亚马逊的随机上架策略
amamzon真的是随机储存。首先要有先进的库存运行后台,每一个Asin对应一种商品,对于上架和捡货来说只要是选取不同的数量即可,一旦进行收货,这个库存就已经在amazon系统后台运行,只要有足够的权限就可以看到每个商品的移动记录,其实对于随机上架来说,会有一些上架规则的限定,是在一定功能区域,商品大类/中类等类别区域内的随机存储,由上架员根据上架规则将商品随机上架到任何货位,当然同样可以看到这个移动记录。当一旦上架成功,客户就可以下单。最后等待捡货员将需求商品捡出。
要研究电商的下架出货,必须先研究如何上架,如何上架可以保证下架的效率。所以,在百万SKU库存下,随机上架更节约上架和下架的总体时间。电商的下架是随客户的订单内容的,客户的下单商品是随机的。这个大家可以详细的查看自己仓库的出库商品,库位和出库量来分析。商品随机储位后,商品越分散,出库效率越高。
Bin系统介绍
货位与库存数量绑定,则有着明显的好处,如果将这一思想贯彻到极致,则会自然而然地出现本篇将要介绍的货位系统,不如称其为库存货位绑定的货位系统,简称为Bin系统,这也就是亚马逊公司的仓库管理系统。
Bin系统操作流程
一、收货:收货时实际是将采购订单看作一个货位,运货车看作另外一个货位,收货员将货品逐个从采购订单的货位转移到运货车的货位上去。这样的操作精度高,而且效率也相当的高。
二、上架:上架实际上也是货品从待上架的货位(运货车)中的货品转移到存储用的货位上的过程。
上架操作按批次进行,每一个运货车作为一个批次,一个批次中包含了多次的上架操作。每一次的上架操作只涉及一个SKU,在操作时,需要输入系统的信息为:上架SKU,目标货位,上架数量(批次号中已经包含了运货车货位的信息)。
在Bin系统下,由于货位和货品数量相绑定,因此在上架操作时,也不要求将一个SKU一次性放到同一个货位上,而是可以根据货架的实际剩余情况灵活安排到两个、三个甚至更多的货位上。
由此可以看到,在Bin系统下,上架员具有相当的灵活性,看到哪里有空隙,就可以将货品放到哪里。这样的库房,虽然在看起来会很凌乱,货架上放着各种各样的东西,杂乱无章,但实际上所有的信息都存储在货位系统中,任何需要都可以随时满足。
三、 盘点:在Bin系统下,每一个存储货位中,分别有几个SKU,每个SKU有多少数量,这些信息都是在货位系统存储的。并且,由于每一次库存实物操作都与在系统中相对应,所以实物与系统是同步更新的。
在这样的情况下,盘点可以在任意时间,任意货位操作。即时,在盘点的同时进行上架、拣货等操作,对于盘点精度也完全没有影响。这是其他的任何系统都无法做到的。
四: 拣货:在Bin系统中,由于货位与货品数量绑定,因此在生成拣货批次的同时,可以指定拣货库位,只有被指定的有拣货需求的货位会被路径规划系统所考虑。
例如,订单中需要10个SKU A,而当前可用库存共计有23个SKU A,这23个货分别位于Location A,B,C上,分别有8个,9个,6个,则系统使用其中的10个,例如从Location A,B上分别占用8个,2个,则Location A上的8个以及Location B上的2个库存属性会设置为“订单占用库存”。
拣货时,根据所有已占用库存货位的位置,自动规划出拣货路径。拣货时,只能检出“订单占用库存”,而不能检出普通库存。
拣货时拣出的货品,放在拣货容器中,同样也是一种特殊的货位。
五:出货:出货时,订单中包含的货品,从拣货容器中转移到包裹,包裹号一样可以追踪。
综上所述,Bin系统将货品、货位、数量的绑定关系做到了极致。这样做的好处可以有目共睹,Amazon所使用的货位系统原理上与上述一致,支持起了每年400亿美元的销售规模,并且完全可以支持到更大的规模。
但是也必须清醒地认识到,Bin系统可以实现库存的精密化管理,但是成本非常高。首先,Bin系统数据库虽然结构相对较简单,但是数据量很大,任何的库存转移的操作都必须与系统同步,造成了数据库的读写负荷极大,对数据库系统的可靠性、稳定性的要求很高;其次,所有库存转移的操作与系统同步都需要设备,这些设备必须具有移动能力,相当于每个操作人员都必须配备,这一投资也是非常巨大的。以最为常用的Symbol的RF移动扫描枪为例,一台就要将近8000元,每个操作员工一台的话,设备投放是非常大的。
Bin系统的要点
1.将整个库房,所有用于放货的物理空间都标记为container(其实就是货位),container与货品,货品库存数量绑定。
以收货过程为例,在Bin系统中,操作人员在收采购订单后,收得的实物往往放在运货车(托盘或者小车)上,这时运货车就是一个容器。运货车有自己的编号(即相当于货位编号),在此运货车上的所有货品及其数量都绑定起来。
运货车和采购订单是多对多的关系,也就是说,若采购订单比较大,其货品可以放在多个运货车上,而采购订单较小时,也可以将多个采购订单放在一个运货车上。
在使用Bin系统后,明显可以看到两个好处:
A. 以前收货时,往往是清点确认数量后,再在系统中确认收货数量;而采用Bin系统后,可以认为采购订单为一个货位,而收货动作就是将货品从采购订单的货位中转移到运货车的货位上。因此,收货操作时可以采用一边扫描一边收货的方式。这样做,将收货和点数结合起来,效率有所提高,更重要的是,逐个扫描的方式实际是系统点数,收货人员可以将精力放在检查货品是否合格,提高了收货质量。
B. 收货后,由于运货车上的货品及其数量在系统中有记录,则上架员可以直接上架。上架时直接按照运货车的数据即可,而不用去匹配采购订单数据。这样有利于上架员工作量的平衡,也提高了精确度。
2. 不同的container(货位),有不同的属性,对应于不同的操作任务。
货品在库房中,实际是处于不断流转的过程中,涉及到的操作有:收货、上架、存储、拣货、发货,其中拣货、发货都可能是由于订单、调拨、退货的需求而发起。各个操作环节时涉及到的容器(货位),设置为不同的属性,只能由相对应的操作对应使用。
这也就是说,收货使用的容器只能用于收货以及其相关的操作,例如是上架,而不能用于拣货;退货拣货时使用的容器也只能用于退货拣货,而不能用于订单拣货。
这样的规定后,某一容器只能用于与其属性相关联的操作中,而不能滥用,减少了操作中的错误。例如,收货完成后,收到的货品只能放到收货处的运货车上,而不能放到拣货使用的运货车中,这样不会发生错乱。
3. 数据结构设计,SKU与Location是多对多的关系,某一SKU可以存放于多个Location,某一个Location也可以存放多个SKU。
每一个SKU在每一个Location的数量都作了记录。另外,还需要再引入当前库存属性的概念。库存数量即对应于库存结构中所指的几种分类。
SKU Location Qty Property
SKU 1 Location 1 Qty 1 Property 1
SKU 1 Location 2 Qty 2 Property 2
SKU 2 Location 2 Qty 3 Property 3
SKU 3 Location 2 Qty 4 Property 4
... ... ... ...
4. 任何货位变更的操作,都必须与系统同步。
例如,在移货(货品从一个货位上移动到另外的货位)操作中,需要输入系统的参数有:移动货品SKU编号,移动货品数量,源货位,目标货位。例如,在某一个SKU O从Location A向Location B移动N个,在移货操作前后,相关货位的数据记录分别如下:
SKU Location Qty
Before SKU O LocationA Qty 1
SKU O LocationB Qty 2
After SKU O LocationA Qty 1-N
SKU O LocationB Qty 2+N
再例如,在拣货时,若拣起一件货品的实物,需要在系统中输入货品所在货位,货品的SKU号,货品的数量。在实际的拣货操作中,往往是系统指定了货位,操作人员是按照系统的提示走到某个货位,取下货品,扫描SKU即可完成拣货操作。
这篇文章对这个随机存储有些解释,太长了,先收着,有空仔细研究。
========
How can something be random on purpose? Well, Amazon, the world’s largest online retailer, stores its goods in a chaotic disorder. But only at first glance, because there’s order behind the apparent disarray. It’s called chaotic storage.
How does chaotic storage work?
A warehouse for chaotic storage – sometimes also known as random storage – is basically a shelving system holding the products. So far, it doesn’t differ from a warehouse with fix storage positions. What makes a chaotic storage system so special is the flow of material.
This starts at the goods-in section: the warehouse staff takes incoming goods to the shelving system, where they are placed in unoccupied shelf positions. Each shelf space has a unique barcode and every product as well. The staff uses handheld scanners to record the shelf space and the corresponding product, thus telling the computer, where the goods are located.
When an incoming order requires these goods to be picked, the computer compiles a picking list. It then sends order pickers to exactly those shelf spaces where the requested products can be found, according to the database. In order to keep this database current, each article that is removed from the shelf needs to be scanned again.
By the way, chaotic storage does not imply automatic storage. Although it is possible to operate a chaotic storage system automatically, it is not always the best alternative. Amazon for instance, still needs quite a lot of manpower, because a simulation of the storage processes showed that hiring warehouse staff was more economical than automation.
What are the advantages of chaotic storage?
Chaotic warehouses are much more flexible than conventional ones and can respond to changes in the product range much easier. This reduces the amount of planning, because neither the range of products as a whole nor the sales volume of particular goods need to be known or planned in advance.
In addition, chaotic storage allows to use the available storage space more efficiently, because freed-up space may be refilled immediately. In a storage system with fixed positions on the other hand, some shelf space is always reserved for certain articles, even if their actual stocks are considerably lower.
Chaotic storage is a time saver, not just when stocking up on goods but also during order picking. Incoming goods are simply placed in free spaces on the shelves. The computer will then create picking lists with optimised routes whenever someone orders products. This way, the distance the warehouse staff needs to cover is shortened. Furthermore, picking lists at Amazon are not sorted by order, which means that the picked products have to be combined to shipments in an additional step.
The amount of training required by new employees is also remarkably lower when using chaotic storage. It is not necessary for them to memorise the entire warehouse layout or even single storage locations. This will allow you to replace staff more easily or hire seasonal workers during peak times.
What are the requirements for chaotic storage?
Intuitively, most people would store similar goods together, virtually sorting them according to predefined characteristics. This would place all books in one section of the warehouse and all toys in another section.
But that’s not necessary in a chaotic storage system. The products only need to share the most basic requirements with regard to storage (i.e. temperature, humidity). Further characteristics don’t have to be considered. In a chaotic warehouse, all kinds of different articles may lie directly next to each other, such as books, toys, sport equipment, electronics, DVDs, jewellery and digital cameras.
Exceptions are made for fast-moving articles, because it wouldn’t be worth storing them, and those items which are too heavy or bulky for normal storage operations. Articles like these have to be stored separately. Perishable goods are also not suitable for chaotic storage.
Needless to say, all the goods have to be barcoded and entered into the database. The same holds true for all possible storage spaces. The computer also needs a kind of map of the entire warehouse, enabling it to compute optimised picking routes.
Chaotic storage is dependant on a reliable warehouse management system. If the computer would freeze or lose data, warehouse operations would need to be suspended until the problem is solved.
This type of storage is particularly interesting for distribution centres handling a large number of items with small stocks each. This usually is the case in the online retail business.
Also, orders with articles from different categories are a common occurrence there, so storing them according to categories would not yield any advantages. Quite the contrary: the staff at Amazon takes care not to place articles from the same category directly adjacent to each other. This improves order picking accuracy because mix-ups are much less likely.
The term “chaotic storage” is by the way only justified from a human point of view, but is not at all correct from the standpoint of a computer. For a warehouse management software, a chaotic storage system is nothing more than a sequence of calculations and database operations.
Do you think that Amazon is a good example for a chaotic storage system or do you know a better one?
2019-05-08 13:57