Dow Chemical has undergone many changes in its long history, the most recent being the transformational merger with DuPont that created the holding company “DowDuPont” with three divisions in agriculture, materials science and specialty products.
For Sharla DeFrain, the batch expertise area leader within Dow’s materials science division, the goal—even before the merger—has been to standardize on a global DCS to streamline operations amid business change. And the platform of choice for Dow Silicones going forward is Siemens SIMATIC PCS 7.
The tight couplings between the architecture of a DCS and its applications, however, created challenges for Dow’s offline testing. This resulted in unplanned disruptions to operations when changes were made to the production systems. In 2017, there were several such incidents. The Dow team wanted to cut these incidents in half year over year, with a goal of having fewer than two incidents per year by 2020.
“For every incident, we completed a root cause investigation and we kept seeing the recommendation of offline validation testing before going to production,” DeFrain says. “So we went to Siemens and they determined a virtual digital twin would be the most effective way to meet our needs going forward.”
The digital twin, which Dow calls the Virtual Development/Simulation/Training (Virtual DST) system, can replicate plant installations. It includes one offline engineering system per production system, multi-user engineering (up to three engineers making changes concurrently), Siemens SIMIT simulation software and up to six virtual controllers, as well as four screen monitors for operator training. Product support includes batch, route control, safety instrumented systems, process historian and SAP Plant Connectivity (PCo).
“Every production plant we have, we want to have a replica in the virtual setup,” DeFrain says. “The benefit for us is, as we get integrated into a bigger company, we can use the pilot system to define system setup and provide plants with a standard solution.”
The virtual system is designed to decouple the hardware and software. It has the PCS 7 system made up of virtual machines on one physical server. A hypervisor manages the PC hardware resources and distributes them dynamically across the virtual machines, which are accessed using thin clients.
Dow piloted two versions of the Virtual DST. The first uses Siemens’ SIMATIC Virtualization as a Service (SIVaaS) powered by an HP ProLiant server with VMware ESXi/vSphere hypervisor for the virtualization layer. This is a turnkey system in which Siemens sets up the digital twin for the production environment and provides aftermarket support. The second system was configured by Dow and runs on Dell PowerEdge server VMware ESXi/vSphere. This do-it-yourself model addressed some of the hardware restrictions in the SIVaaS option, providing Dow with more flexibility in its approach. In both models, Siemens SIMIT V9.1 is the simulation software, including virtual controllers, the FLOWNET piping system library and the CHEM-BASIC equipment module library as the simulation framework.
According to DeFrain, SIMIT makes it easy to develop simulations and has extra features, such as the ability to speed up and slow down time, take snapshots of the automation system and facilitate testing code and training scenarios.
Testing code with SIMIT results in validated control applications that meet Dow’s business and operational needs. DeFrain explained how Dow updates the Virtual DST system with the current production code, makes the desired application changes in the virtual offline environment and then tests the changes using SIMIT. They can also train operators on the changes to the control code from any location, including the control room. After testing and training is complete, the changes are transferred from the development system into production with reduced risk of upset to the process.
“A side benefit is the use of the virtual solution for application development, assembly and testing,” DeFrain says. “It supports multiple project execution strategies and multiple systems with multiple users both local and remote.” In addition, Dow can effectively train operators on a project prior to installation and startup, she says.
As with any deployment, there where challenges, DeFrain acknowledges—specifically around the segregation of duplicate IP addresses in the production vs. the DST system, and the move from production to virtual system was not completely seamless as it relates to workflow. In addition, in the Dow-configured model, understanding the software licenses was a challenge, so Dow went with a datacenter license for unlimited guests, which was cheaper than buying a separate license for each virtual guest.
But the benefits outweigh the challenges. In the area of operations, there are advancements around change management and continuous process improvement. The virtual system provides the ability to code, validate and train on process changes prior to installation and production. The offline project development is used for application development and finding flexible ways to support multiple project execution strategies.
Dow is already seeing the advantages of having a Virtual DST system. “A couple weekends ago, we had our first significant incident of the year. Because of the digital twin, the operations team was able to run tests, reproduce the incident and determine a fix. We were back up and running by Monday morning,” DeFrain says. “In the past, it may have taken us that long just to set up a system for testing.”
While Dow tested the Siemens system as well as its own configuration, DeFrain says there are pros and cons to each scenario. The homegrown model has a lower cost, can be customized, and has the flexibility to turn on and off guests as needed. But it requires extra skills to set up the host and create guests. The Siemens SIVaaS is a turnkey system—which limits customization, but includes technical support.
“Both solutions met our needs, so it all depends on what your business needs and resources allow,” says DeFrain.