InTech

JAN-FEB 2018

Issue link: http://intechdigitalxp.isa.org/i/935787

Contents of this Issue

Navigation

Page 42 of 53

INTECH JANUARY/FEBRUARY 2018 43 Inferring information from data The interesting part is seeing what information can be inferred from all the data. What can you learn if you are willing to spend some time look - ing at the data? Inference provides information on behavior, which can relate to a person or a process, and generates four main benefits for Badger Hill: l self-documents human activities by captur- ing indications of process steps l creates information useful for training by illus- trating current versus ideal practices l provides secondary and tertiary information on top of primary functions, useful for risk management l shows where efficiency can be improved through long-term analysis What does this all mean in actual practice? How did we recognize the potential, and how have we realized these benefits? More than just level The first use of the pressure transmitter was as a DP level instrument on the cold-liquor tank, which is the initial stage for the fresh water to be used for a new batch. In the initial data (figure 2), there was normal data scatter, but in some areas, it was much more pronounced. While this might have been written off as an instrument malfunction, we real- ized that these areas coincided with feeding steam into the hot-liquor tank heat exchanger. The cold- and hot-liquor tanks are next to each other and have interconnecting pipes. Heat- ing water in the hot-liquor tank involves feeding steam through a heat exchanger immersed in the water. If too much steam is being fed into the heat exchanger, steam bubbles form in the water, which shake the tank and rattle the piping. This shows up on the pressure transmitter mounted on the cold-liquor tank. So, from this scatter we were able to infer that the steam regulation to the hot- liquor tank heat exchanger was set incorrectly. This was an interesting realization, but it be- ning. Each improvement extends our vision, ex- posing us to new technologies and applications. When we stir in DIY and Internet of Things (IoT) applications with these technologies, interest- ing things start to happen. Some may find it daunting to take risks and experiment with the new IoT and wireless au- tomation technologies, but it is possible to start small and succeed. The sensors and transmitters gathering operational data are the starting point. These technologies are scalable, making it easy to start small and grow. Rolling our own data historian Badger Hill does not have a traditional superviso- ry automation system or a process data historian. Like many craft brewers, ours is largely a manual operation with basic programmable logic con- trollers driving motors, valves, and pumps—and only a modest amount of instrumentation. When we installed the first wireless pressure transmit- ter, our initial step was to figure out the best way to extract data and post it to the cloud for analysis and archiving. This meant getting to know Modbus, an amaz- ingly forward-thinking protocol given its age, which was not familiar to us. Two wires provide remote data access and automation for dozens of devices. It can also be extended transparently over TCP/IP. Our first tests did just that using an industrial wireless gateway that bundled all of the transmitters into a single virtual Modbus network. As our first experiment, we installed a pressure transmitter on our cold-liquor tank (a brewing water storage tank) to measure the differential pressure (DP) level and post it to the cloud. Given the low cost of cloud storage, we started gather- ing data continuously. The data is requested by a simple Modbus master hosted on a $20 Arduino-like chip called a Particle Photon. It reads the response and posts it to a cloud-based database using a REST- ful interface over HTTP. For data analytics, we have pretty graphs on the Internet, and we can download the data for analysis. In the future, we would like to tap into the big data capabilities of companies like Google or Amazon. New compa- nies, such as Initial State and Meshify, also exist with this type of application in mind. We also have Modbus capabilities in our tem- perature controllers, brewhouse, keg filler, can- ning line, and centrifuge. We are slowly bringing more data sources into our analysis. Security is and should be a concern, but the cloud is no worse, and probably better, than what can gener- ally be achieved in-house by companies like ours. AUTOMATION BASICS Figure 2. The scattering in the continuous level plot of the cold-liquor tank showed a steam flow problem in the hot- liquor tank. This was one of the first recognitions of the information avail- able through inference from data collected by a Rosemount 3051 wire- less pressure transmitter. 0 1 2 3 4 5 6 CLT volume (inches H 2 O) 66 68 70 72 74

Articles in this issue

Archives of this issue

view archives of InTech - JAN-FEB 2018