Can PHA be Big Data?

How to Make PHA Data

Work Smarter

Patrick Nonhof
Managing Partner
27 March 2019
OVERVIEW
Should Process Hazard Analysis
(PHA) Information be part of Big
Data?
The answer is YES!
• Today’s PHA Process
• Big Data
• PHA Data Limitations
• Finding a Solution
TODAY’S PHA PROCESS
& INFORMATION

3
PHA Refresher: Basic Steps

Unmitigated Mitigated
Consequenc
Cause Risk (UMR) Safeguards Risk (MR)
es
Assessment Assessment

4
Case Study: Distillation Example
To Flare

Column MAWP: 175 psig

Operating Pressure: 140-150

Blocked Outlet

5
Case Study – Cause
Tower Example – Typical Cause Information:
Unit HF Alkylation Unit
Node Debutanizer Tower
Deviatio No Flow
n
Cause Overhead liquid Control valve
FC-109 loses instrument air
and fails closed

6
Case Study - Consequence
Tower Example – Typical Consequence Information:
Consequenc Increase column pressure to
e vapor pressure of overhead
fluid at reboiler steam
temperature
i. Overpressure: Greater
than 3 x MAWP
ii. Vessel Failure

7
Case Study – Unmitigated Risk
Assessment
Tower Example – Typical Risk Determination:
Severity Catastrophic
Frequency Occasional
Unmitigated Risk 5
(UMR)

Negligible Minor Moderate Serious Catastrophic

Frequent 2 3 4 5 5
Occasional 1 2 3 4 5
Remote 1 1 2 3 4
Improbabl 1 1 1 2 3
e
Very 1 1 1 1 2
Improbabl
e
8
Case Study - Safeguards
Tower Example – Typical Safeguards:
Safeguard Shutdown PSD-301
1:
Safeguard RV-305 Set at 175 psig
2:

9
Case Study – Mitigated Risk
Assessment
Tower Example – Typical Mitigated Risk:
Severity Catastrophic
Frequency Very Improbable
Unmitigated Risk 2
(UMR)

Negligible Minor Moderate Serious Catastrophic

Frequent 2 3 4 5 5
Occasional 1 2 3 4 5
Remote 1 1 2 3 4
Improbabl 1 1 1 2 3
e
Very 1 1 1 1 2
Improbabl
e
10
 Complete Case Study PHA Scenario
CAUSE CONSEQUENCE UNMITIGATED RISK ASSESSMENT SAFEGUARDS MITIGATED RISK ASSESSMENT
Unit Node Deviation Cause Consequence Severit Freq. Unmitigate Safeguards Severit Freq. Unmitigate
y d Risk y d Risk
(UMR) (UMR)
HF Debutaniz No Flow Overhead Increase column Cat. Occasiona 5 1. Shutdown Cat. Very 2
Alkylatio er Tower liquid Control pressure to vapor l PSD-301 Impro
n Unit valve FC-109 pressure of overhead b
loses fluid at reboiler steam 2. RV-305
instrument air temperature Set at 175
and fails i. Overpressure: psig
closed Greater than 3 x
MAWP
ii. Vessel Failure

Unmitigated Mitigated Risk

Cause Consequences Risk (UMR) Safeguards (MR)
Assessment Assessment

11
BIG DATA

12
Characteristics of “Big Data”

Big Data Defined Big Data Goals

Veracit
Volume Variety Velocity Value
y

13
Characteristics of “Big Data”

PHA
PHA Data as Big Data
Big Data Goals
Volume Variety Velocity Veracity Value
One PHA: Complete PHA Reports every Address Data
1500 scenarios data sets reside 5 years inconsistencie Comparisons
Large in a variety of *Applies when s in PHA
forms: Knowledge
Petrochem you use PHA teams’
plant: 30 PHA’s PDFs, Excel, data to add
capture
wording,
One cycle: PHA facilitation context to risk analysis, Predictive
45,000 PHA systems, plus operational data rationale, etc. model
the which has a connections
scenarios that create
reference/sourc much higher
6 cycles (30 e documents: velocity. uncertainty in
years): 270,000 CAD drawings, the data
scenarios etc.
14
 PHA DATA
LIMITATIONS

15
The Big Picture

The Industry Does The Industry Doesn’t

facilitate its PHAs well manage its PHA data as well

16
The In’s and Out’s of PHAs

What Goes In What Comes Out

Manhours: Up-to-date Process Safety

• Extensive Preparation Information (PSI)
• Multi-week, multi-participant event

PHA Process
Massive Report (filed away between
Expertise Repository PHAs)
• 20 years of institutional knowledge
• From people ready to or already retired
Recommendations list with action
plans
Resources
• Logistics & budget planning (back-filling Independent Protection Layer (IPL)
team members) lists
• Retrieving all documentation for reference

17
What’s the Problem?
A paradigm shift is needed for the industry’s approach to PHAs

Micro Level Macro Level

•Single focus •Multiple altitudes
•Siloed Data •Compare and
•Snapshot in analyze
album •Easy access
18
What are the Barriers?

1. Lack of Accessibility
2. Lack of Structure

19
 Barrier: Lack of Accessibility

• Not easily searchable

• Data stored in stand-alone

software
• Restricted ability to view native files
• Different file formats

• Most common form is hard copy

binders on a shelf

20
 Barrier: Lack of Structure

• “Free text” fields used to document/embed

extremely important data:
• Process fluid
• Equipment involved
• Safeguards
• Consequence details

• Puts undue limits on the

usability of the data

21
Result: Lack of Consistency

Without intentional structure, how consistent are your PHA’s for:

= =

The Same The Same Consider:

Process? Technologies &
Do they have similar
Equipment?
causes,
consequences, or
risk levels?
Envision the Possibilities
Understand the Risk Profiles
for your units?
What if we
could use PHA Match spending with risk
data to: and tolerance levels?

Maintain centralized
documenting
and reporting?

Use analytics to make decisions?

Provide better access to your

knowledge repository?
23
FINDING A
SOLUTION

24
Step 1: Start with Structure

1. Structure improves Consistency

2. Consistency allows for Analytics

3. Analytics directs and supports

Decisions

25
Improving Structure – Cause
Tower Example – Add Cause Categories:
Unit HF Alkylation Unit
Node Debutanizer Tower
DeviationNo Flow
Cause Overhead liquid Control
valve FC-109 loses
instrument air and fails
closed
Technolo HF Alkylation
gy
Cause Blocked Outlet
Category
Process Butane
Fluid
26
Location Overhead of Debutanizer
Improving Structure –
Consequence
Tower Example – Add Consequence Categories
Consequen Increase column pressure to
ce vapor pressure of overhead
fluid at reboiler steam
temperature
i. Overpressure: Greater
than 3 x MAWP
ii. Vessel Failure
Simplified Tower overpressure
Consequen
ce
Equipment Debutanizer V-300

27
Improving Structure – Safeguards
Tower Example - Add Safeguard categories:
Safeguard 1: Shutdown PSD-301
Description
Safeguard Safety Instrumented
1: Type System
Safeguard 1
1: Priority
Safeguard 2: RV-305 Set at 175 psig
Description
Safeguard Pressure Relief Device
2: Type
Safeguard 2
2: Priority
28
Step 2: Compare the Data -
Analytics

1. Structure improves Consistency

2. Consistency allows for Analytics

3. Analytics directs and supports

Decisions

29
Analytics: Identify & Manage
Discrepancies
Now that we’ve structured the scenarios:

Identify Compare Identify

related scenarios that specific discrepancies
using scenario across to be resolved
categories. multiple PHAs or justified at
to review the scenario
patterns or level.
discrepancies.
Analytics: Compare Related
Scenarios
Sort and find related scenarios easily
• Use Cause Categories
Plant 1 Plant 2
Node Debutanizer Tower Debutanizer Tower
Deviation No Flow No Flow
Cause Overhead liquid Control Loss of instrument air to overhead
valve FC-109 loses receiver liquid flow control FIC-
instrument air and fails 315 which fails closed.
closed
Technology HF Alkylation HF Alkylation
Cause Blocked Outlet Blocked Outlet
Category
Process Butane Butane
Fluid
Location Overhead of Overhead of Debutanizer
Debutanizer
31
Analytics: Compare Consequences
Comparing Consequences & Risk
Plant 1 Plant 2
Consequence Increase column Increase column
pressure to vapor pressure to vapor
pressure of overhead pressure of bottoms
fluid at reboiler steam fluid at reboiler steam
temperature temperature
• Overpressure: • Overpressure:
Greater than 3 x Ranges from
MAWP 1.2-2.2 x MAWP
• Vessel Failure • Gasket Failure
Simplified Tower overpressure Tower overpressure
Consequence
Equipment Debutanizer V-300 Debutanizer T-5
Severity Extreme Moderate
Frequency Occasional Occasional
Unmitigated Risk 5 3
32
Analytics: Compare Safeguards
Comparing Safeguards
Plant 1 Plant 2
Safeguard 1: Shutdown PSD- PSV-1 Set at 200
Description 301 psig
Safeguard 1: Type Safety Pressure Relief
Instrumented Device
System
Safeguard 1: 1 1
Priority
Safeguard 2: RV-305 Set at 175 Alarm PAH-115
Description psig
Safeguard 2: Type Pressure Relief Alarm
Device
Safeguard 2: 2 2
Priority

33
Analytics: Compare Mitigated Risk
Compare Mitigated Risk
Plant 1 Plant 2

Severity Catastrophic Moderate

Frequency Very Improbabl

Improbable e
Mitigated Risk 2 1
(MR)

34
Analytics: Big Picture Comparisons
Compare Consequences and Risks

3 1

4 2

4 3

5 3

5 3

3 2

3 2

4 2

4 3

5 3

4 2

3 2

35
Analytics: Compare Safeguards
Compare Safeguards by
Unmitigated Risk Level
View most common/frequently
used safeguards for entire
facility:
• Basic process control system (BPCS) Other
Safeguards
Alarms, Controls and Local
• Pressure Relief Device (PRD)
• Safety Instrumented Function (SIF)
• Standard Operating Procedure (SOP)
• Other Safeguards

36
Analytics: Risk Metrics Example
Looking at Risk in a Plant...
Unmitigated Risk Value Weighted average:

Catastroph
Negligible Minor Moderate Serious
ic
Frequent 2 185 3 117 4 167 5 238 5 142

Occasional 1 233 2 183 3 313 4 460 5 259

Remote 1 24 1 13 2 12 3 50 4 22

Improbable 1 0 1 0 1 0 2 0 3 0
Very
1 0 1 0 1 0 1 0 2 0
Improbable
37
Step 3: Make Informed Decisions

1. Structure improves Consistency

2. Consistency allows for Analytics

3. Analytics directs and supports

Decisions

38
Example Continued: Risk Metrics
Calculating a Risk Profile (weighted averages):

•URV – unmitigated risk value

•MRV – mitigated risk value (after safeguards applied)

•RRV* - reduced risk value: impact of your safeguards

39
Decisions: Risk Metrics Example
•Overall – Corporate Risk Profile
Questions to ask:
• Is Plant 1’s URV
too low?
• Why is Plant 2
•Plant 1 Unit Risk Profile unmitigated risk
so high?
• Is Plant 2 taking
too much credit
for safeguards?
•Plant 2 Unit Risk Profile • Does Plant 2
have more
robust
safeguards?
40
 Decisions: Connect the Big Data
Dots
• Start with an “I wonder” statement:
“I wonder if any alarms are safeguards for more than one
scenario?”
• Examine the safeguard connections to indicators/alarms
• Filter for the facility’s 50 highest risk scenarios
• What does it tell you? Do you have one alarm that is critical for
your facility? Is that wise? Are you maintaining it accordingly? Is
your team training on it appropriately? Routinely enough?

• Don’t set the course, let the data lead – see where it takes
you! 41
Decisions: Connecting the Dots
• Key in on critical scenarios

• Analyses could help determine where to:

• Focus manpower and budget resources

• Emphasize training
• Direct mechanical integrity efforts

42
MOVING
FORWARD
Key Take-aways

• A paradigm shift is needed to

embrace Big Data concepts for
PHA Data
Micro ⇢ Macro

• Realizing Big Data possibilities for

PHA data begins with addressing
accessibility and structure issues.

• Big Data tools can move us 43

 Questions?
For additional questions, contact:
Patrick Nonhof at

Patrick.Nonhof@provpsm.com

www.provpsm.com
© 2019 Provenance Consulting. All Rights Reserved.

Patrick Nonhof
Managing Partner
27 March 2019
Improving Structure – Classify the
Data
• Can use fancy or simple methods to classify data

• Three data sets to classify/categorize:

• Causes
• Consequences
• Safeguards

• Add key words/new fields for categorization

46