Applied Modelling Algorithms the intelligence behind E-Fulfilment

E-commerce, the transacting of sales over the internet, is the fastest growing retail market in Europe, and according to the Centre for Retail Research online sales in the UK, Germany, France, Sweden, The Netherlands, Italy, Poland and Spain are expected to reach a combined total of 111.2 billion pounds sterling in 2014.

Growth in online sales continues to increase as consumers become more confident in the safety of the commercial transaction and, importantly, more accustomed to this new buying process. Buying a dress online is very different to the high street with several trips to the dressing room to try on different dresses in different colours and sizes, then simply handing back the discarded items. Online the same evaluation process is carried out but the consumer instead over orders, ordering multiple dresses in multiple sizes under the website’s assurance that products can be returned and refunded at no cost to the consumer.
Let us now consider the retailer’s perspective and the logistics involved. The high street store is specifically situated near to its consumers. The distribution centre that supplies this and other stores is specifically situated to be able to serve a number of stores. Stores will place orders or receive seasonal products, orders will be consolidated, and shipped to store.  The timing of these replenishments will probably take place out of retail opening hours, when heavy goods vehicle deliveries are free to enter the high street.
Retailers need to consider multiple factors when moving to online trading, some of which include:
Which DC should satisfy the order?
Is there sufficient geographical coverage to deliver the goods in an acceptable lead time? For fresh products such as grocery this is more important than apparel, but consideration needs to be given to the cost of transport and the target demographic’s ability to wait more than 24 hours to get their purchase. Time to fulfil the order and the cost of shipment suddenly become a retailing differentiator.
What changes do I need to make to the DC?
The DC is now no longer a B to B supplier. The DC needs to accommodate some of the tasks previously performed by the retailer. For example, specific customer orders now need to be picked, so a level of quality control is required to ensure the correct items are picked. These orders then need to be packaged and labelled for the consumer. Marshalling and assembly areas may now require space for benches, pickers may now use trolleys, more pick faces may need to be allocated, and more people employed across different shifts. 
The shipment of smaller orders on a more frequent basis now requires more vehicle collections, possibly smaller vehicles. The loading bays may be designed for the loading of larger vehicles and now need to accommodate smaller vehicles or courier collections.
Goods Inward will no longer just handle cases and pallets. It needs to accommodate customer returns. Quality inspections need to be carried out, products may need to be repacked before they can be returned for resale. This will impact on the space required and facilities needed.
What transport is needed?
Potentially smaller vehicles may be needed. A new outsource contract established or postal services engaged.
Consequently, when opening an online store the retailer must decide how to fulfil these orders, whether they can be fulfilled through the existing supply chain, or should the network and stores fulfilment processes be adjusted to accommodate the orders, or indeed if a completely separate channel should be established. 
How can you evaluate the impact on the existing supply chain and validate a new strategy? The most powerful tool available to support the Supply Chain Director is “supply chain simulation”. 
Supply Chain Simulation
Fortunately the same technology that is enabling online retail computers, has also led to the advancement of simulation technology.   Early simulation applications required some dedicated expertise, programming skills and a considerable amount of data. The user needed to build up a model from scratch, which was time consuming and also difficult to validate. Not surprisingly then, simulation was confined to operational research specialists or to large corporations that could invest in the people and skills needed to create, validate and maintain the model. 
Today, with the advent of ever more powerful personal computers, a new class of simulation products has been born, referred to here as Applied Modelling Algorithms, AMA. These simulation applications have been designed by supply chain professionals to be “applied” to solve specific supply chain problems. Simon Shore managing director of Cirrus Logistics, a provider of supply chain simulation applications, said “…………”
Let us consider again the changes needed to the supply chain for e-fulfilment, and how can AMA help design and validate the correct supply chain strategy?
Network Validation
Commercial analysis will provide some expectation as to the volumes anticipated, the key geographical target areas and the desired order to delivery time. Simulation will use this information to determine the optimal location based on the lowest cost to serve the consumers. 
A cost to serve simulation application works by using the geographical co-ordinates of the DCs and the key customer conurbations to then consider different demand profiles.  By creating a model in this way the location of the DC can be changed and the net effect on the transport cost evaluated. 
Equally if the retailer is seeking to meet key lead time to delivery criteria, free next day delivery for example, and the order correlation between the sales effect of free next day delivery is known, then the model can find an optimum network strategy balancing transport costs versus sales.
The dynamic nature of the model allows for different constraints or options to be tested. In this way simulation will assist in determining the best network and transport strategy.
Warehouse Validation
Having established the optimum location strategy the next step is to consider the impact on an existing DC or to design a new DC fit for purpose. 
To commence a warehouse simulation project the first thing to do is to build a virtual warehouse model. In the same way the network model needs to understand the distances and road networks from DC to consumer, the warehouse needs to know the relative distances travelled in the warehouse. 
Warehouse operational processes, storage and pick faces need to be set up to make best use of the available space. Commercial analysis will provide the volume information, enabling the warehouse to be sized and the market information will allow for the determination of anticipated fast and slow moving items. This information can be used to set up the pick faces and configuration of the storage areas. 
The processes needed to dispatch the orders can then be set up, shift patterns established and the model run. So in short, a virtual representation of the warehouse and its operating model can be constructed.
Whilst this may all sound like a lot of work, warehouse simulation tools employing AMA functionality make this task a simple activity enabling complex warehouses to be constructed in hours not days. 
The design can then be tested by running the model to ensure the simulation you use has the capability to model the packing and labelling process. It will also test the yard or site where the DC is situated, as the limitation of the operation will probably be in the e-fulfilment processes, and not storage areas.
Running the simulation will lead to the compilation of KPIs which will enable the model to be tuned and the best DC design and operation determined.