When pundits discuss the Internet of Things (IoT), they often end up talking about the importance of Big Data analytics. Sensors generate tremendous volumes of data. How will all this data be analyzed?
But when I look at the different sensor based systems generating data in a warehouse, I’m beginning to think the bigger challenge may be execution, not Big Data analysis.
Most well run warehouses have a Warehouse Management Systems (WMS) that depends upon bar code data. Warehouse floor level operations personnel and warehouse managers execute their tasks based upon this data.
Some warehouses also have extensive amounts of material handling equipment and will use Warehouse Control Systems (WCS). Equipment can have sensors that provide diagnostics on how well the equipment is running, and if not running well, what the problem is. Maintenance personnel depend upon this data.
Finally, some warehouses also have Building Automation Systems (BAS). These systems control lighting; heating, ventilation, and air conditioning (HVAC), and security and facility access control. Facility and security managers are the primary users of this sensor/actuator data.
Suppliers for each one of these systems – WMS, WCS, and BAS – have built dashboards that do a good job of keeping their primary users informed on the key events they need to know about. But, in the age of IoT there is an increased ability to share key pieces of data across these traditional silos to solve difficult problems.
For example, imagine a warehouse that has ambient, chilled, and frozen sections. Imagine that some of the customers that get goods from this warehouse want assurances that the goods they receive have been kept within a prescribed temperature range. They don’t want the goods if the temperature has gone outside that range.
There are temperature sensors that can be embedded in cartons or pallets. These sensors can show the temperature of goods by time period for the entire duration of the trip to the customer’s site. No warehouse manager would want to be told a temperature excursion occurred at his warehouse. The facility manager, in turn, would not want to be blamed for permitting the temperature in that zone of the warehouse to get too high or dip too low.
On a hot summer day, keeping chilled or frozen products at the correct temperature is not easy. Goods have to be staged at a dock door. When the dock door is opened, heat seeps in. The warehouse manager may want to insure that a particular product be staged further back from the dock than is usual, that dock doors are kept closed until a truck is at the door, and that when the truck becomes available it is loaded with alacrity.
The facility manager may want to keep the temperature of that zone of the warehouse within a narrower temperature range than usual.
In short, for cold chain supply chains, BAS and WMS systems are going to need to coordinate their activities in a way they were not designed for.
The good news is that IoT technologies are, by definition, web enabled and rely on Internet communication protocols. Getting access to the data in a meaningful manner is not as big a problem as it used to be.
Nor is it all that difficult to build custom dashboards and portals which can display key data in the right format by user role.
But how do you make the sensor data actionable? To make cold chain data actionable, the following questions might need to be answered. How far back should an SKU for a particular customer be staged from the dock? Does that depend upon the temperature outside? How fast does the truck need to be loaded? Does that depend upon the temperature inside or outside the warehouse?
I don’t see the answers to these questions being provided by analytics. If you have to use historical shipment data on what has not worked, you have already angered your customers. The better route is to run experiments.
But once experiments provide the answer to those questions, how can these rules be built into the Warehouse Management System and executed in a highly repeatable manner? What actions does the facility manager need to take to help insure success? How is this actualized?
In short, Big Data analytics are going to be important to solve many types of IoT problems. But I think building systems capable of acting on these insights will be the more difficult challenge in many cases.