

The problem approached from the point of view of modelers includes service variables that help the organization of a problem, such as the inventory position, a Boolean variable about the production happening or not. Actually sold items are needed to manage the stockout possibility. Then we start from definition of the target function, we want to maximize the sum for all product families of the sold quantities times the price valid for product I in the period t. All minus the production cost variable and valid in the period, also fixed costs must be considered through the Boolean variable that measures the production (so the setup is happening) times the setup cost à SET*C-Setup. Also, inventory costs must be considered.

Then we have to measure the constraints that we have to be subject to: Production capacity à the sum of any product in the quantity that I am producing times the-time for producing that pieces must be lower the available manufacturing hours. True for all periods t. Production quantities à the sum of production requirements should not overcome the amount of inventory available from periods before. True for all periods t. Balance equation à the sold quantity of product should equal the minimum between demand and inventory available. Also, I have to connect production with available inventory, so I have put equal the steel entering in MT and exiting from the supplier à conservation of steel, inventory. Non-negativity of variables.

This is a mixed integer, non-linear programming problem (due to the min operator). There are tricks to transform this problem to a linear one, but this is basically non-linear due to capacity constraints. But this is pretty useless for working with companies as this can be just a starting point. Also, we are assuming perfect information which does not exist in real cases/projects, as forecasts which may also be wrong.

Problems:

• Data availability
• Data quality
• Data uncertainty (I may have a clear model but results may be stochastic in nature)
• Closed loop time for making decisions à time is critical to get data, model and analyze/run the algorithm, it may require weeks or months to get data and results may appear when they are not meaningful anymore. We are modelers but we are not running the models we create.
• Social issues, collaboration is needed to solve problems, you can change problems to find better solutions (i.e., manage capacity through a campaign to communicate exclusivity) à advanced supply chain planning is about skills to explore problems from new perspectives

Objective function: maximize the profit or minimize the costs? In this case, the price is specified so we go for profit maximization.

Max SUM(i,t) (Sold * Price ) – [(X * Cvi,t) – Seti,t * Csetupi – Choldi * InvP-fin – Choldsc * InvSCt)]

Subject to:

• SUM(i) X * Tprod + SET * Tsetup <= H for each t
• SUM(i) X <= InvSC (t-1) for each t
• SC(t) <= SClimit (t- LT)
• Sold = min (D ; Inv-in)
• InvSC = InvSC (t-1) – SUM X + SC(t)
• Inv-in = Inv-fin + X
• Inv-f = Inv-in – Sold
• X >= 0
• SC >= 0

We assume the cost of setup is constant over the period, otherwise, there is also the t to be considered. This is a non-linear problem because we have capacity constraints. There are some methods to transform it into a linear problem.

The problems here are:

• Data availability and data quality: the assumption of having perfect information does not hold true. There are problems in data acquisition, data cleaning, data integration.
• Data uncertainty (it is different from availability and quality): uncertainty is affecting the businesses.
• Closed loop time decision: time related to data, modeling, running of the algorithm. Running the model is not easy: we need powerful computers, mathematicians.
• Social aspect: we can collaborate to solve the problems. If the problem is too big, I can find many ways to solve it. I can even change the problem: if I have not enough capacity, I can provide a marketing campaign to transfer a more exclusive product, for example. You can reframe the problem and then find the solution.

Exercise

4000*5 years- 1500*5 years- 400*12*5- 1100*13+3% every two years à Retirement of 40 years for the other guys- Sara can pay 20 000 to retire after 35 years- Cost of capital assumed is 2% and her salary will increase every 2 years-Sara could start from a net income of 1500, it is not enough. She should get to 1650. Even combining all earnings for a lifetime discounted, you should be around 500K starting from 1100, so Sara should be better off from any monthly income over 1650 à our motivation.

[Nizar talking a caso: De facto standard: washing machine (during time the standard for washing machine has always been like that, without authorizations) De Jure Standardization: can be done only with authorization (?) Complexity is bad sometimes but it is good when the customers accept to pay for it. Can you make an airplane simple? No, it is complex but the customer wants to pay for it. Unjustified complexity is not good because the customer will not pay for it.]

Vocabulary, frameworks

• Process
• Bill of Material (eBom, mBoM, …)
• Production Cycle
• Lead Time (supply, production, delivery, …)
• Sell in / Sell through / Sell out
• Tier / Echelon
• Opportunity cost
• Supply Chain Surplus and Supply Chain Profitability
• Flows in a Supply Chain
• Bullwhip effect
• Product architecture
• Decoupling point
• Postponement and Delayed Differentiation
• Risk pooling
• Fisher’s Model
• Other reference frameworks

Supply chain surplus and profitability

The profitability is the difference between price and costs generated in all stages of the supply chain. The surplus is the difference between SC cost and customer value, which represents the willingness to pay that characterizes how much a customer values the product/service. The surplus for a customer is the difference between the price paid and the customer value (so how much more you were willing to pay OR how much less than expected you paid).

Flows in a supply chain:

• Material/service flow (from supplier to OEM)
• Information flow (from supplier to OEM)
• Money flow (from OEM to supplier)

For example, the target of S&OP is to align demand with operations and align demand and capacity. This is for aligning sales and operations department with other departments. Also, making sure to have the turnover that you want is crucial (especially if you are a listed company). Note that time buckets are different for every industry.

Material Requirements Planning

Both approaches use data about expectations for the future. MRPs are more complex, but companies are using it. For example, when filling the car, you are looking for a stock-based approach, but if you know that there will be no opportunities to refill in the near future, you consider future availability. You manage pasta in a home through stock, while Dom Perignon is pull because it is expensive. Stock needs stable supply, reliable suppliers, low obsolescence (so you can keep it long), and stable demand. You have to create conditions to use pull and lean manufacturing; if they are not available, you should enable them, not look away. MRP is hard to use, but you are forced to use it if there aren’t the conditions to use stock-based planning. MRP limitations assume infinite capacity and so uses a fixed lead time (this is not real as it depends on the workload and available capacity). It was designed to tackle variable demand as you are forecasting the future and order to have material the time it is needed. MRP requires lots of data (demand, lead time, scrap %, …) and as one parameter changes, everything must be computed again as everything is made to reach a delivery date (the planned one).

Bullwhip effect: amplification of variability from the downstream supply chain to upstream supply chain. Very problematic for the planning of the supply chain. Its causes are:

• Promotions: in order to prevent it, we can use collaboration between the actors of the supply chains. Also, fixed prices are a solution, of course (like in Walmart)
• Long lead times
• Demand forecast: the standard deviation for the calculation of the forecast may be different for different actors in the supply chain (Retailer, distributor, manufacturer…). It depends also on the period of the forecast: bigger period of time, the demand will be more stable, so the variability will be less in the forecasting. A solution is that the retailer shares its forecast with the whole supply chain.
• Volume and transportation discounts: full truckload does not represent the real demand of customers. The customer demand is highly fluctuating, but putting all the demand in one full truckload reduces the variability.
• Inflated orders: due to stop in delivering like in the Christmas period, there are amplifications of demand before these periods.

The vendor management inventory can solve the problem of bullwhip effect: the manufacturer can see at the retailer the fluctuations. The inventory is managed by the supplier.

Use of stock

Stock management is simpler and it is cheaper because you look only at the inventory. So why do companies use requirements-based planning? Because there are four conditions: Managers have to create the conditions in order to go pull. The conditions can be built. MRP is evil, and stock management is heaven, but we are forced to stay in MRP if we do not create the conditions. MRP is assuming infinite capacity and the lead time is fixed (?) (this is completely false in real life: lead time depends on the work load and on capacity). Then there is the MRP nervousness problem: MRP needs a lot of data. If you change one of the parameters, the program is completely changed. Strong dependency on parameters.

What is the use of stocks?

• Cycle stock: for optimizing the process (batching)
• Safety stock: to face uncertainty
• Speculative stock: to have best price
• In transit stock: in order to fill the pipe

But the most important objective is to guarantee a certain service level. And there is also the psychological level (vetrine piene di roba). Bad side of inventories: they hide problems! Sometimes stocks are there to hide problems. You cover problems putting them in stock. You have to chase this one.

Distribution planning

Decoupling point

Decoupling point: separates the part of the system managed "to order" from that the part managed "on forecasts(*)".

Decoupling point: point in which you switch from push logic (producing according to your plans) to pull logic (producing according to customer orders). It can be placed in different points of the supply chain, defining different supply chain fulfillment strategies. (Remember the chicken run case: the company wants to put the decoupling point right after the packaging and right before the labeling and not before because you cannot store chicken all together because of the risk of cross-contamination.)

Product architecture

Product architecture is related to functional and physical components of a product. It is the scheme by which the function of the product is allocated to physical components. For function 1, you can have component 1 and 2; for function 2, you can have component 2, 3, 4, etc. Product architecture is about the decisions about the system of functions and parts. In order for a car to move, it is necessary to have many physical components to ensure that a car goes from A to B; many functions need to be performed, these are performed through components. The mapping that pairs components and functions is the architecture of a whole product.

Common types of architecture are either modular (1 element = 1 function) or integral (1 piece = 1 or more functions). Modular design can generate more variants as each module can be swapped easily as long as the interfaces are standardized. It is possible to measure modularity level, but we will see it later. Modularity can influence the supply as it can enable postponement/delayed strategies, more importantly, architecture of the supply chain changes to look like a pyramid as in each step parts that are linked to 1 function are grouped together in components that move downstream.

• Modular: 1 to 1 mapping
• Integral: a single function can be linked to many components.

We can define different degrees between modular and integral. We have not only 2 choices (modular and architecture) but there are also possibilities in between. What is better? It depends. With modular, you can generate variety and it has a big impact on supply chain for standardization and postponement. Integral can have better performances.

Postponement and delayed differentiation

Delaying the differentiation. An example is Benetton which delays the coloring process (here comes the name united colors of Benetton?)

Risk pooling

It is about using standardized products and aggregate inventories. It is advantageous because the safety stock in one big warehouse will be lower than the sum of the safety stock needed in more warehouses. This is due to the fact that the fluctuations of demand are adjusting each other (keeping the same service level). Safety stocks are useful to fulfill the demand during the lead time.

Fisher’s model

The characteristics of a product are related to the characteristics of the supply chain. Products and their characteristics should be linked to the characteristics of the corresponding supply chain. Functional products (low variety, stable demand) à efficient/lean supply chain, cost1. minimization focus. Innovative products (high variety, unstable demand) à responsive supply chain, focus on reacting fast to variation in demand.

Efficient SC

• Leagile model composed of both lean and agile, depending on the position of the decoupling point.

IT for SC planning

Definition

A set of:

• Data processing functions
• Relying on company databases: we need a set of data to be processed by the functions
• Support decision making in supply chain planning tasks

IT for SC Planning is a set of Data processing functions which means algorithms or very simple functions, Relying on company databases, to Support decision making in Supply Chain Planning tasks. Decisions such as were to keep inventory, when to replenish and so on. Numbers are impossible to manage without the support of IT systems.

The past

History: Historically the evolution followed the presented steps:

1. Make = Ad-hoc IT system (1960/70s), the first MRP procedures were defined and it was the biggest available IT support in planning
2. Best of Breed = Systems made up of applications chosen among the best ones available, many options appeared and it was possible to choose the most suitable solutions amongst many
3. ERP = Enterprise Resource Planning (1990s), this phenomenon is very relevant and systems like this are still alive and many are used in the business
4. ERP + APS = Integration of ERP systems and advanced planning and scheduling systems
5. Next Big Thing

In 1964, MRP (material resources planning) was introduced for the first time, while in 80s, DRP (distribution requirement planning) were deployed. This is a very relevant part of the story but it is very famous too. What can be done with these tools? What a best in class in the 80s would be able to do?