3.4.2 Data and a network model
The network model is dependent on data about each element of the supply chain network which contributes towards the system wide costs or performance of the supply chain network. This requires the building up of a very large database. If a firm is planning to upgrade its existing network, it should be able to find most of the data from the central database in the ERP system or in the departmental databases. If the logistics network is being modelled on a 'green field' basis, the data will have to be generated by the firm based on its market research and strategic planning process.
Section 2.2 in your text deals quite extensively with the type of data and you will need to study this section very thoroughly. The basic requirement is the collection of data relating to all relevant cost items and encoding and aggregating this data in a way which makes computation feasible.
Data source. Ballou suggests the following as sources of data (Ballou1999):
- business operating documents
Documents such as sales orders, activity reports regarding purchasing, selling, manufacturing, shipping, storing, handling will provide useful and relevant data. - accounting reports
Accounting data focus on identifying the costs of operations, including the costs for logistics activities Although traditional accounting reports will not capture all logistics costs as required by the network designer, accounting reports remain the primary source for cost data. - logistics research
Research can be the source of valuable data for defining relationships in network planning such as the sales-service relationship and the transport rate-distance relationship. - published information
Industry reports, academic research in logistics and trade journals can provide useful primary and secondary data on cost and industry trends, technological advances, activity levels, and forecasts. - judgement
All personnel within and outside the firm connected with the supply chain are valuable sources of data and should be consulted in the network design.
Data encoding. Computation is facilitated by encoding the data. The data type which require encoding would usually be the product and sales data. The bar coding of products greatly facilitates automatic data entry for such information as sales, shipment and inventory location, and this greatly facilitates the processing of product related data. The sales data is usually collected on a customer basis with reference to the name and address of the customer. The logistics network planner requires that the data have reference to a geographic base. This facilitates the analysis of the transportation decision, facility location decisions and inventory decisions. One convenient way of encoding a customer and sale s data would be on the basis of all sales within a postal code or within a suburb which has been given a particular code by the business or planner themselves.
Data aggregation. The text covers the concept of data aggregation quite adequately. As a large firm would normally have a very large number of customers and many products, aggregation is required to bring down the amount of data to be handled by the computer to an acceptable level. The main concern is whether the loss of detail affects the quality of solution. Consult your text for data aggregation techniques.
Aggregation of customer data. This relates to demand or sale forecasts of individual customers and is the basis on which the supply chain is designed. The data is aggregated on the basis of a geographic zone so that all individual customers within the zone are replaced by a customer at the centre of the zone. There are many ways of defining this zone, but a convenient way is to define a zone according to the postal code. As the service requirements of all customers within a zone are not same, the aggregation of customers is made in specified classes. This means that each zone may have two or three different and independent aggregated demand and sales data.
Aggregation of product data. The data regarding the products have to be aggregated. There is no fixed rule for aggregation but the idea is to aggregate products into groups which make the information useful in network analysis. The aggregation can be done on the basis of such characteristics as distribution pattern, product type or transport class of merchandise.
The aggregation of data not only provides manageability for the designer, it also reduces the uncertainties associated with these data. Moreover, the error associated with transport costs due to aggregation of customer data into 150 to 200 points is said to be no more than 1% (Simchi-Levi et al. 2000).
Transportation costs. Transportation is one of the key functions of a firm's logistics activities and the costs associated with a particular logistics network have to be accurately captured in any analysis of the network. Transport costs are associated with all inbound and outbound transport of goods through the network. The network model will require information regarding transport rates for various distances between different echelons of the network and between the distribution centre and demand centres. The rates will reflect the mode, shipment size, class of merchandise, product weight to volume ratio, handling characteristics and hauling distances.
Transport rates. Transport rates determine what a particular shipment will cost per unit weight for transporting a certain distance. The calculation of costs when the transport function is managed internally by the firm's own fleet is different from costs incurred when the transport is provided by a commercial operator. Transport rates are usually linear with distance, but not with volume. This means that for the same distance and same commodity, unit freight rates will depend on shipment size. This is particularly true for LTL freights. LTL rates are usually based on commodity classification, and are offered on specific weight break. (A weight break is a range within which a particular shipment falls; for example, a weight brake of 500 kg - 750 kg, for a particular commodity, for a particular distance. A shipment which falls within this weight break will usually cost more per unit weight compared to a larger weight break of say, 1000 kg -1250 kg.) Therefore, it is imperative to apply to a particular demand centre a transport rate which reflects all these different attributes.
For incorporating rates in a database, Ballou (1999) has cited the use of shipment profile s based on past data, and the use of this profile to accurately estimate the transport rates to a destination. The following reading from Ballou (1999) deals with estimation of transport rates quite extensively.
Reading 3.3
Ballou, RH (1999) extract from Chapter 14 'The network planning process', in Business Logistics Management , Prentice Hall, pp.545-564.
Mileage estimation. The estimation of mileage for freight cost calculation is vital for supply chain modelling. As different locations of facilities will affect the distances from the retailers, the distance will have to be considered along with the freight rate. In most cases distances can be obtained from road maps or distance tables. The use of latitude and longitude with a circuity factor is a good way of estimating distance in areas with a good network of roads and railroads. Ballou (1999) provides an approximate circuitry factor of 1.21 for road and 1.24 for railroads in well developed networks. These figures are perhaps workable for the USA and EU, but definitely not for Australia which has a different pattern of spatial layout of urban centres.
There are a number of ways in which distances between two points can be found. Gilmour (1993) wrote in 1993 about Transit , a computerised vehicle scheduling system in use in Australia by trucking companies. One of the feature of Transit was a road network incorporated in the computer model based on the Australian Grid Reference System.
Another example cited by Gilmour is Roadnet, another computerised system used by Esso Australia , which incorporates a digital map of all Australian roads and can accurately calculate the distance between Esso's terminals and customers. With GPS and GIS technology widely in use today, estimation of distance is not a challenge anymore. (Refer to chapter 2 and Chapter 12 in your text.)
Warehousing costs. The warehousing or facilities costs have to be available for network analysis. The costs related with a facility can be represented in terms of fixed costs, storage costs and handling costs.
Fixed costs. These are costs which do not change with the level of the activity in the facility. Examples are real estate taxes, rent, supervision and depreciations. The fixed cost is also related to the size of the facility.
Storage costs. Storage costs are those that vary with the amount of stock stored in the facility. Typical storage costs are inventory holding costs or costs of the capital tied up in inventory and the insurance costs associated with the inventory.
Handling costs. Handling costs vary with the facility throughput and include the costs of labour and equipment to handle the inventory while receiving, handling, sorting and dispatching.
Facility capacity. The capacity of each facility determines the flow through the facility. The capacities also have to be optimum in order that there be no unwarranted excessive surplus capacity, as this will raise fixed costs associated with the facility. The important concept is the inventory turnover ratio , which is defined as the annual sales divided by the average inventory level of the facility. Assuming regular shipping orders and replenishment of inventory, the maximum storage space required is twice the average inventory level. The physical space required for this maximum inventory level has to be calculated keeping in mind storage, handling and peaking requirements. Reading 3.3 provides a more elaborate practical approach towards capacity determination.
Potential warehouse locations. In real life there are many other factors which influence network design decisions. Chopra and Meindl (2001) points at following factors as having influence on logistics network decisions.
Macroeconomic factors. Taxes, tariffs, exchange rates etc.
Political factors. Political stability is a prerequisite in facility decisions in global supply chain.
Infra-structure factors. A good infra-structure is a pre-requisite.
Competitive factors. A strategic decision based on competitors strategy and firm's own policy.
Logistical and operational factors. The performance of the logistics network with respect to inventory, transportation costs and service level of the firm.
The above factors are relevant when considering facilities within an international dimension.
Activity 3.2
Refer to reading 3.1 and 3.2 and find out the factors considered by Timberland and Toyota regarding location of their facilities.
Service level requirements. The service level is a very important consideration in a supply chain and a design must meet the service level criteria set by the management. The service level in the context of a supply chain network can be defined in various ways. In network design models, the specification of service level is usually considered in terms of distance of a facility from a customer. This distance represents time to service a customer and the costs associated with this service. It is often the case that all customers cannot be served from an warehouse with the same level of service. In such a case, the target service level is set to serve a certain percentage of customers with a specified level of service.
Future demand. The decisions regarding the configuration of the network has significant effect on the overall performance of the firm in the medium term. The facilities, once built or established at specific locations, cannot be upgraded, relocated or closed on short notice and without commitment of heavy financial input. This, of course, is not desirable. The projected and anticipated future growth of the company in terms of sales and customer demand have to be taken into account. The network to be designed must not only be growth friendly. The optimum network to be selected must provide the best results on the basis of these future scenarios.