Order picking is no mundane subject. Okay, it may fall short of winning the enthusiastic attentions of an acquaintance at a cocktail party, but in business terms, order picking strategies can quite easily make or break a company.
Nowhere is this more readily illustrated than in the burgeoning online grocery market. The catastrophic failure of US online grocer, Webvan, resulted from the financial burden of huge investment into automated picking facilities founded on unrealistic optimism in the expansion of this new market. The UK’s established grocery retailer, Tesco, has capitalised on its brand and stores coverage to offer home deliveries by adopting a less capital intensive approach, picking from the shelves of local stores and delivering locally. The picking strategy has enabled Tesco to move to profit in a very short period of time. The company is now turning over €15m a week and processing 3300 orders per hour, having established a customer base of two million people.
However, how long can such a model support this growing market? Surely, store based picking staff are at some point going to hinder the busy shopper? Logistics Europe had the opportunity to put these questions to John Higgins, head of e-commerce development at Tesco.com at a recent retailing conference in London.
‘In a few years we’ll build a warehouse in a busy area, but right now store pickers don’t get in the way and can be easily released for other duties in the stores when not required to pick,’ answers Higgins.
Interestingly, competitor in the UK home delivery groceries market, Ocado, claims that it is only twelve months away from operational break-even using dedicated picking facilities.
But you don’t have to be a dotcom to see the critical role picking strategy can play in achieving operational success and enhanced customer satisfaction. Advanced highly automated picking processes have been adopted by grocery retailer, Sainsbury’s, to bring down logistics costs and improve shelf availability at their stores. These ‘fulfilment factories’, four in total, represent the major part of a €1.5bn investment over three years into Sainsbury’s supply chain. It will be interesting to see if the others follow this bold initiative.
Order picking is the function that warehouse professionals consider as being the most critical in their distribution operations. It requires the most resources, is the most customer sensitive, and can be the most complex of the warehousing functions. But getting it right requires clear and objective thinking, coupled with careful analysis and planning.
For those about to embark upon the testing and absorbing process of implementing or re-configuring their order picking operations there are many fundamental questions to consider which hold profound implications for the efficiency and cost effectiveness of the operation. Questions such as: Should you bring the part-to-man or man-to-part? What about automation – when is it appropriate? Should paper be eliminated? What about single order picking versus zone picking? Or, should we be considering wave picking?
There are many pitfalls. One of the most common of which is to make a strategic order picking decision based on what was seen at some other company that may not be remotely relevant to the operation at hand.
According to Mike Cogger, managing director of automated systems company, Knapp UK, ‘The first step in the successful implementation of a new order picking system is to review the entire supply chain, both up and down the line of operation under consideration.’ He believes a long-term view of the operation should be taken, looking at, for instance, waste packaging which may in the future be subject to stricter legislation.
The more data the better
Rob Ward, systems director at Savoye Logistics, says, ‘key information required revolves around two areas: the products themselves and the popularity of the products. For every item you need to look at the order picking characteristics and the typical order – the larger the amount of historical data the better.’ But he warns, ‘You would be surprised at the poor quality of dimensional information on companys’ systems, such as length, height width of sku’s and the size of outers they arrive in.’
Modelling is an essential part of the planning process ‘but it doesn’t need to be that complicated,’ suggests Ward. ‘It really helps in identifying the splits for most popular through to least popular items.’
The 80-20 rule, or Pareto’s law, applies with uncanny regularity here. As Ward points out, ‘80 per cent of the work tends to be satisfied by 20 per cent of the population of stock items, in general – there are always a relatively few items that are key and then a really long tail.’
Jeremy Clouston-Jones, also of Savoy Logistics, explains the next step,’ We would put them in order of accession rate and then add quantity and volume.’
Pick face profiling is a critical planning activity. According to Simon Tomlinson, director of consultants, The Logistics Business: ‘Most companies are really bad at this, which leads to losing picking efficiency as fast movers are not in the right place and some products may not be available, or may not be in the most effective sequence.
Store friendly picking
Store friendly picking is an important hot topic at the moment, but for many, has not been practical because DC’s can’t keep the pick face in order long enough. It is seen as an extra task but should be planned and systematic – and carried out regularly,’ says Tomlinson. ‘Most warehouse management systems don’t help very much and extra software may be needed.’ He believes, ‘There’s a long way to go here – there are a lot of costs with shelf replenishment in stores and this can add up. Store friendly picking very much ties in with pick face profiling.’
Maximise picking time
The golden rule in planning order picking operations is to maximise the time the operative has picking rather than walking or moving between picks. Order picking can be a labour intensive operation and with the high costs of labour and difficult staffing issues in Western Europe it’s clearly an advantage to keep man power to a minimum and to make the most productive use of staff.
Goods-to-man systems obviously take wasteful leg time out of the equation, but require far greater investment in equipment. High throughput is needed to justify the investment and often this can only be achieved by consolidating inventory at one central location. An example of this would be Argos in the UK, where automated picking did not stack up at DC level and only reached critical mass with a national distribution centre.
Man-to-goods operations are less expensive and perhaps, more flexible. The major lift truck manufactures have an extensive range of order picking trucks, from low level models for speeding movement in the aisles, to man-up trucks for using the full height of the warehouse. In low level order picking the best working height for a person to have to lift, carry and place goods is in the range of 75 to 125cm from the floor. Of course, picking operations cannot be limited to this narrow range in practice. The usual compromise is to use heights of up to two metres, meaning that the picker has to bend, stretch and sometimes climb on ladders. According to information from lift truck maker, BT Industries, a picker in the food industry picks between five and ten tonnes a day – tough work without the right equipment.
Various picking methods are used: discrete, zone, batch, wave or combinations of these.
Discrete order picking relies on one person picking just one order, one product at a time over only one order scheduling window. It has the advantages of simplicity and low picking errors, and is the most common. It is, however, the least productive because the picker must complete the total order and so travel time could be excessive compared with other methods.
Zone picking relies on one or more pickers assigned to a zone for picking all the lines for each order located in that zone, and so cuts wasteful walking times. Lines from each zone are then brought to an order consolidation area to complete the order before shipment.
Batch picking, also has a higher productivity rate. It relies on one picker selecting a group of orders at the same time, one line at a time. Where a product appears on more than one order, the total amount required for all orders combined is picked at one time, and then segregated by order.
Wave picking is similar to discrete picking but the difference is that a selected group of orders are scheduled to be picked during a specific planning period and there is more than one order scheduling period during each shift.
But then, let’s not forget replenishment. ‘By concentrating too much on picking, replenishment can be overlooked. Often the determining factor for where we place stock items can be the replenishment rates, not purely the accession rate,’ says Clouston-Jones. ‘If you have the luxury of plenty of stock that means you can look at longer replenishment gaps – real-time replenishment is a real problem, avoiding people bumping into each other.’ Flow racking can be a useful solution to this but there is a space penalty to pay.
Many mechanical devices for bringing goods-to-man are available. Typically, paternosters and carousels can deliver 250 lines/hr compared with 150 lines/hr for miniload crane systems and 50-80 lines/hr for unassisted pedestrian pickers. Some manufactures of pick-to-light systems claim to be able to deliver a remarkable 1,000 picks/hr.
Picking robots and layer picking systems have also been the subject of much interest over recent years, an area where automated systems company, Univeyor of Denmark, have been heavily involved. According to Christian Dohn, the company’s sales and marketing director, ‘Robots are good in a well defined picking application, but I see the robot being surpassed by layer picking devices and perhaps, robots being used for packing more. Robots are too slow to pick.’ Dohn adds, ‘There has been a lot of experiments with robot pickers in groceries but there is such a wide variation in pick sizes etc that it’s a problem.’
Batch picking car
‘We have been involved in the dry goods side of food distribution and have been working on a batch picking car system where goods are elevated to a conveyor system which takes away picked items. We have been able to increase pick rates from 4-500 picks per hour up to 8-900 picker per hour,’ claims Dohn.
Univeyor is to go live with a layer picking application at a major grocers this October. Layer pickers will feed products to 500 lanes over seven levels, sorting out product in the right number to pack stations. An impressive pick rate of 3000 items per hour is expected of the system – the operation will be fully automated for infeed and outfeed and the only manual operation will be the packing.
Swisslog are very much involved with robot picking technology. Roland Schwab, business development manager of robotic solutions, describes the technology: ‘We have special robots for every order picking structure, designed for picking single packages, layers, or multi-use packaging. Our RoboPick solutions pick boxes, and trays of varying sizes, weights and configurations, removing the packages directly from the pallets and routing them to a consignment conveyors which transport the packages to the consolidation zone.’ Bakeries have benefited significantly from this technology.
Voice picking systems are another area of strong interest at the moment. This technology has come on a long way and seems to have overcome some of its earlier teething problems related to recognising regional accents. Two key companies active in the market are Vocollect and Voxware – Logistics Europe hopes to bring you a case study soon on this ‘handsfree’ method of picking.