Managing Automation Project Risk

Getting a "MAC" involved, early in the process, can save headaches and costs

By Bill Robertson, P.E., PMP, St. Project Manager, Emerson Process Management

1 of 2 < 1 | 2 View on one page

The upgrade project looked good. All the vendors were lined up, everyone knew the schedule, and all the milestones were posted and agreed to. But one thing nagged at the chief engineer: what about the documentation?

Sure, all the vendors swore they had their ducks in a row, but with so many involved, how could he be sure that all of them had the same documentation standards, and that they adhered to them? Had the prime contractor or end user reviewed all the skid vendor design submittals and compared them to the automation device specifications/standards? How could he be sure that all the vendors had adequate internal testing standards? And what would this mean for validation?

The upgrade was built around skids, which can speed a project to completion but can cause difficulties because of the increased number of vendors involved and the need for coordination and consistency in the supply of devices. Validation! The worry just wouldn’t go away. No validation, no product; just tanks full of useless material that couldn’t be sold. And the company president was starting to get curious. As it turned out, there were good reasons to worry.

Documentation standards and internal testing requirements varied from vendor to vendor, which led to longer validation document development and execution times than expected. It also led to widely varying validation results. Some protocols took longer to execute and resolve than others. It was also discovered that some vendors’ internal testing was inadequate in terms of content and rigor. One vendor did not provide any evidence of internal testing that was assumed to be a “standard” requirement and the customer believed should have been conducted by the vendor.

On the plus side, the OQ validation protocol was written to include the required tests. But on the minus side, there were many test errors that required vendor remediation. This caused several iterations of re-testing and protocol amendments to be executed, putting a strain on the company’s validation resources. As a result, other validation tasks slipped and system water batching had to be delayed several weeks.

What should have been done? The use of skids — standard plant modules, complete with documents and drawings, based on known pre-tested process cells — can help avoid long lead times on new plant builds. The skids are produced in parallel by separate skid vendors, while the site is prepared to receive them and is fitted out with utilities and buildings. The skids are then integrated at the site. Done well, this method is a good way to handle the project and time pressures brought about by market shifts, delayed investment decisions, regulatory requirements and patent expiration deadlines.

Skids are supposed to be plug-and-play, but for that to work all vendors must be on the same page. The main control system (DCS, or distributed control system) must contain accurate information on each skid; if skid vendors veer off track, or use outdated standards or information, the success of the whole project can be endangered, and the schedule pushed back. The chief engineer in our story should have made the decision to use a more programmatical approach with activity diagrams, and to require the vendors to take responsibility for their scope of work and how it related to the overall project, so he could track everything and prevent the project from “skidding” out of bounds.

An activity diagram is essentially a network/block diagram representation of a Gantt chart. Its value is to show a knowledgeable project engineer or project manager that planning has taken place to indicate what work has to occur and in what sequence to mitigate the risk. potential cost and schedule impacts What could the scenario above do to your project? Let’s try to put it into numbers. Assume that 25 percent of the vendor skid testing documentation does not align with or is inadequate to meet validation requirements, and that a typical vendor skid carries 200 devices. Further assume that this occurs with four skid vendors. Table 1 summarizes what can happen.

The scenario played out above is one of several that may face a company doing new installations or upgrades.

Others could include:

  • Lack of oversight over contractors
  • Late process changes resulting from safety design reviews
  • Software delivery issues
  • Commissioning responsibility disputes.

Any one of these can derail or seriously delay a project and add considerably to costs. contrAct A mAc The common thread running through all the risk scenarios above is the lack of someone to take overall responsibility for automation: a main automation contractor (MAC). The MAC approach involves picking the strategic suppliers before final design. This allows for the inclusion of ideas and application expertise to improve the project’s financials and the long-term operating cost of the plant.

It makes sense to contract a MAC as partner for front-end planning to include development of the risk management plan — to do risk identification and response, to act as an application software supplier, to manage I/O databases including skid vendor supplied devices, and to provide validation services, including review and comment to skid vendor validation protocols.

Figure 1 shows typical project organization with a MAC involved. A project without a MAC involved would include many more steps, the possibility of additional complications, and consume many more days. Benefits of the MAC approach The MAC approach involves earlier involvement by the automation contractor in the project, which allows for greater influence over the costs and benefits of a project.

1 of 2 < 1 | 2 View on one page
Show Comments
Hide Comments

Join the discussion

We welcome your thoughtful comments.
All comments will display your user name.

Want to participate in the discussion?

Register for free

Log in for complete access.

Comments

No one has commented on this page yet.

RSS feed for comments on this page | RSS feed for all comments