Case Study: Saving $20,000,000 annually with a Simple API

Things that are clickbait about the title:

1. $20,000,000 Projected Savings. Regardless, it's going to be big.

2. It's an API plus a Web App

3. The API is simple, but the customer needs were unique.

This post walks thru meeting the needs of a chemical company utilizing the Phast Web Services API in a web application. Please connect with Risk Space (mike@riskspace.io) if interested in learning more.

Process Hazard Analysis (PHA), mandated by OSHA and similar legislation in other jurisdictions, is the process of identifying hazards and quantifying how severe their consequence as well as how likely the hazardous event is to occur. All scenarios are mapped to a risk matrix. The greater the severity and likelihood combination, the greater attention is needed.

Wlofab, CC BY-SA 4.0 <https://creativecommons.org/licenses/by-sa/4.0>, via Wikimedia Commons

Capital expenditure and required turnaround time for a "Red" item is intensive. In some cases, the final cost may be over $1,000,000 to mitigate a single PHA finding . When a corporation doesn't have the appropriate tools for consequence assessment, they have to rely on conservative estimates and screening tools such as RAST or other spreadsheet-based correlations. These conservative estimates will result in large, unnecessary projects that do not focus capital where it is truly needed. Making recommendations for high-capital intensive systems based on screening tools is not a sound business strategy.

Typically, a site will have a small group of staff proficient in consequence analysis using a comprehensive tool such as DNV Phast. Phast is an incredible tool. However, given its depth and breadth of capability, it has been shown that consistency is challenging. Phast allows for complete control over model parameters. However, for a PHA, this full set of controls is not required. To perform the type of consequence modeling required for PHAs, only a small number of figures will change between scenarios: pressure, temperature, hole size, elevation. Most if not all other parameters are repeated every time a model is run. In addition, the other parameters are in a myriad of locations within the Phast application. This makes accuracy and consistency difficult across a user base.

The small group of dedicated users will also have a difficult time keeping up with PHA demand. In a given year, a company with multiple sites can have thousands of scenarios identified. This is of course Pareto-dominant where 80% can be handled on a qualitative basis (e.g. what is the impact of a small water pump deadheading in a remote area). However, the remaining 20% will require a deeper evaluation. This can be a daunting task for a group of 3-5 people. Some corporations rely on a single person for this modeling.

Other corporations will train all PHA analysts to use Phast. While I cannot personally speak to the success rate, it seems that this will be exceptionally challenging. Again, Phast is quite comprehensive, and attempting to have the right models performed accurately and consistently across 50+ analysts seems to present a large challenge.

To remedy the issues of consistency and overburdening, one company commissioned a customized consequence analysis web application. Inside the application was a "core module" consisting of a call to the DNV Phast Web Services (PWS) API. The API contains nearly all of the modeling capability of the desktop Phast application. However, the application was developed to keep hundreds of parameters "behind-the-scenes". Again, all that is truly needed for 95%+ models are the temperature, pressure, hole size and elevation of the leak. The specific company also had requests for parameters around proximity to certain potentially impacted locations, which could be included as inputs.

The final web application took the simplified user interface and connected it to a SQL database backend. Every revalidation or question on a PHA model can be retrieved within seconds. Anytime a PHA must be revisited per OSHA PSM mandate, the existing models can be retrieved and updated as needed.

The largest benefit of this application was that, analysts now didn't have to rely on screening tools. They didn't have to request a model from the dedicated team and wait in a queue. They have the full power of the Phast application without days of required training. They did not have to worry about ensuring that the models "averaging time" is correct, that the weather conditions are representative or a multitude of other questions.

Phast Web Services can drive consistency and reduce conservatism in PHA recommendations. This has the potential for enormous cost savings. If interested in learning more about custom systems please contact us (mike@riskspace.io).