Why is machine learning important?
Incorporating machine learning into the manufacturing process is critical for IVL because of three specific production risks that impact its ability to operate effectively: (1) the company is unable to influence commodity prices directly, so a key part of their competitive advantage comes from being able to maintain efficient operations and be a low cost provider, (2) operations are dependent on the availability of raw materials, and (3) there are potential operational risks that come with this type of manufacturing, such as environmental damage.
In 2017, IVL budgeted USD 300 million in maintenance capital expenditure over the next two years . It can be argued that by leveraging machine learning in predictive maintenance, the amount of money set aside in the future for this type of work can be significantly reduced. By combining data obtained from machines (i.e. through the use of sensors) with an artificial intelligence component, it is possible to more accurately predict when a machine is going to go through downtime and take preemptive measures to curtail this issue in advance. A report by McKinsey suggests that this could lead to savings of up to 10% per annum in maintenance costs and up to 20% in downtime reduction – thereby increasing IVL’s competitive edge over the industry .
Another example of machine learning’s importance is the possibility of more accurate forecasting of external and internal events, providing increased supply chain visibility and planning for inventory optimization. This is particularly important in the PET industry, where problems at the supplier level can compromise IVL’s ability to meet customer demand .
What is IVL currently doing?
Many of the larger chemical companies have already invested heavily in the machine learning space. For example, Sinopec, a Chinese petrochemical conglomerate started researching ways to incorporate this megatrend in its operations back in 2011 and has already reaped benefits in some of its factories, including increasing average labor productivity by 10% . For IVL to stay ahead in its niche space of the petrochemical’s industry, the organization’s management has set out an action plan in both the short- and medium- term.
The company has just recently completed a diagnostic at its test plant to discover potential areas for improvement on its way to building “a factory of the future”. Over the next one to two years, it will look to implement these initiatives at the test-plant and measure the impact they will have the plant’s ability to operate more efficiently and reduce costs (without compromising on quality). These initiatives take on various roles from improving production planning, to reducing the failure rate at key steps of the process (such as, acid burn reduction), to improving overall equipment reliability. 
In the medium term, IVL will look to track the benefits the approved initiatives have on the test plant and look to roll out the learnings to the other plants within the organization, across geographies. With that being said, the company will look to create a digitally-enabled organization that will strive for continuous improvement and look to capture opportunities for innovation, moving forward. 
What else can IVL do to capture the potential of machine learning?
One key area in which IVL can look to capture further opportunity from machine learning and other transformative megatrends in the future is by investing time in building a culture of individuals that are constantly looking to help the organization improve and remain ahead of the curve. This can be done by making necessary changes in the design of the organization to foster collaboration between teams across geographies, identifying and enabling internal talent to drive a new digital team and identifying and filling gaps in current skill levels need to drive change in a more digitally-enabled organization.
Another important consideration for the future is trying to tie key metrics for environmental sustainability with machine learning. For example, this has been done with improvements in Siemen’s operations through deep learning to reduce gas turbine emissions by 10-15% . As the company looks to make its operations more efficient, it will be beneficial to track how these changes impact key environmental issues, i.e. reductions carbon footprint.
Lastly, they can look to apply machine learning beyond the factory, i.e. in R&D, improving collaboration with partner plants or optimizing transportation.
What are some considerations to keep in mind going forward?
For a traditional manufacturing organization, what is the most efficient way to implement digital factories across IVL’s85 existing plants across geographies?
How can we achieve buy-in from traditional employees at the factory-level? What new skills / working conditions are required in a “factory of the future”?
 Indorama Ventures Annual Report 2017. (2017). [ebook] Bangkok, p.105. Available at: http://ivl.listedcompany.com/misc/ar/ivl-ar2017-en.pdf [Accessed 11 Nov. 2018].
 Columbus, L. (2018). 10 Ways Machine Learning Is Revolutionizing Manufacturing In 2018. [online] Forbes. Available at: https://www.forbes.com/sites/louiscolumbus/2018/03/11/10-ways-machine-learning-is-revolutionizing-manufacturing-in-2018/#57842ec923ac [Accessed 11 Nov. 2018].
 Defang, L. (2018). Sinopec: ProMACE Builds a New Smart Factory – Huawei Publications. [online] Huawei Enterprise. Available at: https://e.huawei.com/us/publications/global/ict_insights/201802151130/features/201806011600 [Accessed 11 Nov. 2018].
 Telephone interview with IVL Management (2018). Machine Learning in Manufacturing.
 Siemens.com. (2018). The Future of Manufacturing. [online] Available at: https://www.siemens.com/global/en/home/company/innovation/pictures-of-the-future/fom.html# [Accessed 12 Nov. 2018].