Chemicals & Energy

Constar is one of those companies whose products are ubiquitous but whose name is little known by the consumers using them. A major supplier of PET (polyethylene terephthalate) plastic containers for food and beverages, Constar applies other companies’ labels and brands to the 8 billion containers that it produces each year.

“In terms of production, ours is really not a very complex business,” says Deep Parmar, senior director of sales administration and customer service at Philadelphia-based Constar. “We purchase resin from our suppliers, which we convert into test-tube-shaped containers that are then blown out into bottles of various sizes. We are talking about 6,000 to 6,500 SKUs.”

The complexity for Constar lies in the huge volumes that it produces and in the variability of demand. “We primarily supply the beverage industry so there is a lot of seasonality involved,” he says. “Also, whenever there is a natural disaster of any sort, additional production is required. On top of that, we have promotional activity that impacts short-term demand. These are usually not very predictable and show up in peak season and usually with only five or six weeks notice. This always throws a wrinkle in our operations.”

Getting a better handle on forecasting and improving forecast accuracy was one of Parmar’s top priorities when he joined the company two years ago. “Our forecasting was not up to standard so the company wanted to focus on that and understand the problem,” he says. It soon became evident that focusing on the forecast was really the first step in a larger sales and operations planning (S&OP) process that the company needed to implement. “My background was with pharmaceutical and bio-tech companies and I had worked with S&OP in these jobs and elsewhere,” he says. “At Constar, we sort of knew intuitively that our lack of having a formalized S&OP process was causing a lot of ad hoc activities and reactive jumping to meet certain demands.” To document that gut feeling, Parmar formed a project team that mapped out a full end-to-end supply chain process and identified gaps. “We realized that there is only so much we could do to influence the demand forecast because we do not control the fluctuations that our customers are facing,” he says. “We determined that improving our demand forecast was important, but the bigger opportunity was in improving our ability to react to a changing forecast.”

This shift in focus required some education, Parmar says. “I found that even the terminology is important. I avoid using ‘forecast accuracy’ and talk instead about ‘variability in forecasting.’ Forecasts are like budgets—as soon as you finish them they are wrong.”

As Constar moved ahead to develop an S&OP process, it brought in a consultant who highly recommended S&OP Azimuth, a workbench tool from Supply Chain Consultants, Wilmington, Del. “At a very early stage, we asked SCC to come in, not to sell us anything, but just to educate us a little more about how they have seen S&OP initiated and implemented in industries similar to ours,” Parmar says. “They did an excellent job of bringing us up to speed on what the market is doing and the state of the industry. It really helped us a lot.” That led to further discussions about SCC’s solutions and Constar ultimately decided to implement the S&OP workbench tool, which standardizes the collection and reporting of data to support the S&OP process.

“It was pretty clear that if we wanted to do this on a full-scale basis, we needed a solution that could access all the data we have and put it together in a meaningful, user-friendly format,” Parmar says. “We needed to eliminate the labor intensive work of populating and maintaining spreadsheets and reallocate those resources toward actually solving problems.”

Another criterion for Constar was the ability to do “what-ifs” in the context of S&OP meetings. “We wanted to be able to identify problems and then model solutions. What would happen if we supplied this demand out of another plant? Would it be better to pull production from this customer and give it to that one? Those are the types of questions we want to be able to answer in meetings or even on the fly, so we can get decisions that move us in the right direction.”

Looking ahead, Constar also wanted a solution that could help it meet its long-term goal of allocating demand among plants based on “some sort of an optimizer model. I would like to take my total demand and allocate it on the basis of true cost at the individual plant and true cost of supplying a particular customer,” he says. “Currently we are not looking at the overall true cost. We primarily focus on proximity to a customer in order to minimize freight costs, but this is not necessarily the most optimal solution.” Constar doesn’t have the resources to go ahead with that now, “but it is one of the clear needs that was identified in our search for a solution,” he says.

Constar fully used the Azimuth solution for the first time during the May S&OP cycle, which was still in process when this article was written. Based on trial runs, Parmar expects that the process will help the company better understand demand fluctuation and demand history in a way that is easy to see and easy to communicate to others. “One of the bigger things that we are expecting is better response to demand changes,” he says. “If we can identify problem situations ahead of time, we will be better prepared to address them without resorting to rush orders.”

RESOURCE LINKS:

Constar, www.constar.net
Supply Chain Consultants, www.supplychain.com