Datasets:

hndc

/

AI4Hazard-small

like 0

66 GUIDELINES FOR MANAGING ABNORMAL SITUATIONS the response should be dictated by a procedure that is simplified and prescriptive as much as possible. Ho wever, it should be noted that the development of and training for such procedures is difficult to do and often includes mostly table-top/wh at-if drills, which do not properly mimic mindset stress. In cases requiring notification of staff outside the unit, or public responders, it may be useful to de velop a hand-held flip chart that directs the supervisor regarding the order of actions and notifications, provides contact numbers and other cr itical information needed for the response. While the intent is not to eliminate judgment and initiative from the response, the flip chart reduces the number of decisions the responder needs to make, thus making the responses more likely to be correct. Example Incident 3.12 and Exampl e Incident 3.13 include a loss of power event and a serious incident fo llowing an extended plant outage. Example Incident 3.12 – Lo ss of Site Power Supply The electrical power supply to an ethylene cracking plant failed, resulting in a cascade of failure of site services including steam and instrument air. The Uninterruptib le Power Supply (UPS) to the DCS failed after a matter of minutes, and lighting to the control room, which was designed to be explosio n-resistant and therefore not fitted with windows, was lost, plunging the control room into darkness. The hydrocarbon feeds to the crac ker tripped as designed in the emergency response system. Despit e this trip, the cracker tubes ruptured due to thermal stress from a loss of temperature control. The loss of temperature control resulted from a loss of steam flow that should have allowed for contro lled cooling. Cracked gas flowed back into the crackers from th e downstream quench system and ignited in several of the crackers , causing fires and further damage. The site had identified that loss of power to the site was a key safety consideration. The plant had two po wer feeds that they considered independent, and the site could oper ate on just one supply. However, they both failed due to a fault at a common substation owned by the power company.

310 Figure 12. 1 Example of Potential Design Solutions for Reactor Failure (Ref 12.3 CCPS 1998)

2.4 Ensure Open and Frank Communications |35 In summary, the process of developing mutual trust often starts with words – a declaration of intent or a request for cooperation. However, trust is truly created by deeds – living up to the words. Trust is a valuable but fragile com modity. It is hard to create and easy to violate, and once it has been violated it is difficult to regain. Therefore, process safety leaders should pay attention to trust and work to earn it every day. 2.4 EN SURE OPEN AN D FRAN K COMMUN ICATION S Over Texas and Louisiana, USA, February 1, 2003 The space shuttle Columbia broke up upon re-entry, killing all seven crew members (Ref 2.15). During the initial minutes of flight, insulating foam detached from the external fuel tank, striking and dam aging the shuttle’s heat resistant tiles. Without the tiles’ protection, the heat of re-entry melted the structural support of the wing. The resulting dam age destabilized Columbia, and the resulting stress led to its disintegration. The shuttle’s design specification stated, “No debris shall emanate from the critical zone of the External Tank on the launch pad or during ascent.” However, investigation revealed that foam loss had occurred during more than half of previous m issions, in m any cases damaging tiles. Employees and contractors at several NASA sites had noted this as a concern among their local groups, but the culture discouraged Open and Frank Communication of this concern to m anagement. One mem orandum, com posed but never sent, said, “I m ust em phasize (again) that severe enough damage… could present potentially grave hazards… Remember the NASA safety posters everywhere around stating, ‘If it’s not safe, say so’? Yes, it’s that serious.”

126 PROCESS SAFETY FOR ENGINEERS: AN INTRODUCTION Table 7.2. NFPA 704 hazards and rating Flammability (red) 0 Materials that will not burn under typical fire conditions 1 Materials that require considerable preheating, under all ambient temperature conditions, before ignition and combustion can occur 2 Materials must be moderately heated or exposed to relatively high ambient temperature before ignition can occur 3 Liquids and solids (including finely divided suspended solids) that can be ignited under almost all ambient temperature conditions 4 Material will rapidly or completely vaporize at normal atmospheric pressure and temperature or if readily dispersed in air and will burn readily Health (blue) 0 Poses no health hazard, no precautions necessary and would offer no hazard beyond that of ordinary combustible materials 1 Exposure would cause i rritation with only mi nor residual injury 2 Intense or continued but not chronic exposure could cause temporary incapacitation or possible residual injury 3 Short exposure could cause serious temp orary or moderate residual injury 4 Very short exposure could cause death or major residual injury Reactivity (yellow) 0 Normally stable, even under fire exposure conditions, and is not reactive with water 1 Normally stable, but can become unstable at elevated temperatures and pressures 2 Undergoes violent chemical change at elevated temperatures and pressures, reacts violently with water, or may form explosive mixtures with water 3 Capable of detonation or explosive decomposition but requires a strong initiating source, must be heated under confinement before initiation, reacts explosively with water, or will detonate if severely shocked 4 Readily capable of detonation or explosive decomposition at normal temperatures and pressures Special hazard (white) OX Oxidizer, allows chemicals to burn without an air supply W Reacts with water in an unusual or hazardous manner SA Simple asphyxiant gas Other symbols that are not included in NFPA 704 are sometimes used including for strong acids and bases, biohazards, radioactivity, and cryogenics

2.3 References | 23 2.9 CCPS (2007). Guidelines for Risk Based Process Safety. Hoboken, NJ: AIChE/Wiley. 2.10 CCPS (2020a). CCPS and Process Safety Publications. www.aiche.org/ccps/publications#books (accessed May 2020). 2.11 CCPS (2020b). PSID: Process Safety Incident Database. www.aiche.org/ccps/resources/psid-process-safety-incident-database (accessed May 2020). 2.12 CCPS (2020c). Process Safety Beacon Archives. www.aiche.org/ccps/ resources/process-safety-beacon/archives (accessed May 2020). 2.13 CCPS (2020d). Moving from Good to Great: Guidelines for Implementing Vision 20/20 Tenets & Themes. www.aiche.org/ccps/moving-good- great-guidelines-implementing-vision-2020-tenets-themes (accessed June 2020). 2.14 CCPS (2020e). Conferences. www.aiche.org/ccps/resources/ conferences (accessed June 2020). 2.15 CSB (2006). Combustible Dust Hazard Investigation. CSB Report No. 2006-H-01. 2.16 CSB (2016). Combustible Dust Safety. www.csb.gov/recommendations/ mostwanted/combustibledust. (Accessed January 2020). 2.17 CSB (2017a). ExxonMobil Refinery Chemical Release and Fire. CSB Report No. 2016-02-I-LA. 2.18 CSB (2017b) Didion Milling Company Explosion and Fire. www.csb.gov/didion-milling-company-explosion-and-fire-/. 2.19 DECHEMA (2020). DECHEMA ProcessNet. processnet.org (accessed June 2020). 2.20 DSS (2020). Resources. dustsafetyscience.com/resources (accessed June 2020). 2.21 EASHW (2020). Napo: Safety with a Smile. osha.europa.eu/en/tools-and- resources/napo-safety-smile (accessed June 2020). 2.22 EFCE (2019). International Symposium on Loss Prevention and Safety Promotion in the Process Industries. lossprevention2019.org/ (accessed June 2020). 2.23 EI (2020). Toolbox: Putting Safety in your Hands. toolbox.energyinst.org (accessed June 2020). 2.24 Elsevier (2020). Journal of Loss Prevention in the Process Industries. Amsterdam: Elsevier. 2.25 European Commission (2020). European Commission Major Accident Reporting System. emars.jrc.ec.europa.eu (accessed June 2020).

Pipes 89 The direction of the slope can be decided during P&ID development, but slope magnitude should be calculated during the design. The goal of sloping a pipe is to prevent stagnant liquid. In making a decision on the slope direction, the general rule is to put the slope toward the more tolerant resource. For example, in Figure 6.47a, there is a need to slope the pipe in the outlet of pressure safety valves in liquid services or two‐phase services (potentially or actually). In the majority of cases, the outlet flange of pressure safety valves dictates the horizontal pipe, and also the flow through the pressure safety valves are not continu-ous (the pressure safety valves will be opened during emergencies). Therefore, the outlet pipe should be sloped. It is not wise to have the remaining liquid stay at the outlet of the pressure safety valve, because the accu-mulated liquid at the outlet of pressure safety valve impacts its operation (refer to Chapter  12 for more details). Therefore, the slope of this pipe should be away of the pressure safety valve outlet. Similar logic is used to put slope on a horizontal pipe, that is, normally no flow (NNF). This pipe goes to a ves - sel that is considered a more tolerant system for contain-ing liquid; therefore, the slope is toward the vessel. Another example is in steam piping. In steam distribu- tion networks, there is always a chance of generating condensation in the pipe. Therefore, steam traps are installed at specific intervals to remove condensation from the steam flow, and the pipe is sloped toward each steam trap (Figure 6.48). The concept of a steam trap will be discussed later in this chapter. The last example is the pipe that directs released fluid from a pressure safety valve to a collection flare header (discussed more in Chapter  12). After collecting the release from different safety devices, the flare headers direct the liquid to the knockout drum before going to the flare. As the liquid knockout drum is placed to cap-ture the liquid from the release gas, it is more tolerant to liquid and the main header is generally sloped toward the liquid knockout drum. If the slope calculation is missed during the design phase of project, the process engineer responsible for P&ID development may decide to put a slope magnitude on pipes with triangles on them based on rule of thumbs. The minimum practical slope is about 0.08%, but the slope of the pipes can go up to 5%. The typical range of a slope is 0.5% and the typical range is between 0.2 to 1.0%. For hard‐to‐move or high‐viscosity liquids, 2–3% is not rare. Underground pipes tend to be at the lower side at 0.2–0.5%. The reason is that if the slope of underground pipe is a large value, the pipe will be in deep ground at the destination point. Choosing a lower slope value is more important when the pipe is long or the underground water level is high (like near lakes). 6.7.2 No Liquid Pock et A pipe fully carrying a liquid flow with the chance of excursion of gas or vapor can be a candidate for no liquid pocket, means “design in a way that liquid pockets natu-rally flow and exit the pipe route during the routine operation of the system” . This phrase directs the pipe modeler to design a pipe route wherein the gas flow cannot trap a pocket of liquid anywhere in the pipe route (Figure 6.49). To respond to this requirement, a piping modeler specifies a pipe route that is vertically down-ward or has a direct slope. 6.7.3 No Gas P ocket A pipe fully carrying a gas or vapor flow with the chance of generation of liquid can be considered no gas/vapor pocket if the intention is to avoid a stagnant gas pocket in the pipe during the routine operation of the system. The term no gas/vapor pocket directs the pipe modeler to design a pipe route that the liquid flow cannot trap a pocket of gas anywhere in the pipe route (Figure 6.50). To respond to this requirement, a piping modeler speci-fies a pipe route that is vertically upward or has a reverse slope. 6.7.4 Fr ee Draining (Self‐Draining) This note dictates the same requirements of no liquid pocket but during a system shutdown. This note act ually means to do the piping in a way that no liquid  r emains in the pipe after a system shutdown. NNF(a) (b) Figure 6.47 (a, b) Two examples of sloped pipe. T Steam Condensate Figure 6.48 Example of sloped pipe in steam pipe .

Application of Control Architectures 287 non‐simultaneously), but in reverse mode, while one con- trol valve is going toward opening, the other one is going toward closing (simultaneously or non‐simultaneously). Table 14.10 gives a graphic illustration of the difference between split‐range and parallel control operating on two control valves. Therefore, there could be four different modes of “multi‐valve” control. Here we explain two of them, and the two others are easy to interpret. Let’s look at the split‐straight control type. When the process parameter is at its normal level, both control valves are fully open. This means that CV1 (control valve 1 on stream 1) is fully open, and CV2 (control valve 2 on stream 2) is fully open too. When the process parameter starts to deviate from its normal level, CV1 starts to close, and CV2 remains open until CV1 closes fully. At this point, CV2 starts to close. CV1 remains closed while CV2 is operating. Now let’s look at the parallel‐reverse control type. In this mode of control, CV1 is fully open and CV2 is fully closed when the process parameter is at its normal level. Then, when the process parameter deviates from normal level, CV1 starts to close while at the same time CV2 starts to open. When reading a P&ID, we need to make sure to under - stand which types of control are used out of the four types we have introduced. Not all P&IDs will indicate whether the control is split‐range or parallel, or whether the operating mode is straight or reverse. The most com-plete P&IDs show a diagram below the control system to show the intent of the control. One distinguishing difference between split and paral- lel control is that for split control, the middle point (X%) must be mentioned on the P&ID. However, the mode of control – straight or reverse – is generally not mentioned on P&IDs. Now let’s see some examples of parallel/split control.In Figure 14.29, we have blanket gas at the top of a tank and it is important to control the pressure through two control valves, CV1 on the inlet (blanket gas stream), and CV2 on the outlet (vapor stream). We know that the type of control used is split‐range because it is written there. Some people just write 50% on the P&ID and then you know that it is split‐range control. Now, how do we work out if the two valves work in straight or reverse mode? We do this by analyzing the oper - ation. If the pressure of the blanket gas in the vessel goes too high, we open CV1 to relieve the pressure and CV2 gradually closes until the pressure is stabilized at its set point. Then as the liquid level in the tank drops, the pres - sure starts dropping and CV2 will open gradually to increase the pressure again. So it is a reverse mode operation. Table 14.10 Types of split r ange control. Straight Reverse Split CV1 CV2 CV1 Or:X% X%CV2CV1 CV2 CV1 CV2Or:X% X% Parallel CV1 CV2 CV1 CV2Or:CV1 CV2 CV1 CV2 Or: CV2 CV1PCSPLIT RANGE PT FBSplit, Reve rseCV2CV1Figure 14.29 Examples of split‐range c ontrol – blanket gas.

NOTIFICATION , CLASSIFICATION & INVESTIGATION 83 that it does not consider the pote ntial worst-case co nsequences - what could have happened. Potential severity is mu ch more difficult to determine. Hence personnel responsible for incident classi fication should be knowledgeable in process operations and receive classifica tion training to ensure consistency between different personnel. A broad knowledge of other incidents across industry is also helpful. In addition , the actual severity may not adequately reflect the complexity of the system involved, which could impede selection of the most appropriate investigation team. Examples of severity classifi cation are illustrated below. i. CCPS Guidance CCPS developed guidance on the classification of process safety incidents in 2007 as an industry laggi ng metric that would become the benchmark across the chemical and petroleum industry for measuring process safety performance. The docu ment (CCPS, 2011) was later updated to broadly align with the first edi tion of API Recommended Practice 754 published in 2010. Subseque ntly, API revised RP 754 (see Section 4.2.1.ii) in 2016 and CCPS updated thei r guidance to align with API (CCPS, 2018). The CCPS guidance is based on a tiered approach representing the severity of the incident (referred to as “process safety event”) ranging from Tier 1 as the greatest consequence (i.e., lagging metrics) to Tier 4 as proactive performance evaluations (i.e., leading metrics). Tiers 1 and 2 cover process safety incidents with co nsequences affecting safety /human health, property damage, material release, communit y impact, and offs ite environmental impact. The classification of Tier 1 incidents at four consequence severity levels is illustrated in Table 5.2. These consequence severity levels were selected primarily for reporting company and industry process safety performance purposes, and include a po ints system to indicate incident severity, which is additive if a single incident impacts several consequence categories.

78 Guidelines for Revalidating a Process Hazard Analysis Secondly, routine maintenance records can also provide insight on failure mechanisms overlooked by the previous PHA team. Consider this example of a simple piping circuit. Routine inspection detected pipe thinning, and it was replaced in- kind. A couple of years later, routine inspection detected pipe thinning, and the same piece of pipe was again replaced in-kind. No MOC was required, because in each case, the pipe was replaced in-kind. Yet the pipe is being replaced at a much- higher-than-expected frequency. Perhaps this particular piece of pipe was improperly specified for its actual service conditions. Perhaps the pipe is subject to higher flow velocities or cavitation and eroding at an unexpectedly high rate. In this example, the affected node can be Updated during the revalidation to address the actual operating history. However, additional pipe failu res were in different pipe segments, suggests a more widespread damage mechanism unanticipated by the previous PHA, or that the current production rates are exceeding the expected flow velocities throughout the unit. Such evidence in the routine maintenance records could justify the Redo approach. 4.2.4 Audit Results A five-year PHA revalidation cycle will typica lly include one or two site audits that examined the management systems for PH As and other relevant elements of the site’s process safety management sy stem (e.g., Process Safety Management [PSM], Risk Management Plan [RMP], Ri sk Based Process Safety [RBPS], Seveso). It is a good practice to look at those audit results for any findings against the particular PHA being revalidated, or ag ainst the PHA management system in general. If there were particular findin gs against the current PHA, those were likely addressed to close the audit. But that should be confirmed by the team. If there were more general findings against the PHA management system, then the study leader should evaluate the extent to which they are relevant to this particular PHA. General audit find ings like “PHAs are not documenting consequences assuming failure of all the safeguards,” or “PHAs are not using the current risk matrix,” if relevant, will almost certainly require the Redo approach for revalidation. Example - Customer Complaints The customer complaint database included several records of recent, unexpected problems with a new type of shipping container. This alerted the revalidation team to issues that might also arise when the containers were being filled in the process unit. Ultimately, the revalidation team recommended additional ITPM to reduce the risk of loss of containment events.

Appendix C – Example RAGAGEP List Recognized and Generally Accepted Good Engineering Practice (RAGAGEP) is a term used by OSHA, stemming from the selection and application of appropriate engineering, operating, and maintenance knowledge when designing, operat ing and maintaining chemical facilities with the purpose of ensuring safety and preventing process safety incidents. OSHA does not provide a specific list of RAGAGEP practices. Inst ead, these are inferred from OSHA letters of interpretation and OSHA audit findings. Once a company specifies a RAGAGEP standard, then it is committed to implementing it. An exam ple RAGAGEP list is provided in Table C.1. Table C.1. Example RAGAGEP list Topic Code or Standard Atmospheric Tanks API 620: Design and Construction of Large, Welded, Low-pressure Storage Tanks Fired Equipment NFPA 85: Boiler and Combustion Systems Hazards Code NFPA 86: Standard for Ovens and Furnaces FM 6-0: Industrial Heat ing Equipment, General FM 6-9: Industrial Ovens and Dryers FM 6-10: Process Furnaces FM 7-99: Hot Oil Heaters API 521: Pressure-Relieving and Depressuring Systems API 537: Flare Details for General Re finery and Petrochemical Service Flammable Liquids NFPA 30: Flammable and Combustible Liquids Code NFPA 77: Recommended Practice on Static Electricity Heat Exchangers TEMA: Standards of the Tu bular Exchanger Manufacturers Association API 510: Pressure Vessel Inspection Code: In-Service Inspection, Rating, Repair, and Alteration Instrumentation and Controls ISA-18.2 Management of Alarm Systems for the Process Industries ISA-84.91.01 Identification and Mechanical Integrity of Safety Controls, Alarms, and Interlocks in th e Process Industry ISA-84.00 Functional Safety: Safety Instrumented Systems for the Process Industry Sector ISA-101 (Draft) Human Machine Interfac es for Process Automation Systems Plant Buildings API 752: Management of Hazards Associated with Location of Process Plant Permanent Buildings API 753: Management of Hazards Associated with Location of Process Plant Portable Buildings Pressure Vessels ASME Sect ion VIII – Pressure Vessels API 510: Pressure Vessel Inspection Code: In-Service Inspection, Rating, Repair, and Alteration

234 INVESTIGATING PROCESS SAFETY INCIDENTS Investigators should keep developin g the tree until they find issues, such as: • What was the management system involved in this failure? • Why did the management system for plant surveillance, test, or inspection programs fail to detect the incipient failure? • Why did the preventive maintena nce program at the plant not prevent the failure? • If the failure resulted directly from a human error, what was the underlying reason for this error? For components or devices suppli ed by outside manu facturers, the downward progression is usually stopped at the component level, unless the device is normally opened, repa ired, calibrated, adjusted, or inspected by in- house personnel. Electronic black boxes (similar to those under the hood of our automobiles) are good examples. Owners may have occasion to manipulate the connection points (wires , attachment, and securing brackets) but typically do not open them or attempt to diagnose an internal malfunction. Alternatively, certain systems are assembled and maintained by operators of chemical plants. For exam ple, various components of a control valve system may be purchased separately and then assembled and configured by plant personnel. The incident investigation team would investigate possible accident causes associated with the methods of integration, assembly, maintenance, inspection, and calibration of the control valve system. Nevertheless, if a malfunction of a factory-sealed sub- component were involved, the incident investigation team would seek out the appropriate expertise. The team would usually not attempt to analyze any factory-supplied components that normally remain sealed without additional help. If the malfunction co ntributed to the in cident, it should be investigated until it is understood, especially if similar components are in use elsewhere. Another guideline is to stop the development of the tree when the events become external to the point that they can no longer be controlled by the organization. There are significan t differences in the ability to control internal events as opposed to external events. Company “A” may experience a massive explosion and toxic vapor re lease that injures employees at the adjacent plant of Company “B.” Investigators and mangers at Company “B”

EDUCATION FOR MANAGING ABNORMAL SITUATIONS 105 Initial and refresher training that is appropriate to the HMA Maintenance training that covers the specific maintenance requirements for the selected HMA 4.3.2.3 Emergency Procedures Some abnormal situations in the pl ant or process could result in the need to initiate an emergency r esponse. Training on historical emergency situations, prepared and proven emergency procedures, and establishing a line of communicati on is therefore recommended. With this in mind, responding to emergencies is typically less effective if the response requires lo cating, reading, and executing an emergency procedure. Best practice involves operating teams conducting drills on emergency proc edures during off-shift hours. The drills are typically scheduled monthl y, or more often, so that all emergency procedures are drilled at least once per year. A typical drill could consist of an experienced oper ator leading the shift team through a brainstorming session that uses team input to recreate the emergency procedure without actually referring to it. A scribe records the shift team interaction and creates a step-by-step emergency procedure that, at the end of the exercise, is then compared to the actual procedure. The team then discusses any steps that were missed or added, and the updated procedure is provided to the operat ions trainer for further review and critique. Each shift team completes a similar exercise for each emergency procedure. The scribed procedures from each shift are collected by the operations trainer and reviewed for new and appropriate or missed steps. At the end of the exercise for each emergency procedure, the procedure has been: Drilled from memory Reviewed for missing steps Reviewed for correct order Reviewed for adequate time to respond Revised based on all operator teams’ inputs Applied MOC process to ensure the revised procedure is formally updated and recorded.

5.5 References |201 and approval should be conducted with the sam e diligence the approver would use if they were going to have a fam ily member perform the task. Anyone involved in the permitting process or in conducting the work itself should be able to feel secure in voicing objections or pointing out potential risks or flaws in job preparation activities. Leaders should use audits and informal walk-throughs to verify that the safe work practices element is functioning correctly. Field visits also im portant to allow leaders to correct any errors, reinforce good behaviors, and identify improvement opportunities. 5.5 REFEREN CES 5.1 Center for Chemical Process Safety (CCPS), Guidelines for Risk B ased Process Safety, American Institute of Chemical Engineers , 2007. 5.2 American National Standards Institute/American Petroleum Institute, Process Safety Performance Indicators for the Petroleum and Petrochemical Industries , ANSI/API RP-754, 1st Ed, 2010. 5.3 Center for Chemical Process Safety, Process Safety Leading and Lagging Metrics … You Don’t Improve What You Don’t Measure, American Institute for Chemical Engineers, 2011 5.4 Center for Chemical Process Safety (CCPS), Hazard Evaluation Procedures, 3rd Ed., American Institute of Chemical Engineers, 2007. 5.5 Center for Chemical Process Safety (CCPS), Guidelines for Asset Integrity Management , American Institute of Chemical Engineers, 2016. 5.6 Center for Chemical Process Safety (CCPS), Guidelines for Writing Effective Operating and Maintenance Procedures , American Institute of Chemical Engineers, 1996.

7.6 Interpersonal Intelligence | 99 signs along the way depicting how the incident unfolded. As workers walk the path, they learn how the incident occurred and how to prevent it. The learning model scenario in Chapter 14 uses this method of learning with a hands-on contest where the challenge is how to handle a simulated ammonium nitrate incident. An inexpensive and fun way to help colleagues appreciate fire and explosion hazards is to take them to dinner at a Japanese steakhouse. After the chef performs the ritual of igniting a few milliliters of 190 proof alcohol (a practice that has been determined by experience to produce a burst of flame that makes patrons uncomfortable but doesn’t hurt them), ask him to share with the group what quantity he used. Then have the participants guess the volume of flammable materials in a typical industrial spill and how far they would need to be from the deflagration in order to be safe. After dinner, walk with the team away from the restaurant, stopping at the distance where the radiation from the hypothetical fire would have dissipated to the level they felt during the indoor demonstration. Participants will likely be surprised how far away they had to walk. Here is another fun way to promote kinesthetic learning of the proper response to person-down incidents. While you are training a group or leading a meeting, collapse to the ground at a random time, pretending to be overcome. If you are indoors, stand near a source of outside air such as a vent. Before collapsing, draw attention to the vent by pointing at it and asking the group, “Do you smell that?” When someone rushes to help you (for the first person, you could enlist a helper in advance), hand them a card that looks like the text box to the right. Depending on how well employees understand person-down procedures, you may need to have several cards ready to hand out. After an appropriate number of mock fatalities, revive yourself and discuss with the group what everyone should have done. 7.6 Interpersonal Intelligence People with interpersonal intelligence learn best where they can discuss the lessons learned with a group. This fits well with the format of safety meetings and toolbox talks. It can also fit well as part of safety moments at the start of regular business meetings, providing that the moderator engages the participants and doesn’t simply lecture. You’re dead. Fall down. Hand this card to anyone who tries to help you.

390 PROCESS SAFETY FOR ENGINEERS: AN INTRODUCTION The risk assessment in an MOOC should focus on the competence of the people involved and the system in which they conduct their wo rk including the time available. The risk assessment should consider not only the risks of the change but also the risks during the change. An organizational change puts demands on top of uncertainties resulting in increased stress and the potential for more human pe rformance issues (Chapter 16). The risk assessment should include consideration of the change on all operating conditions including emergency response, e.g. impacts on emergency response crew numbers. What a New Engineer Might Do New engineers will likely be involved in the Ma nagement of Change process. They may be requesting a change, participating in the hazard review of a change, closing action items created by the change, or tracking changes as they progress through the MOC system. In all cases, a new engineer should understand the MOC sy stem at their facility. A key part of this is understanding what is, and what is not, a change. It can be easy to believe that a change is minor enough that it won’t impact process safe ty; however, many sign ificant process safety events have resulted from what was consid ered to be a minor change at the time. Tools Resources to support Management of Change include the following. CCPS Guidelines for the Management of Change for Process Safety . This book provides guidance on the implementation of effective and efficient Management of Change (MOC) procedures, which can be applied to improve pr ocess safety. In addition to introducing MOC systems, the book describes how to design an MOC system, including the scope of the system and the applications over a plant life cycle and the boundaries and overlaps with other process safety management systems. (CCPS 2008) CCPS Guidelines for Managing Process Safety Risks During Organizational Change . This book provides an understanding of the management of organizational change which is essential for successful corporate decision making with little adverse effect on the health and safety of employees or the surrounding community. Addressing the myriad of issues involved, this book helps companies bring their MOOC systems to the same degree of maturity as other process safety management systems. Topics include corporate standard for organizational change management, modification of working conditions, personnel turnover, task allocation changes, organizational hierarchy changes, and or ganizational policy changes. (CCPS 2013) Summary It is inevitable that changes, permanent or te mporary, will be made to a facility over its life cycle. The intent of Management of Change is to ensure those changes don’t inadvertently introduce new hazards or remove any existing risk prevention or mitigation measures. The MOC process starts with identifying a change. This is a key step as changes that are not identified will not be managed. The MOC involves a hazard review which can be simple or can be as complex as a HAZOP. Action items identified in the MOC process may be required either

9 • Other Transition Time Considerations 171 The specific commissioning steps and the reviewed and approved procedures for eq uipment and process start-up steps, which include the safe operating limits, the consequences of deviation, etc. ( Note : These procedures are, at best, thorough drafts that ty pically need updating based on the situations detected and resolved during the commissioning and start-up execution. Any changes should be reviewed and approved th rough a change management system.); And, once normal operations have been achieved for the continuous or batch process (s ee definitions in Table 2.1), performing test runs to verify that the performance goals, such as throughput and product quality, have been achieved. Additional guidance when preparing for the initial start-up of a process unit includes using a checklis t to inspect the following [87]: personnel safety, vessels, heat exch angers, columns, reactors, piping, machinery, electrical, and instrumentation. Rotating equipment checks include rotational direction, bearing temperature, and vibration during the “run-in” of the motors before they are coupled to their respective drives. If hydro testing or moist air was used, they may hold residual water that needs to be removed by drying the system, especially if the system needs to be dry during normal operation. When commissioning furnaces, refractories need to be heated slowly to expel residual water and help le ngthen the heater’s life. Special steam heating and drying protocols shou ld be used on the fuel-gas lines before lighting the pilot burners and raising the burner temperatures. General guidance for catalyst loading includes low-density and high-density catalyst sy stems. Tightness tests are used to confirm that process units handling hazardous materials do not leak. Vacuum systems must be leak tested , as well. The final piece of guidance focuses on Nitrogen inerti ng systems, describing different methods such as evacuating with steam ejectors or steaming out the

9 Other Transition Time Considerations 9.1 Introduction This chapter introduces the types of projects that are associated with other transition times, when the processes are not in normal operations, providing an overview of the equipment and process unit life cycle (Section 9.2) and the t ransition times associated with these life cycle-related projects. The commissioning and initial start-up projects, the first transition time discussed in this chapter, involves new equipment, new process units, or greenfield facilities (Section 9.2.1). The second transition time involves the end-of-life projects (Section 9.2.2). Effective handovers between groups, as noted earlier in this guideline, are essential fo r effectively managing the process safety risks during these transition ti mes, as these projects often have specialized contractors who have specialized technological or decommissioning expertise (Section 9.2.3). Special commissioning and initia l start-up considerations are discussed in Section 9.3, followed by a review of incidents and lessons learned during the transition time between construction and start-up (Section 9.4). Section 9.5 provides gui dance on specific end-of-life shut- down considerations. The two decommissioning-related transition times discussed next are th e temporary—mothballing—or the permanent, decommissioning s hut-down times. Mothballed processes and equipment are temporarily shutdown for an unknown period of time, requiring some fo rm of preservation during the shutdown period (Refer to Section 5.3.4). Mothballing considerations are covered in Section 9.6, follow ed by a review of incidents and lessons learned which occurred du ring these transition times in Section 9.7. Specific decommissioning considerations are discussed in Section 9.8, with corresponding incidents and lessons learned Guidelines for Process Safety During the Transient Operating Mode: Managing Risks during Process Start-ups and Shut-downs . By CCPS. © 2021 the American Institute of Chemical Engineers

CONSEQUENCE ANALYSIS 273 Reduction in flow: Valves, pumps, or other restrictions in the piping that might reduce the flow rate below that estimate d from the pressure drop and discharge area. Inventory in the pipe or process between the leak and any isolation device. Hole Sizes A primary input to source calculation is the le ak hole size. Holes occur in process equipment due to corrosion, impact, fatigue, brittle fracture, and other mechanisms. The mechanism can influence whether a small hole or a full-bore pipe rupture is likely. No general consensus exists for appropriate hole sizes. Analysts use a vari ety of approaches for hole size including the following. World Bank (1985) suggests characteristic hole sizes for a range of process equipment (e.g., for pipes 20% and 100% of pipe diameter are proposed). Some analysts use 50 and 100 mm (2 and 4 in) holes, regardless of pipe size. Some analysts use a few hole sizes to repr esent the full range possible, such as 5, 25, 50, and 150 mm (0.2, 1, 4, and 6 in) and full-bore ruptures for pipes less than 152 mm (6 in) in diameter. IOGP data set provides a means to estima te leak frequencies spreading the total frequency across any number of hole sizes selected. (IOGP 2019) Discharge Phase Source models require careful consideration of the discharge phase. This is dependent on the release process. Standard texts on vapor-liquid equilibrium or commercial process simulators provide useful guidance on phase behavior. The st arting point is defined by the initial condition of the process material before release. This ma y be normal process conditions, or an abnormal state reached by the process material prior to th e release. The end point will normally be at a final pressure of one atmosphere. Table 13.1 is a partial list of typical scenario s grouped according to the material discharge phase, i.e. liquid, gas, or two-phase. Different models are appropriate for each of these. Figure 13.5 shows selected discharge scenarios with th e resulting effect on the material's release phase. Gasket failure, either full or partial, is often the cause of liquid or gas discharges from equipment leaks.

318 PROCESS SAFETY FOR ENGINEERS: AN INTRODUCTION Each of these steps has an associated frequenc y or probability. Typically, the initial loss of containment or initiating event is described by a frequency value and the subsequent steps in the scenario story are described by probabilitie s. Frequency analysis develops the specific frequencies associated with the specific potent ial scenario outcomes, as illustrated in Figure 14.4. Combining each frequency with the relevant consequence yields the risk. Historical Records This approach is typically used to quantify th e initial step in the story, e.g. the loss of containment. However, it is not as simple as it sounds. Locating the data can be challenging. Companies may develop their own data sets; howeve r, this requires significant effort and the availability of a large database in order to create a data set that is statistically valid. Publicly available data sets are described in Section 14.9. Once data are located, they should be validat ed that they are applicable to the scenario being analyzed by considering points such as the following. Did the data come from analysis of the same type of equipment in the same enviro nment? Is the data current or does it reflect equipment technology that is no longer predominately used? Fault Tree Analysis Fault tree analysis was described in Section12.3.6. A fault tree can be used to estimate incident frequencies. Fault tree analysis can calculate th e hazardous incident (top event) frequency using the fault tree model of the system failu re mechanisms. For example, the analyst may wish to calculate the frequency of a toxic releas e for a reactor overpressure, but this frequency is not available in the historical records. A fa ult tree could be constructed of the event using the frequency of loss of reactor cooling, the pr obability that safety interlocks fail, a runaway reaction occurs releasing toxic gas, and the probability that the emergency scrubber system fails. The fault tree describes the potential causes leading to the top event and that the logic (and/or gates) used. Once the fault tree structure is validated qualitatively, then the frequency of the top event may be calculated from the frequency values, probability values, and logic gates in the fault tree. FTA is well described in Guidelines for Chemical Pr ocesses Quantitative Risk Assessment . (CCPS 1999) Event Tree Analysis Event tree analysis was described in Section 12.3. 7. An event tree can be used to quantitatively estimate the distribution of incident outcomes (e.g. explosion, flash fire, VCE, safe dispersion). In an event tree, the branches are typically a yes/no decision such as the following. The release ignites immediately, or it does not. The vapor cloud ignites sometime later, or it does not. People are in the impact area, or they are not. These yes/no decisions can be described with probabilities as their sum must be equal to one. The branches may also be probabilities su ch as wind towards a vulnerable population. The initial incident is typically described by a frequency. The product of the initial frequency

GLOSSARY xxxiii Incident An event, or series of events, resulting in one or more undesirable consequences, such as harm to peop le, damage to the environment, or asset/business losses. Such events incl ude fires, explosions, releases of toxic or otherwise harmful substances, and so forth. Incident investigation A systematic approach for determinin g the causes of an incident and developing recommendations that address the causes to help prevent or mitigate future incidents. Independent Protection Layer (IPL) A device, system, or action that is capable of preventing a scenario from proceeding to the undesired consequence without being adversely affected by the initiating event or the action of any other protection layer associated with the scenario. Note: Protection layers that are designated as "independent" have specific functi onal criteria. A protection layer meets the requirements of being an IP L when it is designed and managed to achieve the following seven core attributes: Independent; Functional; Integrity; Reliable; Validated, Maintained and Audited; Access Security; and Management of Change. Individual risk The risk to a person in the vicinity of a hazard. This includes the nature of the injury to the individual, the likelihood of the injury occurring, and the time period over which the injury might occur. Inherently Safer Design (ISD) A way of thinking about the design of chemical processes and plants that focuses on the elimination or reduction of hazards, rather than on their management and control. Inspection, Testing, and Preventive Maintenance (ITPM) Scheduled proactive maintenance activities intended to (1) assess the current condition and/or rate of degr adation of equipment, (2) test the operation/functionality of equipment, and/or (3) prevent equipment failure by restoring equipment condition. Jet fire A fire type resulting from the discharg e of liquid, vapor, or gas into free space from an orifice, the momentum of which induces the surrounding atmosphere to mix with the discharged material. KSt value The deflagration index of a dust cloud. It is a dust-specific measure of the explosibility, in units of bar-m/s. Not that it is not a physical property of a substance, but dependent on particle size, test conditions, etc. The equation is the so-called cubic /cube root law. Interlock A feature that makes the state of tw o mechanisms or functions mutually dependent. It may be used to prevent undesired states in a finite-state machine, and may consist of any elec trical, electronic, or mechanical devices or systems. (Wikipedia) Intrinsically safe Equipment in which any spark of any thermal effect produced. Including normal operation and specified faul t conditions, are not capable of causing ignition of a given explosive gas atmosphere.

211 8.81 Vernon, R., Cohrssen, B., Patty's Industrial Hygiene and Toxicology, Sixth Edition. John Wiley and Sons, 2010. 8.82 Wells. G., Rose, L., The Art of Chemical Process Design, 266. Elsevier, 1986. 8.83 Yaws, C., Chemical Properties Handbook. McGraw-Hill, 1999. 8.84 Yoshida, T., Wu, J., Hosoya , F., Hatano, H., Matsuzawa, T., and Wata, Y. Hazard evaluation of dibenzoylperoxide (BPO), Proc. Int. Pyrotech. Semin.. 2 (17), 993-98, 1991. 8.85 Zabetakis, M., Bureau of Mines Bulletin 627, Flammability characteristics of combustible gases and vapors. National Technical Information Services, 1965.

(VJEFMJOFT
GPS
*OWFTUJHBUJOH
1SPDFTT
4BGFUZ
*ODJEFOUT
5IJSE
&EJUJPO By 5IF
"NFSJDBO
*OTUJUVUF
PG
$IFNJDBM
&OHJOFFST Copyright
¥

5IF
"NFSJDBO
*OTUJUVUF
PG
$IFNJDBM
&OHJOFFST 366 APPENDIX C. PROCESS SAFETY EVENTS LEVELING CRITERIA The following table is an example of a logic tree approach to determine the incident classification level (“leveling”) described in Chapter 5. The table is used as guidance to determine whether or not a safety injury or fatality precursor or potential event should be in vestigated. It applies only to Process Safety processes and equipment that are determined to be high risk by local regulations and company policy.

238 the heart of the inherently safer de sign philosophy – to first ask if hazards can be eliminated from the process, or if they can be significantly reduced. De signing of safety systems to manage and control hazards should not proceed until it has been firmly established that it is not feasible to eliminate or redu ce these hazards from the start. Basic process selection is at the core of inherent safety at the pre- design stage, for this is the point where the basic chemistry and selection of unit operations is established. Where one or more alternative processes for producing the desired product exist, they should be compared in terms of their inherent hazards, such as raw materials used, intermediates and wastes produced, and operating conditions including temperature and pressure. Selection of the inherently safer process is not always straightforw ard, and a number of conflicts may exist that need to be reconciled to op timize the level of risk against other factors (see Chapter 13). Ankers (Ref 10.2 Ankers) describes a software application for identifying inherently safer process options, with an emphasis on early hazard identifi cation. A number of methods have been developed for measuring and comparing the relative level of inherent safety between two or more processes. These are discussed later in this chapter. The use of corporate design standa rds, which incorporate the use of applicable inherently safer design fe atures can also be beneficial in providing guidance for designers of new processes and helping to establish company expectations and decisions regarding risk minimization. Many companies have es tablished internal design guides that incorporate the use of IS design s for new facilities, as well as for major modifications to existing facilities. Many companies rely on engineerin g contractors for design work at various levels, from detailed engin eering of a process developed by the customer, all the way to “turnkey ” licensed technology and plants. Companies should clearly define expectations for IS considerations in the design process, before contracts are awarded. Some contractors may not be aware of IS concepts and ho w to implement them, so providing an engineering standard on how these concepts are to be considered is a crucial factor. The company should provide guidance as to how IS

122 Human Factors Handbook • Commitment to continual improvement of the Competency Management Systems: Warwickshire Oil Storage Ltd was aware that competency requirements and job roles can change and recognized the importance of regular job and task analysis. 10.5 An example of gaps in operational competency The Esso Longford gas explosion in Austra lia in 1998 was an industrial accident with severe consequences [20]. It is summarized in B.3 (page 387). Several factors contributed to the accident, including: • Engineers being relocated off-site; • A focus on lost time injury rates; • Poor audits; • Management control failure; • Inadequate regulatory systems; • Government failure to provide alternative gas supply; and • Market forces leading to a cost-cutting business strategy [48]. Deficiencies in operators’ knowledge, due to flaws in training and operating procedures, were reflected in their actions on that day. Operators and supervisors focused on Gas Plant 1 (GP1), specifically on the leak in flanges of heat exchanger - GP922. Operators’ steps to restart the lean oil pumps were intended to restore heat in GP922, and to reduce the temperature differential across flanges. This was thought to be responsible for the leaks. The operators and supervisor present on the plant on the day of the accident were highly experienced individuals, yet no-one recognized the hazards associated with the plant conditions. The gaps in knowledge were due to failure of training programs, as noted in the Royal Commission Report [20]. “Though the existence of a link between this failure and the occurrence of the accident is hard to evaluate, a ppropriate management of change risk assessment may have exposed important and relevant weaknesses in the level of operator knowledge, in training programs, in communication systems, in operating procedures and in other aspects of Esso’s management system.” [20]

167 common type is the printed circuit heat exchanger in which a channel is chemically etched into a plate. Hydraulic diameters in the 50-200m range are possible. The plates are then stacked and bonded together. Very high temperature (900 C) and pressure (500-1000 bar) applications are possible with this type of heat exchangers. The main lim itation of a microchannel heat exchanger is high pressure drops across the channel. One caution associated with compact heat exchangers, particularly plate and spiral exchangers, is that they often contain a significant amount of gasketed surfaces, which may increase the likelihood of leakage. Additional work has been performed in recent years to develop multi-function heat exchangers th at combine heat exchange and reaction unit operations in one devi ce (Ref 8.74 Stankiewicz). Several such devices already exist, includin g catalytic plate reactors, where a plate-type heat exchanger is coated with a reaction catalyst. A heat exchange reactor has to meet several design objectives: The residence time in the device must be sufficient to complete the desired reaction. The fluid temperature must be controlled, implying high heat transfer coefficients. If the feed and reactant are not pre-mixed well, the channel geometry must create turbulence that is sufficient to accomplish adequate mixing. The pressure drop across the device must be acceptable. Some compact heat exchanger desi gns meet these characteristics and have high heat and mass transfer coefficients even at low flows and have flows that are turbulent enou gh (Re > 300) to ensure adequate mixing. However, additional design wo rk needs to be done before such devices are ready for widespread app lication in the process industries. Piping . Inventory in piping systems can represent a major risk. For example, a quantitative risk analysis of a chlorine storage and supply system identified the pipeline from the storage area to the manufacturing area as the most important contributor to total risk (Ref 8.45 Hendershot). To minimize the risk associated with transfer lines, their lengths should be minimized by careful attention to unit location

336 INVESTIGATING PROCESS SAFETY INCIDENTS 15.6 INVESTIGATION FOLLOW -UP REVIEW Table 15.5 offers prompts to evalua te the effectiveness of incident investigation follow-up. Not all options are appropriate for all investigation management systems or every investigation. The reader should determine which sh ould be used and where. Table 15.5 Example Follow-Up Checklist Follow-Up Issues Addressed? Yes No 1. Are the incident investigation follow-up ex pectations clearly stated in the incident investigation policy statement? 2. Does the incident investigation management system include: – Strong encouragement for near-miss reporting and investigation? – Requirements for formal periodic status reports of recommendations? – Requirements for documentation of a formal plan for sharing lessons learned? – Provisions for providing appropriate report information to various levels as needed? – Provisions for modifications of original recommendations? 3. Are appropriate levels of upper management aware of and involved in monitoring the implementation or reso lution of recommendations and resultant action plans? 4. Have audit protocols been esta blished that include examination of effective implementation of: – Investigation follow-up measures? – Recommendations? 5. Are incident investigation follow-up expectations included in training and competency systems? 6. Are actions from investigations being co mpleted within the specified timescale? 7. Was the implementation of the recommendations effective? 8. Has the investigation team leader prov ided the members of the investigation team and their supervisors structured f eedback on their performance throughout the investigation?

14. Operational competency assessment 157 14.3.3 Post learning assessment Assessment of learning should be conducted during learning opportunity, and (for example) 4-6 weeks after, to allow fo r implementation of knowledge. This assessment would aim to evaluate whether competency has been reached at this stage. Examples of post-learning assessment methods are shown in Figure 14-1. The figure provides suitable assessment methods for “acquired learning” and for “application of learning into practice”. Figure 14-1: Learning assessments 14.4 Reassessment Individual and group competency should be maintained over time and reassessed to prevent skill fade. For example, refresher tr aining is important for safety critical roles and infrequent tasks. The reassess ment requires use of methods that are suitable for assessing competency and human performance, as shown in Table 14-1. Reassessment should focus on technical knowledge or expertise, and skills application. It should cover the regular activities and tasks that individuals perform. For example: • A highly critical task that is perf ormed infrequently may require regular reassessment and refresher training, such as every year. • A medium critical task that is performed very frequently may require reassessment every three to five years.

290 | Appendix E Process Safety Culture Case Histories What m essages did the Engineering M anager send about process safety culture? Defer to Expertise, Combat the Normalization of Deviance. E.3 Taking a Minim alist Approach to Regulatory Applicability A specialty chem ical facility that produces many products has several product fam ilies that involve highly exotherm ic reactions. The facility has several norm al and emergency cooling systems for the reactors that produce these products, including back-up diesel emergency generators. The feed m aterials are both toxic and/or flamm able and are highly volatile. The reactors process chemicals addressed by regulation, but the final products are not covered, are not highly toxic or flammable and have low vapor pressures. The reactors have m ultiple B PCS and SIS system s that m onitor and control reactor temperature, pressure, and level, as well as dual relief devices. The facility defined the regulatory boundaries of the facility to include all equipm ent from raw m aterial storage to just before the first valve downstream of the reactors. They argued that since the products were not regulated, the equipment handling them need not be addressed in the PSMS. Note that the valve is a remotely operated by instrum ent air and opens and closes autom atically based on the tem perature in the reactor. The facility also excluded the cooling system s for the reactors, including the backup power system s from the PSM S, since water and power are not regulated, and in any case, other reactor safeguards protect the reactor in case of therm al runaway. The regulatory manager corporate legal have reviewed and approved the PSM S boundaries. What could be the impact of excluding utilities, back-up power, and the downstream valve from the PSMS? B ased on Real Situations

(VJEFMJOFT
GPS
*OWFTUJHBUJOH
1SPDFTT
4BGFUZ
*ODJEFOUT
5IJSE
&EJUJPO By 5IF
"NFSJDBO
*OTUJUVUF
PG
$IFNJDBM
&OHJOFFST Copyright
¥

5IF
"NFSJDBO
*OTUJUVUF
PG
$IFNJDBM
&OHJOFFST 278 12 DEVELOPING EFFECTIVE RECOM M ENDATIONS Using structured approach es such as those descr ibed in the preceding chapters, an investigation team identifi es the causal factor s and root causes of the incident. These approaches pr ovide the mechanis m for understanding the interaction and impact of managem ent system defici encies. When the investigators understand what happened, how it happened, and why it happened, they can develop recommenda tions to help prevent a recurrence of the incident. Effective recommendations can reduce risk by improving the process technology, upgrading the operating/ma intenance procedures or practices, and most critically, improving the mana gement systems. Recommendations that correct management system failures should either eliminate or substantially reduce the risk of recurrence of the incident as well as other similar incidents. This chapter describes the char acteristics of high quality recommendations necessary to prevent future incidents, as detailed in Chapter 14. The first section is a presentation of the major concepts related to recommendations, such as attributes of good recommendations, management of change, and inherent safety. The second section expands on the attributes and presents a systematic discussion of the flowchart for recommendations. 12.1 KEY CONCEPTS Figure 12.1 presents an over view of the activities in this chapter, beginning with the system-related causes already identified. The cause(s) should be addressed by recommended preventive or mi tigative action item(s). In some cases, the incident investigation team is responsible for developing the recommended actions, and then presen ts these recommendations to the management team responsible for accept ing, modifying, or rejecting these recommendations. Consultation with the management team is important in order to establish ownership of the re commendations and to address issues such as priority and timeline. In other cases, the responsibility for developing some or all of the recommendations lies with the management team,

153 Cyanides Toxic and flammable gas generation Fluorides Toxic gas generation Epoxides Heat generation, polymerization Combustibles Oxidizing agents Explosion Anhydrous Chromic Acid Spontaneous ignition Potassium permanganate Spontaneous ignition Sodium peroxide Spontaneous ignition Alkali Nitro compounds Easy to ignite Nitroso compounds Easy to ignite Ammonium Salts Chlorates Explosive ammonium salts formed Nitrites Explosive ammonium salts formed Alkali Metals Alcohols, Glycols Flammable gas and heat generation Amides, Amines Flammable gas and heat generation Azo- and diazo- compounds Flammable gas and heat generation

4 • Process Shutdowns 70 Guidance to help prevent ma intenance-related incidents during the shutdown, and some examples of incidents occurring during maintenance due to inadequate preparations after the shut-down [20, pp. 12-17]. Guidance for removing the hazards from equipment before the shutdown activities begin [20, pp. 17-23]. Addition maintenance-related guidanc e is provided in more detail, as well [2, p. Chapter 7] [20, p. Chapters 1 and 23]. Some specific incidents during the shut-down and start-up transition times are noted next. The incident summary is provided in the Appendix. 4.7.1 Incidents during shut-downs for planned project-related shutdowns C4.7.1 -1 – Delaware City Refinery Company (DCRC) Equipment Preparation [39] Incident Year : 2015 Cause of incident occurring du ring the preparation time : Leak of hydrocarbons through a “closed” sing le block valve while personnel were preparing equipment for maintenance work by de- inventorying and draining vessels located between two isolation points. Incident impact : Release of hydrocarbons into the sewer, which ignited and caused a flash fire that injured an operator with second - and third -degree burns Risk management system weaknesses : LL1) Operational tasks for preparation for maintenance can be uncommon and non -routine. The hazards and risks should be assessed when preparing the equipm ent, including establishing clear procedures to perform the ta sk safely. Any changes to the pre- plan during equipment preparation should follow a structured change management protocol.

Evaluating Operating Experience Since the Prior PHA 73 and approved?” The MOC and PSSR records should provide the answers to those questions. Usually changes can be categorized as: • Resolutions of process safety or environmental recommendations (e.g., from PHAs, incidents, or audits) • Resolutions of process improvement recommendations • Temporary modifications or impairments When deciding on a PHA revalidation approach, resolution of a safety or environmental recommendation alone rarely warrants a Redo . For example, if a recommendation called for providing additional protection against overheating a reactor, a high-high temperature interlock to shut off the steam supply could be installed as resolution of the recommendation. This type of change can be handled by Updating the safeguards in the reactor node. The resolution of process improvem ent recommendations may also be amenable to the Update approach if they resulted in relatively few, simple changes. However, as previously discus sed, numerous and/or complex changes that were analyzed or documented differently than the PHA may require that the Redo approach be applied to the new or otherwise affected nodes. The status of temporary changes. Temporary modifications should be restored to their original config- urations within a specified, approved timeframe. From a revalidation perspective, two situations might occur. One possibility is that a temporary modification might coin- cidentally be active at the time of the revalidation, such as an experimental run of a new product or an engineered clamp on a leaking pipe awaiting replacement during the next shutdown. In most cases, the revalidation team simply notes that change and the date on which it will be reverted to original operating conditions. The temporary MOC addresses the hazards of that change for the authorized period, and the Example - Temporary MOCs Company A approved a temp- orary modification of its heat tracing until the earliest known freeze date for that location. If the restoration work could not be completed by that date, the MOC would require revision and formal approval of an extension via the MOC process. Since the change is not permanent and hazards are addressed in the MOC, the PHA need not be updated if the revalidation happened to occur during the duration of the change.

230 | 6 Where do you Start? to stop an unsafe process. As discussed previously, operators have a front row seat to view the process, so they are typically the first to detect a problem . They are also the closest when an incident occurs. Em powering shut-down authority and assuring there will be no reprisals for doing so goes a long way to creating good chemistry for process safety culture. Sim ilarly, technical and process safety experts are also well- placed to understand if processes or anticipated changes or start- ups are unsafe. Respecting that expertise and assuring no reprisals also is key to establish the right chemistry. Control the change. As the process safety culture, and with it PSM S perform ance improves, it is important to monitor conditions closely. In addition to the various metrics that are defined, tracking the normalization of deviance can serve as a useful control point. As culture improves, deviance should begin to decrease. Similarly signs of slippage in the culture can quickly be observed through increased normalization of deviance . Reassess and Im prove. Chapter 7 will address the sustainability of the process safety culture. Some of the above themes deserve additional discussion. Leadership Two things distinguish effective leaders: 1) the am ount of time spent monitoring worker performance (work sam pling) and providing appropriate feedback, and 2) listening to em ployees and contractors, and providing them with an environm ent that m akes it easier for them to succeed. Generally, leaders can best achieve this with “Leadership-by-walking-around.” Quite simply, leaders cannot interact with operational personnel while seated in their offices.

LESSONS LEARNED 353 Another “Learning Event Report” ex ample is shown in Figure 16.5. Figure 16.5 Learning Event Report Example

2.1 Establish and Imperative for Process Safety |25 The 10 core principles have som e overlap. Readers m ay note, for example, the sim ilarity of Core Principle 7 (Em powering individuals) and Core Principle 8 (Deferring to their expertise). Nonetheless, the activities associated with these related elements are different, and that differentiation helps provide clarity in the presentation of these Guidelines. The order of the Core Principles shows the dependency of each Core Principle on others. Ultimately, to successfully implement the later principles, a solid foundation should be built upon the earlier principles. Indeed, com pany and site leadership should m ake a conscious business comm itm ent to process safety and internalize it personally before making significant efforts in the other Principles. With these in place, leaders then have the possibility to build trust and communication, and start implementing the remaining Principles. 2.1 ESTABLISH AN IMPERATIVE FOR PROCESS SAFETY Illiopolis, Illinois, USA, April 23, 2004 An explosion and fire of vinyl chloride monomer killed five workers and severely injured three at a polyvinyl chloride (PVC) m anufacturing facility (Ref 2.2). A worker overrode the interlock to prevent opening the bottom valve on a pressurized reactor. As a result, hot vinyl chloride m onomer spewed into the building, ignited, and exploded. The explosion destroyed most of the plant. Smoke from the fire drifted over the local community. As a precaution, local authorities evacuated the community for two days. To override the interlock, the worker used a dedicated, labeled “Em ergency Air Hose.” This hose had a specific emergency use, requiring authorization from a senior manager designated to approve such variances. However, the plant had

130 | 4 Applying the Core Pr inciples of Process Safety Culture health and safety of their fam ilies, particularly their children. Value of their property and possessions. The public worries about incidents and environmental im pacts that could dam age their property. They are also concerned about the potential influence of the facility on their property values. Environmental protection. Most people regard themselves as pro-environment in some way. This m ay take many forms, from sim ply appreciating nature to actively protesting. They are concerned about “What YOUR plant is doing to OUR environment.” Quality-of-life. This core value encom passes three objectives: 1. Pride in Community, including the aesthetics of their neighborhood and nearby businesses, 2. Absence of Conflict. People do not enjoy fighting over issues such as chem ical releases (including nuisance odors), frequent truck traffic, or rail crossing delays, and 3. Freedom from Fear, the absence of constant concern about what events might occur in the middle of the night or while their children are in school. Economic security. B eyond the value of property and possessions, the public is concerned with how the facility affects the overall economic condition of the comm unity. This can include em ployment of family members and friends, contributions to local com munity organizations, and producing products or raw materials needed by other local businesses. Anything that happened to the facility could potentially im pact com m unity economic security. Peer pressure conflicts. If a friend or neighbor feels that the facility threatens a value important to them, they m ay start som e type of com munity action, such as a petition against the facility. This can then lead to additional discom fort in • • • • •

1 • Introduction 15 Part III of this guideline covers other items that should be considered when effectively manag ing the risks associated with transition times. Chapter 9 covers the transition times related to the facility life cycle, such initial sta rt-ups and decommissioning—the steps essential for a final shut-down. The last chapter, Chapter 10, provides a brief overview of the CCPS Ri sk Based Process Safety (RBPS) approach, providing guidance on ho w the RBPS elements apply during transition times.

EVIDEN CE IDEN TIFICATION , COLLECTION & M ANAGEM ENT 165 Examples of such needs include: • Microscopic analytical views • Magnetic Particle Inspection • X-rays • Infrared • Complex sequences • Extremely close-up views of machinery or equipment • Nighttime shots • Drone video and still photography It is obviously desirable to photograph objects of interest before they are disturbed in any way. This includes moving, turning over, or even lifting to tag or affix an identification number. A thorough and up-to-date log of all photographs is invaluable. Whenever possible, identify the data as part of the photograph itself. Data preservation concepts can and should be included in the initial an d periodic refresher training given to personnel involved in incident investigation. Photographic equipment containing electrical components should be intrinsically safe if used in any location with potentially flammable concentrations of vapors. Plant safety procedures will frequently dictate the atmo spheric monitoring requirements and types of equipment that can be used. Cameras fo r use in electrically classified areas are available, although these still need to be used within the site safety regulations. Digital cameras are standard tools for investigations. They are relatively simple to use, inexpensive, reliable, and can perform most tasks needed by the incident investigation team. Digital SLR (single lens reflex) cameras with good close-up capabilities may be needed for specialized documentation, such as fracture surfaces. Compact digital cameras are available which are rugged and weatherproof, with built-in flash and automatic focus and settings. These smaller cameras are more easily carried and suitable for general documentation and ma ny macro photography needs. For incident investigation documentation, a camera with a resolution of at least 5 mega pixels is recommended and resolution of 10 to 20 mega pixels is suggested to allow for enlargements with out significant loss of clarity. Ideally, the camera will al so have the capability for extreme close-ups and a zoom capability for pictures of distant objects. Although digital photography has many advantages for most investigations, digital photographs may be challenged as admissible in court proceedings.

110 Human Factors Handbook • Response is not defined – it is not clear how the operator should respond. • Alarm flooding – too many alarms or too many alarms in quick succession going off can make it difficult to determine the underlying issue. • Nuisance alarms – alarms that sound, but that do not require a response, due to poor calibration of sensors, can cause confusion. In some cases, operators actually respond to an alarm in the field believing that it is a nuisance alarm from past experience and are not prepared or ready to manage the situation. • High number of shelved alarms – alarms may be silenced even though they have a purpose. This commonly occurs with nuisance alarms or with standing alarms, where alarms remain in an active state for a long period of time (usually due to malfunction or poor design of the alarm system or processes). Alarms should be designed to accommodate the limitations of users and should apply good Human Factors design principles. Further guidance on the management of alarm systems and alarm design is provided in Publication 191 by The Engineering Equipment and Materials Users Association (EEMUA) – Alarm systems: Guide to design, management and procurement EEMUA [47]. This guidance is very comprehensive and explains the overarching philosophy of an alarm system, setting out what alarms systems are and how they should function. The guidance provides key principles of alarm system design, advice on measuring alarm performance, and how to make improvements. It also includes an extensi ve appendix of advice and tools to tackle all aspects of alarm system management. EMMUA 191 sets out a very simple performa nce metric to help assess alarm system performance that shows how this can impact on operator ability to respond. These are as follows: • Average alarm rate in steady operation = less than one alarm annunciating per 10 minutes. • Total number of alarms annunciating in the 10 minutes after a plant upset = under 10 alarms. • Average number of standing alarms = under 10 at any one time. • Average number of shelved alarms = under 30 at any one time. The EMMUA 191 guidance goes on to su ggest that if these benchmarks were achieved, operators would find alarm sy stems more manageable. A summary of the key principles of good alarm design is shown in Figure 9-8.

224 PROCESS SAFETY FOR ENGINEERS: AN INTRODUCTION Figure 11.37. Comparison of HF Unit incident scene pre-and post-incident (CSB 2019) Example 3. Piping may be vulnerable to impact du e to its location. Section 8.3 described an incident involving a piece of equipment that was dropped on an HF Alkylation storage tank t r a n s f e r l i n e . A n o t h e r f o r m o f i m p a c t i s v e hicular impact. Because piping is virtually everywhere in a facility, it is frequently loca ted adjacent to roadways. Protecting piping at

EVIDEN CE IDEN TIFICATION , COLLECTION & M ANAGEM ENT 173 Figure 8.8 presents some tips on timeline development. Figure 8.8 Timeline Tips Determining Conditions at the Time of Failure Conditions are included on the timeline. Determining conditions at the time of the failure is an activity bridging the gap between evidence gathering and root cause determination. Failures rarely occur without some prior indications or precursor information. However, unless someone specifically is charged with looking for it, the information is frequently overlooked in an investigation. Therefore, someone should be assigned the project of timeline development and should update it pe riodically as new information comes available. A goal of the incident investig ation team is to search back in time, find this information, and correlate it with the failure occurrence to confirm or refute a postulated failure hypothes is. This circumstantial evidence may be short-term (that is, immediately preceding the failure), or may be long-term and include anecdotal information from earlier fa ilures or from previous operating experience . It should also include post-incident occurrences that may have affected emergency response, mitigation actions, or secondary damage. The information that is gathered will be used to accurately determine conditions at the time of the incident and immediately preceding it. Analyzing evidence and determining pre-incident conditions begin as parallel efforts but converge as the investigation progresses. The incident investigation team should look specifically for evidence that provides the point of initial failure, it s progression path, and the pre-existing conditions that led to the initiation. Having an understanding of a fundamental failure mode and the sequen ce of events, the investigator then seeks evidence that indicates the actual failure mechanism. For example, the incident investigation team could analyz e to confirm material properties and

Conducting PHA Revalidation Meetings 141 on a flip chart or marker board for in-per son attendees or electronically), so all the team members have a common understanding. Table 7-1 Sample Kickoff Meeting Checklist The study leader should first explain the physical scope of the review. That description may be as simple as, “everything within the unit battery limits,” or as specific as enumerated pieces of equipment. If there is a difference between the minimum scope required by regulation and by company requirements, the leader may need to identify where the team will be going beyond regulatory minimums. This distinction will help team members understand why they may observe other differences, such as the MO C records, the selection of scenarios for LOPA, or the priority assigned to recommendations. A highlighted process flow diagram is a useful visual aid for this discussion, and it is something tangible for later reference. The leader should also explain the operational and analytical scope of the revalidation. In addition to normal oper ation, the leader should explain how to address the hazards during other operating modes, such as startup or maintenance. Depending on the core an alysis method, there may be specific checklists, questions, deviations, or node s applicable to other operating modes. Regardless, the leader’s primary purpose is to engage the team in identifying process hazards in any mode of operatio n within the scope of the revalidation. Item Completed Topic Introductions/Schedule Revalidation Scope Team Training Meeting Rules or Revalidation Team Charter Revalidation of PHA Core Methodology Consequences of Interest Previous Incidents Complementary Analyses (checklists to be used) Risk Tolerance/Risk Matrix Supplemental Risk Assessment Tools (e.g., LOPA) Recommendations from the Prior PHA Unit Tour Other Topics for Review and Questions

Application of Control Architectures 277 Second, in the schematic Figure 14.10(b) the FB signal can be used as a set point for the controller on the FF loop. In this arrangement, the FB loop determines the set point for the FF loop. This arrangement is very similar to cascade control loops. One thing to remember is that the FF loop is always the main “driver” of the system in any type of FF + FB combination. A very good example of an FF + FB con trol system is a GPS in a car when you are fairly familiar with the route to get to your destination. You set the GPS with the address of your destination, and off you go. This is the FF loop, with the GPS as the controller. If you make a wrong turn along the way, the FB loop will inform you (and the GPS), which will then use your new position as a set point and recalcu-late your route to get back on track to your destination. Let’s look at an example in a neutralization tank.The following schematics show various control mech- anisms for the neutralization of water in a tank. The first schematic, Figure 14.11, shows a classic FB control loop, with the sensor located on the resultant stream and a sig-nal to a control valve on the line from an acid/base tank to control the desired pH in the water tank.The second schematic, Figure  14.12, shows an FF control loop for the same operation. There is no differ - ence in the position of the control valve  –  it is still located on the pipe from the acid/base tank. However, the sensor element is situated on the inlet stream to the water tank, and sends a signal to the controller. The controller will calculate an adjustment based on a mathematical formula, f (x), in this case a titration curve generated in the laboratory. It will then send a signal to the control valve. Since we can’t always rely on the accuracy of the FF system, it is better to use a combination of FF and FB control. This is shown in the final schematic in Figure 14.13. In some cases, you will find FF + FB in combination with cascade control, as seen in Figure 14.13. This concept is not very difficult to understand: “if a parameter is so important that it deserves FF control, then it deserves cascade control too!” The schematic below shows the bottom of a distillation tower, where we need to control temperature. The tem-perature controller adjusts a control valve on the steam pipe. However, because temperature control is very slow, SPTC TC/uni03A3+ – STMSPSP(b) (a) TCTC STMSP Figure 14.10 Differ ent types of FF + FB con trol. PNC PNTFigure 14.11 First a ttempt to control a neutralization vessel‐feedback control.

GLOSSARY xxix Standard Operating Procedure Written, step by step in structions and information necessary to operate eq uipment, compiled in one document including operat ing instructions, process descriptions, operating limit s, chemical hazards, and safety equipment requirements. Stop Work Authority A program designed to provide employees and contract workers with the responsibility and obligation to stop work when a perceived unsafe condition or behavior may result in an unwanted event. Subject Matter Expert A person who possesses a deep understanding of a particular subject. The subject in question can be anything, such as a job, function, process, piece of equipment, software solution, material, or historical information. Subject matter experts may have collected their knowledge th rough intensive levels of schooling, and/or throug h years of professional experience with the subject. Transient Operation HAZOP (TOH) A specialized HAZOP that focuses on hazards during transient operations such as commissioning startup, and shutdown. The TOH process centers on identification of required unit-specific activities (tasks) with a potential for an acute loss of containment and an in-d epth review of the procedural controls necessary for safe and successful completion of those tasks. What-If Analysis A scenario-based hazard evaluation procedure using a brainstorming approach in which typically a team that includes one or more persons familiar with the subject process as ks questions or voices concerns about what could go wrong, what consequences could ensue, and whether the existing safeguards are adequate.

86 | 6 Implementing the REAL Model knowledge base. Ideally, these case studies will be described in variety of ways that consider different learning styles (see Section 5.2.1; more details are provided in Chapter 7). 6.8 Embed and Refresh Company and site leadership must now manage the changes as implemented. Assuming the PSMS has been updated, much of the work of maintaining continuity of the change will happen via routine conduct of operations and management review activities. However, experience has shown that without regular reminders, including ongoing verification of performance, the organization will gradually forget the reason for the change. Normalization of deviance will then set in, the sense of vulnerability will diminish, and ultimately the knowledge will be forgotten. Strategies for providing plant and corporate personnel with regular reminders of key lessons learned will be presented in the following chapter. 6.9 References 6.1 API (2016). Process Safety Performance Indicators for the Refining and Petrochemical Industries. ANSI/API RP 754 2ND ED (E1). 6.2 CCPS (2007). Guidelines for Risk Based Process Safety. Hoboken, NJ: AIChE/Wiley. 6.3 CCPS (2011a). Guidelines for Auditing Process Safety Management Systems, 2nd Edition. Hoboken, NJ: AIChE/Wiley. 6.4 CCPS (2011b). Recognizing Catastrophic Incident Warning Signs in the Process Industries. Hoboken, NJ: AIChE/Wiley. 6.5 CCPS (2016). Guidelines for Integrating Management Systems and Metrics to Improve Process Safety Performance. Hoboken, NJ: AIChE/Wiley. 6.6 CCPS (2018a). Essential Practices for Building, Strengthening, and Sustaining Process Safety Culture. Hoboken, NJ: AIChE/Wiley. 6.7 CCPS (2018b). Process Safety Metrics: Guide for Selecting Leading and Lagging Indicators, Version 3.2. New York: AIChE. 6.8 CCPS (2019a). Process Safety from the Boardroom to the Frontline. Hoboken, NJ: AIChE/Wiley. 6.9 CCPS (2019b). Guidelines for Investigating Process Safety Incidents, 3rd Edition. Hoboken, NJ: AIChE/Wiley.

6.9 References | 87 6.10 CSB (2007). BP America Refinery Explosion. Report No. 2005-04-I-TX. 6.11 DSB (2014). Explosions MSPO2 Shell Moerdijk. Dutch Safety Board report. 6.12 Rae, S. (2013). A Survivor’s Experience on Piper Alpha [Video]. www.youtube.com/watch?v=1wNG3LfEg6o. [Accessed June 2020) 6.13 US Army (1942). The Doctrine of Completed Staff Work. govleaders.org/completed-staff-work.htm. (Accessed January 2020).

96 GUIDELINES FOR MANAGING ABNORMAL SITUATIONS Figure 4.1 NASA Control Room – Engine Research Building A well-designed analogue control room would provide operators with an overall “feel” for the process that was, to some extent, lost when digital control systems were first introduced. In the late 1970s and early 1980s when DCS systems were first used in the process industry, display screens were large (deep cathode-ray tubes) and expensive, and it was impractical to replicate the layout of the previous analogue control rooms. Color printers, to obtain hard copies of trend data were also expensive and some operators even used instant-print cameras to take screen pictures so that trends could be examined. The DCS systems provided major benefits, but some early systems were not well designed for the operators and failed to provide ready access to alarm screens and system overviews. It was often necess ary to page throug h several windows before the required information coul d be obtained, by which time the original window had been closed. Another feature of the DCS system with both positive and negative aspects was that users were able to add an alarm to the system at zero cost. This led to a large increase in the number of alarms and several incidents where alarm overload (flood) was a significant contributing factor.

Plant Interlocks and Alarms 355 local mode (L/R status), and another to indicate the run status. 16.12.6 Examples F igure 16.34 summarizes the different symbols that can be added to the MCC of a motor, based on an HOA switch. Table 16.8 shows examples of P&ID representation for two types of signal in PBCS and SIS. For example, in Table 16.8, you see in the BPCS that there is a command to adjust the RPM of the motor to a specific value. This command comes through an SC, or “speed control system” (some companies use the acro-nym VFD instead). When the motor performs this com-mand, it generates a signal, SI, to show what the RPM of the motor is after performing the SC command. SI stands for speed indicator. In the same Table 16.8 a SIS action can be seen too. The SIS function asks a motor to shut down through an interlock shown in a diamond. In return, the motor generates a “run status” signal to show what its running status is – running or not running – after the command. “Health” reports by motors were discussed in the pre- vious section. In Figure 16.35 you can see its examples. In Figure 16.35(a) there is a report signal for a com- mand signal of turning on or turning off of the motor. This report shows the running status of the motor. Figures 16.35(b) and (c) show report signals related to the health of the motor. Figure  16.35(b) reports total hours that the motor is running and Figure  16.35(c) is only an alarm for every little failure of the motor, or com-mon trouble alarm. Now let’s go through a complete example of electro- motor control in Figure 16.36. MMCCXA 115MMCC(b) (c)(a) XROIHOURS COMMON TROUBLEALARM115MMCCXSR RUN S TATUS 115 Figure 16.35 Repor ting by motor. MMCC HSH/O/AHSS/SS/S COMMANDRUN STATUSCOMMON TROUBLE ALARML/R STATUS XL 115 E1 231XA 115XSR 115XCR 115X, Y, MFigure 16.36 Example of electr omotor control.

HUMAN FACTORS 367 Facilitate the communications. Keeping th e team on topic will keep everyone’s attention focused and mind engaged. Ensure that everyone contributes. In some cultures, it might be necessary to encour age people to speak up. Control dominant personalities so that the entire team ca n contribute to the brainstorming. Avoid groupthink. Human Factors Engineering. Figure 16.4 show ed a model of the three-part system of people, facilities and equipment, and manage ment systems. The design of the human interfaces between these three areas is human factors engineering. Human factors engineering aims to support the human in completing a task successfully. Human factors engineering can include designing facilities, equipment, and systems so that the human can access, operate, understand, and use them effect ively. It can be seen in valves that are accessible, display screens that are easy to interpret, and procedures that are clear and concise. Two examples of human fa ctors engineering are as follows. A car is a classic example of a man-ma chine interface. Car design has been optimized over the years to improve the mechanical design, and also to improve the operability, comfort, and safety of the dr iver and passengers. For example, the car radio may be in the dashboard, but now the controls for volume and station selection are also often on the steering wheel . This makes it easier to reach, and it also allows the driver to keep their ey es on the road – which improves safety. In a process plant emergency, it may be important to quickly isolate the process flow. Emergency isolation valves (EIV) are in stalled for this purpose. The valve itself may not be sufficient during the busy and crit ical time of an emergency. The type of valve and its location may also be importan t. For example, the EIVs can be grouped in a single location outside of the hazardou s area, labeled clearly, located at grade or provided stair access (not a ladder), and, if they are large, automated to make them easier to close and minimize the time at risk for the operator. Critical Task Analysis. Simply put, critical task analysis is a human factors tool that dissects a task into individual steps, analyzes how the task is completed, what could go wrong, and what are the opportunities for improvement. Crit ical task analysis is typically conducted on those tasks with the potential for a higher risk outcome if not performed correctly. A critical task analysis will typically involve a walk-through of the part of the process plant where the task would be carried out. This allows the analysts to see the lighting, signage, and accessibility. It also allows the team to envisi on, first-hand, what the task entails much more directly than working with a paper procedure in an office. Critical task analysis can identify specific ways to improve the likelihood of human success in completing a task. This is much more he lpful than continually trying to write better procedures, but with little guidance on how to make them better. Critical task analysis can identify improvements in many of the areas in fluencing human performance discussed in this chapter such as labeling to help operators quickly identify equipment, managing lighting and noise levels to enable better sensory signals, th e appropriateness for tools such as a checklist for critical steps, and potentially improved human machine interfaces.

APPENDIX A – PHOTOGRAPHY GUIDELINES 359 their permission. 15. Consider the location of the sun and the accompanying glare, reflections, and shadow s generated during outside shots. It may be necessary to take photographs at di fferent times of day to avoid glare and shadows. Sometimes a specially timed series of photographs may be needed to document the approximate lighting conditions at the time of the incident. 16. One disadvantage of an autofocus camera is that the camera does not always focus on the desired object. If the object of interest to the photographer is in the background but another object is in the foreground, the camera may se lect and focus on the closer object instead. A familiar example is the out-of-focus picture where the camera has focused on some back ground object in the gap between two people. Most autofocus cameras are now equipped with selectable focus features to overcome this limitation, including spot focus and manual focus. A common avoidable mistake is to expect the camera to duplicate the ability of the human eye to focus in low light conditions such as dusk or heavy shade. The performance of cameras represents a compromise of several factors. These include lighting conditions, technical quality, and image resolution. The camera syst ems are designed to perform in a specific envelope. Operating near or beyond the edge of these specifications will produce correspondingly lower performance. When shooting in difficult conditions, try a variety of camera settings to find a combination that provides a good quality image. External lighting may be necessary. Side lighting is often helpful to make surface features on an object stand out, which may not be apparent with an on-board camera flash. 17. A fresh and complete spare set of batte ries is a necessity rather than a luxury. If the camera is part of a seldom used supply kit, before traveling to the site, check that fresh primary and spare batteries are available and that a memory card is installed. 18. Some type of portable background is often desirable when shooting data in the field. A light colored pastel cloth will usually give better results than black or white. 19. When documenting a wi tness statement, the photograph should be taken from as close as possible to the actual viewpoint used by the witness. 20. Backlighting can cause major proble ms, especially when using an automatic or semiautomatic exposure control camera. Backlighting is the condition where the subject of interest (in the foreground) is in relative

APPENDIX D – EXAM PLE CASE STUDY 397 Logic Tree (9 of 9)

Preparing for PHA Revalidation Meetings 119 If an Update is being performed, the resolutions of prior PHA recommendations are critical inputs to the Update. During the PHA revalidation sessions, the team will address each pr evious recommendation by updating the affected scenarios. Table 6-3 is an ex ample of how one might document prior PHA recommendations in a revalidation. Note that in this table, the Rec. No. and Resolution columns are completed during preparation, and the PHA Revalidation Team Comment should be filled out during the meetings with the team. Table 6-3 Example of Prior PHA Recommendations with Comments No. Resolution PHA Revalidation Team Comment 1 Status 1 detail The team inserted new Node 12 containing the equipment added from this recommendation. 2 Status 2 detail The team updated in safeguards in Node 1, Deviation 1.6 per completion of this recommendation. See MOC 1 for details. 3 Status 3 detail This recommendation was open and in progress. It should be completed at the next opportunity. 4 Status 4 detail This recommendation was closed shortly after the prior PHA without action. The team concluded that it should be completed and made a new recommendation in Node 3, Deviation 3.1 to address this concern. 5 Status 5 detail This recommendation was closed shortly after the prior PHA without action. The team agreed and removed it from Node 4, Deviation 4.3. MOCs for Completed PHA Recommendations Implementation of a prior PHA reco mmendation should involve an MOC with associated hazard review. Identifying the associated MOC number for each such recommendation prior to th e meetings will expedite the meeting by having this information more readily accessible to the revalidation team. If recommendations from the prio r PHA were implemented without a documented MOC, the hazards associated with the changes will need to be considered during the current revalidation meetings. The fact that any changes were implemented without an MOC might initiate a discussion between the facilitator, the revalidation team, and management prior to start of the meetings. Is a Redo warranted? Are there other issues?

222 | Appendix: Index of Publicly Evaluated Incidents NPO Association for the Study of Failure (ASF) of Japan Incident Database (Continued) (For incident reports J1–J163: see www.shippai.org/fkd/en/lisen/cat102.html) Code Investigation J65 Explosion of Acetylene Gas Accumulated in a Drum Can of Calcium Carbide On Taking Out (1991) J66 Eruption Due to a Runaway Reaction from Incorrect Charging Quantity in the Preparation of Acrylic Resin Adhesive (1991) J67 Fire During Hot Melting Work for a Valve Blocked With Hydrocarbons (1991) J68 Explosion Caused By Friction On Manufacturing an Air Bag Inflator (1991) J69 Explosion and Fire Caused By Accumulation of Methyl Hydroperoxide at a Methanol Rectification Column of a Surfactant Manufacturing Plant (1991) J70 Leakage and Fire of Gas from Lower Piping of a Heating Furnace for Start-Up at an Ammonia Manufacturing Plant (1991) J71 Explosion and Fire Caused By Insufficient Agitation of Excessive Charging Quantity at a Multi-Purpose Drug Manufacturing Reactor (1991) J72 Explosion of an Organic Peroxide Catalyst During Circulation Before Use at a Crosslinked Polyethylene Manufacturing Plant (1990) J73 Fire Occurred Due to Dispersion of Molten Nitrate Caused By High- Pressure Steam Entering a Molten Nitrate Vessel of a Phthalic Anhydride Reactor After Opening a Steam Generation Tube (1990) J74 Run-Away Reaction Occurred During Vacuum Distillation of Epichlorohydrin Waste Liquid Including Dimethyl Sulfoxide (1990) J75 Dust Explosion and Fire While Feeding Bisphenol a to a Dissolution Drum (1990) J76 Explosion in an Intermediate Tank During Turnaround Shutdown Maintenance at a Dimethylformamide Manufacturing Plant (1990) J77 Rupture of a Reactor Caused By an Abnormal Reaction Due to Lowered Cooling Ability of the Reactor On Manufacturing a Pharmaceutical Intermediate (1990) J78 Rupture of Metal Drum Cans Containing Extracted Reaction Liquid at a Manufacturing Plant of Phenolic Resin (1990) J79 Explosion and Fire of Benzoyl Peroxide (BPO) (1990) J80 Fire in an Electrical Graphitization Furnace for Carbon Fiber Production (1990)

OPERATIONAL READINESS 375 Detailed Description The chain of events started when a standby pump was taken off-line for maintenance. The relief valve had also been removed for maintenance with blind flanges isolating the pipework connections, but with far fewer than the numbe r of bolts required to hold full operating pressure. Later, a condensate pump, reinject ing hydrocarbon liquids from the gas/liquid separation process back into the oil export line, stopped in the late evening. Attempts to restart it were unsuccessful, and a decision was made to start up the standby pump as liquid levels were rising rapidly in the process vessels. If not reversed, this would have resulted in total shutdown of the platform. The night shift crew was aware that the standby pump had been taken out of service for maintenance earlier th e same day but believed that the maintenance work had yet to commence. They re-energized the pump motor, which had not been locked out, and started the pump. Within seconds a large quantity of condensate and gas began to escape from the pump discharge pressure relie f valve location, in the module above and out of sight of the pump. The condensate pumps were located at the 21 meter (68 ft) deck support frame level, below the modules. The condensate pump re lief valves were located inside the Gas Compression Module “C”, with the connecting pipework entering and exiting Module “C” through the floor. Module “C” wa s separated from Module “D” containing the control room and emergency facilities with a non-structural firewa ll consisting of 3 sheets of a composite plating with mineral wool laid between steel sheets designed to be fire and blast resistant. The fire walls between modules “C” and “B”, and “B” and “A ” were built from a single plate coated with a fireproofing insulation material. The firewa lls installed between the modules were not designed to withstand blast from within any of these modules. An explosion blew down the firewall containing the processing facility and sepa rating it from the control room. As a result, the control room was destroyed, and important emergency control was lost. Large quantities of stored oil were quickly burning out of control. Figure 17.3. Schematic of Piper Alpha platform (Cullen 1990)

APPLICATION OF PROCESS SAFETY TO WELLS 63 Permanent abandonment has several objectives: 1) provide isolation between hydrocarbon zones, 2) protect freshwater aquifers, 3) prevent migration of formation fluids through the wellbore, and 4) remove all surface equipment and, for offshore wells, cut pipe to below seabed. A process safety event occurs when any of these objectives is not met. Plugging is normally achieved by multiple barriers. A dense abandonment fluid is pumped into each isolated zone with su fficient hydrostatic h ead to exceed any formation pressure. Cement plugs are set at the bottom of the well isolating any perforations. Higher plugs isolate higher sections of the well. These plugs can be supplemented with a mechani cal plug. A surface cement plug is also set. Each region will have plugging and abandonment requirements. 4.2 WELL CONSTUCTION: RISKS AND KEY PROCESS SAFETY MEASURES 4.2.1 Overview The major process safety hazard associat ed with drilling, completion, workover, and interventions is a loss of well control. This can result in a loss of containment event where subsurface hydrocarbons have th e potential to escap e uncontrolled into the atmosphere, land, waterways, ocean, or sub-surface strata. Consequences can include fire, explosion, toxic gas exposure, pollution, and aquifer contamination affecting people, the environment, asse ts, and company reputation. The primary cause for a loss of well control is failure of or lack of adequate barriers and/or lack of well control management and well data monitoring. Formation fluids can flow into the wellbore if the hydrostatic mud pres sure is insufficient or is compromised (e.g., due to a loss of mud into the formation) to below the pore pressure. While kick events are a part of well construction, these are normally managed by closing the BOP and circulating out with higher density mud or another response. However, if not recognized, kicks can lead to a loss of well control. Other causes of loss of well control include, but are not limited to, intercepting an existing well with new drilling; casing or drill string separation; corrosion or mechanical erosion; drilling into a higher pressure zone; earthquakes; fault movement; and premature detonation of shaped charges. Other hazards associated with well co nstruction include loss of containment events from improper opera tion or well equipment failures at the surface such as in mud separation rooms, separation and treatment facilities, and pumping and compression. This can re sult in the accumulation and potential ignition of flammable gas concentrations or liquid poo ls in equipment spaces such as the drill cabins, the control rooms, and other occupied spaces. Depending on the well location, offsite persons can also be impacted. Loss of control of energy associated with well construction is also a serious issue, but the outcome is more often a personal injury event rather than a process safety event, and so is not covered here. Examples include loss of control of heavy

150 INVESTIGATING PROCESS SAFETY INCIDENTS Paperwork may be recovered from locations exposed to an explosion, fire, chemical release, fire-fighting materials, and the weather. Wet or contaminated documents sh ould be dried and/or decontaminated. Some of these documents may be partially destroyed and very fragile. Commercial services are available to facilitate document drying and preservation. As part of the investigation process, there is often a need to collect a vast amount of documentation. It may be necessary to dedicate one full time person to execute and manage the documentatio n associated with the investigation, to free up the team members for othe r investigation activities. This individual would be responsible for a document control and chain of custody procedure for all documents th at enter or leave the site of the incident investigation. NFPA 921 provides guidance on chain of custody (NFPA 921, 2017). Maintaining accurate records of the documents distributed to outside agencies during the investigation is essential when legal or regulatory issu es are involved. Examples of specific paper data reso urces that may be useful during an investigation are shown in Table 8.2Table 8.4.

Appendix B – Relationship Between Book Content and Typical Engineering Courses This book is intended to support both the te aching of a process safety course and as a materials resource for the inclusion of process sa fety topics in typical engineering courses. To support the later, this matrix relates the chapters in this book with typical engineering courses. Table B.1. Typical engineering course relationship with book contents

SUSTAINING PROCESS SAFETY PERFORMANCE 453 Figure 22.4. Crude oil price versus upstream losses by year (Marsh 2016) Metrics will be required from a corporate leve l; however, they may not be focused on the problems at an individual facility. At a facility level, consider what problem warrants solving. This may be indicated through, fo r example, incident trends or production data. Then consider what leading metrics could be created relative to this problem. For example, production data could indicate that production levels are be ing reduced because pressure relief valves are relieving frequently which diverts product to th e flare. Leading metrics could be created to track relief valve lifts and high operating pressu re limit excursions. Lagging metrics could track the number of relief valve lifts. Through the atte ntion that these metrics focus, it might be identified that the alarms and safety instrumented systems are set too close to the relief valve set pressure giving insufficient time for operat ors or the instrumented systems to respond. This could lead to an action to reset set pre ssures on the alarms and instrumented systems. Figure 22.4 shows the relationship between oil price and the value of losses in the upstream hydrocarbon industry. Historically, as the oil price declines, the resources allocated to maintenance and training are reduced. The figure shows the correlation between these reductions and increased process safety incidents. The Guidelines for Risk Based Process Safety chapter 20 provides many examples of metrics related to sustaining process safety performance. (CCPS 2007) Auditing The purpose of process safety auditing is to identify management system and performance gaps in the process safety management system and allow correction of those gaps before an incident occurs. Audit - A systematic, independent review to verify conformance with prescribed standards of care using a well-defined review process to ensure consistency and to allow the auditor to reach defensible conclusions. (CCPS Glossary) Auditing employs a well-defined review proce ss to ensure consistency and to allow the auditor to reach defensible conclusions. An au dit involves a methodical, typically team-based,

Appendix 213 Table A.2 1Phases for the tran sient operating modes.

36 Human Factors Handbook 4.2.2 Supporting attention – where to find more information Successful selective attention can be supported by methods such as: • Minimize distractions. • Minimize late information. • Training to recognize deviations and drift. • Developing cognitive skills to be aware of personal tendencies and drift. • Training staff to focus on relevant information (Chapter 13). • Avoiding alarm overload by alarm prioritization. • Developing psychological skills (Chapters 21 and 22). 4.3 Vigilance 4.3.1 Vigilance and performance Vigilance is the ability to keep watch for possible danger. An example of vigilance is monitoring the level in a storage tank while it is being filled, to ensure it is not overfilled or monitoring process control screens looking for a spike in temperature or pressure. Other examples include confined space attendants and fire watchers. In the absence of stimulation, attention may be limited to tens of minutes. An absence of stimulation would include a long period of time where someone has to pay attention (remain vigilant) without performing any actions. The experience of the mind wandering, where vigilance begins to decrease, is known as a “vigilance decrement”. Attentiveness, the ability to fully pay attention, is likely to decline the longer a person is required to be vigilant. A typical vigilance decrement is shown in Figure 4-1 (adapted from [22]). (adapted from [22] Vigilance can decline within 15 minutes, especially in an unstimulating and uneventful work environment. Figure 4-1: Typical vigilance decrement

Pumps and Compressors 171 One main point of the above discussion is that specify - ing a pump as a “pump with 100 m3 h−1 and a differential pressure of 200 kPa” is not ideal. Each pump can work over a wide range of operating points, but there is a small operating window in which they work best from a tech-nical and/or economical standpoint. The other aspect of this concept is the example below. If you have a compressor with a capacity of 100 m 3 h−1 operating in a system that requires a differential pressure of 200 kPa, and you remove this compressor from the current system and try to reuse it in another part of the plant with roughly the same flow rate, the new differen-tial pressure of the compressor in the new position could be different from than in the old position! To explain this another way, each fluid mover operates not just at one point, but on a curve, which is called the “pump or compressor operating curve. ” This fact shows that there is a need for a control sys - tem to bring the operating point of the pump to the best point on its curve. Without any control system, a fluid mover “runs over the curve” away from the high effi-ciency point and without any limitation, and this is not good operation. This concept is shown in Table 10.4.What happens at the endpoints of the fluid mover operating curve? These points are explained in Table 10.5. There is one point regarding the unit for differential pressure of dynamic fluid movers. Dynamic fluid movers transfer fluids by throwing out packets of fluid and this type of fluid transfer gives them a specific feature: dynamic fluid movers generate a specific discharge pres - sure irrespective of the density of the fluid. For example, if a dynamic fluid mover can throw water a distance of 2 m, then it can do the same with liquid mercury, which is a heavy liquid. Because of this, instead of reporting dif - ferential pressure (in psi, kPa or bar) for centrifugal pumps, it is more common to report the injected energy to the fluid not in pressure unit as DP but in length unit as head. 10.5 Fluid Mover Identifiers Based on the concepts stated in Chapter 4, the identifiers of fluid movers are symbol, tag, and call‐out. 10.5.1 Fluid M over Symbol There are plenty of different symbols for various types of fluid movers in different companies. Table  10.6 shows some of them. 10.5.2 Fluid Mover Tag The necessity of putting fluid mover tags on the body of the P&ID is mentioned in the project documents. If fluid movers tags need to be shown on the body of P&IDs they are generally placed below the fluid movers. Table 10.4 Demonstr ation of the pump/compressor operating curve in differing types of fluid movers. Dynamic type Positive displacement type Non acceptable BEPOperatingwindow Best ef ficiency DP Non acceptable Best efficiency flowrateAcceptable withacceptance of decreased efficiency Flowrate (gpm)Differential pressure (psi) Non acceptable Rated DPExplosion point Operatingwindow Rated flow rate Flow rate (gpm)AcceptableDifferential pressure (psi) Summary: the operating curve of a dynamic fluid mover is “dropping and concave. ”Summary: the operating curve of a PD fluid mover is fairly vertical.

EVIDEN CE IDEN TIFICATION , COLLECTION & M ANAGEM ENT 155 Figure 8.3 As-found Position of Valves—Example Photo Taking photographs as soon as po ssible after the occurrence helps to document the “original” condition of the equipment and site right after the incident, before post-event response ac tivities such as site clean-up and demolition activities potentially alter the data. Document the position of all witne sses (including injured personnel), immediately before, at the time of, and immediately after the occurrence, with special attention given to determining the direction they were facing at the time they first became aware of th e occurrence and what first drew their attention to the event. The investigators should attempt to determine and/or confirm what each witness could or could not see from their respective positions thro ughout the occurrence. The locations of marks such as scratches, dents, paint smears, and skid marks that could possibly be associ ated with the inci dent should be identified and documented. It is important to determ ine if such marks were made before, at the time of, or after the incident as part of the emergency response or clean-up. Stains or discoloration can be the result of numerous causes, including heat exposure, overflow, releas e of material from adjacent equipment, or some internal occurrence. Again, it is important to determine when the stain

310 | Appendix E Process Safety Culture Case Histories E.25 Post-M OCs A large facility with m ultiple units processed approximately 100 MOCs each m onth. The PSMS Manager for the facility ran the MOC program in addition to being directly responsible for several PSM S elements and being deeply involved in the remaining elem ents. An audit revealed that a several MOCs had been approved after the physical change had been made. During interviews, with the PSMS Manager and others were not m uch concerned about this and it apparently had been the norm for years. The prevailing belief was that MOC was satisfactory if the documentation was com plete. How can a facility cope with a large flow of MOCs and still treat each one with the appropriate sense of vulnerability ? E.26 M ergers & Acquisitions A large chem ical facility was in the process of being sold to a com petitor. The acquiring com pany was in the process of a due diligence review of the organization’s operations, including a thorough review of the status of EHS programs. The acquisition was being closely m onitored by the local community, labor unions, political leaders, and the media because of the long history of operations by the facility and the m any jobs that were at stake if the acquiring com pany decided to withdraw from the deal. A regular audit that had been scheduled came due just as the negotiations and due diligence process began. There were recomm endations to postpone the audit but there were regulatory implications of doing that so the audit was conducted as scheduled. The PSM S was found to be in fairly good shape, but the auditors did discover a few im portant findings. One PHA revalidation was several months overdue, several PHA and incident investigation recommendations were Actual Case History Actual Case History

E.6 KPIs That Always Satisfy |293 Separate ITPM monitoring systems were also maintained for vibration monitoring, electric power distribution equipment, and equipment required for the emergency response plan, and in all systems, m any im portant ITPM tasks tracked by this system were found to be either overdue, m issing from the system, or both. The Plant Manager was surprised and upset when these findings were presented at the audit’s daily debriefing. When the ITPM KPI was updated to include all the m issing data, the perform ance was much poorer. More importantly, m uch work and expense were needed to catch up. Failing to include the data from the other sources was found to be an innocent m istake. However, why was the definition of the KPI not reviewed for com pleteness? Why were positive results not challenged to ensure they reflected reality? Combat the Normalization of Deviance, Understand and Act Upon Hazards/Risks. E.6 KPIs That Always Satisfy A facility tracks an overdue ITPM m etric monthly. The data is reported to a corporate process safety m etrics program , and the KPI is analyzed and published for everyone in the com pany to see. The values for all facilities, since the metrics program was established three years ago have been consistently above 99% completed on time, which the com pany was proud about result. The facility had just undergone a m ajor turnaround that had been planned to be 3 weeks but had been shortened by 5 days due to production pressures. The month following the end of the turnaround, the overdue ITPM KPI still showed 99.6 % ITPM com pletion. Upon closer review it was discovered that 75 ITPM tasks scheduled for the turnaround had not been performed due to the shorter time. This included many proof tests of SIS and B PCS functions. B ased on Real Situations

2.3 References | 25 2.40 OSHA (2005). Grain Handling. US Occupational Safety and Health Administration. www.osha.gov/SLTC/grainhandling (Accessed March 2020) 2.41 OSHA (2020) Grain Elevator Explosion Chart. www.osha.gov/SLTC/ grainhandling/ explosionchart.html (accessed March 2020). 2.42 ZEMA (2020). Infosis ZEMA. www.infosis.uba.de/index.php/en/site/ 13947/zema (accessed April 2020).

APPLICATION OF PROCESS SAFETY TO OFFSHORE PRODUCTION 115 can escalate to adjacent modules (horizontally or vertically) if rapid isolation is not possible or if firewalls or blast walls ca nnot withstand the load. A short primer on the potential for flame accel eration and blast impacts to occur was provided in Section 5.3.2. Key Process Safety Measure(s) Hazard Identification and Risk Analysis : During engineering design, HIRA identifies prevention and mitigation measures to manage various loss of containment scenarios. Prevention controls normally relate to good operations, and maintenance and inspection activities. Mitigation measures include ventilation systems to dilute or extract smaller leaks, ignition controls, fire and gas detection system, emergency shutdown system, a depressurization and blowdown system to de-inventory the affected area, and a drainage system. Fire and blast walls, if installed, mitigate the potential for escalation. A backup battery power supply system provides power for some period of time to the control room and key facilities if power is lost. The firewater system usually has at least one diesel powered fire pump that operates without electrical power. The complex altern atives and rapidity of event progression tend to encourage at least some automated response systems. The need for some or all these barriers would be determined by Hazard Identification and Risk Analysis . Emergency Management : If ignition occurs, the active and passive firefighting system comes into effect, and personnel follow evacuation, escape and rescue procedures (see Section 6.3. 5) to reach a safe refuge or evacuate the facility. 6.2.4 Oil Storage Tanks Risks Most offshore facilities export oil by pipeline once it passes through the separation system. However, some designs include local oil storage, such as with FPSOs, which store oil in large tanks in the body of the vessel prior to periodic offloading by tanker. The risks relate to spills of large volumes of oil from a tank if it is punctured due to a collision or major pipe rupture, and subsequent fire on the sea surface or environmental pollution. Additional storage of hydrocarbons offshore may include helicopter fuel and diesel fuel for power generators. Leaks fr om these tanks can cause a process safety event. Key Process Safety Measure(s) Compliance with Standards : FPSOs are ship-shaped facilities and are covered by marine classification requirements (e.g., from ABS, DNV GL, or Lloyd’s Register) if they are capable of self-propulsion, even if they are permanently moored. This is an International Maritime Organization rule, enforced by maritime regulators globally. Classification rules have detailed requirements for safe storage of oil in onboard tanks and environmental protection, like those for oil tankers. The rules are prescriptive and focus on design requiremen ts. Periodic surveys are required to verify that the facilities remain fit for duty.

Chapter No.: 1 Title Name: <TITLENAME> c16.indd Comp. by: <USER> Date: 25 Feb 2019 Time: 12:31:49 PM Stage: <STAGE> WorkFlow: <WORKFLOW> Page Number: 333 333 Piping and Instrumentation Diagram Development, First Edition. Moe Toghraei. © 2019 John Wiley & Sons, Inc. Published 2019 by John Wiley & Sons, Inc.Companion website: www.wiley.com/go/Toghraei_PID 16.1 Introduction In this chapter we cover SISs, alarm systems, discrete control, and electric motor control. SISs and alarm systems are two layers of steering processes that come after BPCS. Discrete control is a type of BPCS control, but it is covered here because its actions are similar to SIS actions. Electric motor control, including through BPCS and SIS, will be discussed as an important example later. 16.2 Safety Strategies It is mentioned in Chapter  13 that “taming” process parameters is mainly done through the concepts of BPCSs and SISs. When the control of a parameter passes from the hands of the BPCS to the SIS, an alarm is raised to warn the operators. As mentioned in Chapter 12, there are four method- ologies to cope with safety issues: inherent design, pas - sive action, active action, and procedural action. A SIS is set of active actions implemented in process plants. A SIS is a highly regulated component of process control because of its effective action of mitigating safety issues. In addition to codes generated by regulatory bod-ies there are well‐known standards regarding SISs that can be agreed upon to be followed in a process plant. Instrumentation and control practitioners are the pro-fessionals looking after such issues and in this chapter there is no intent to provide a complete approach to SISs. 16.3 Concept of a SIS The concept of a SIS is shown in Figure 16.1. The impact could because of operator error, the pro- cess went out of control, equipment failure, power loss or other things.After the impact, one or more process parameters go beyond the safe operating range. In the last step one or more functions of SIS are trigger ed to bring the process back into its dedicated “playground. ” A SIS is basically a set of SIFs (safety instrumented functions) while a PBCS is a set of control loops. Essentially, a SIS does whatever an operator is su pposed to do when an alarm sounds. 16.4 SIS Actions and SIS Types SIS actions are mainly two types: action on switching valves (on or off) and/or actions on rotary machines (start‐up or shutdown). A SIS triggers one or more SIFs to mitigate the impact. If there is more than one SIF to be triggered they could be “connected” to each other by logical operators like “ AND” or “OR” or other more complicated logic. A SIS action band is between a BPCS band and a mechanical relief band. Thus the first objective of a SIS can be defined as “bringing a stubborn parameter back to the BPCS band. ” The other point here is that triggering a mechanical relief system is not a good thing. It is because conse-quences of each time functioning of a mechanical relief displacing the process fluid from inside of process units to other part of “systems” which is not a good action. If the mechanical relief system is the overflow pipe of a tank, when it works, the liquid comes to the outside area, which is not a pleasant event and could be danger ous, depending on the type and temperature of the liquid. When a mechanical relief system is a pressure safety valve (PSV), it may release to the atmosphere or an emergency release system like a flare. It could be said these systems are designed for such emergency release 16 Plant Interlocks and Alarms

377 specific facility. Furthe rmore, NJDEP suggests that IS analysis is simply good business practice for any facilit y storing or utilizing extraordinarily hazardous materials from an econom ic, worker safety and regulatory compliance standpoint. (Ref 14.12 So ndermeyer). Experience with IS under the Prescriptive Order served as justification for expanding the IS program for all TCPA facilities. In 2010 NJDEP published a summary of the IS reviews performed by TCPA registrants in 2008 and 2009 after the requirement for IS reviews was added to the TCPA beyond the Pr escriptive Order IS reviews. The results of these IS reviews are show n in Table 14.2. (Ref 14.14 NJ IS Summary) Learnings from New Jersey . The State of New Jersey has viewed IS as an option in their security, as well as th eir process safety regulatory arsenal, believing that a facility with less hazardous chemicals will be less attractive to terrorists and, if atta cked, have a reduced probability of a serious accidental release an d/or reduced consequences. One way in which the TCPA has driven the concept of inherent safety has been by inspiring companies to seek ways to eliminate or reduce the amount of EHSs (i.e., the covered ch emicals) handled to a level below the threshold quantity to which the regula tions apply. In the almost 30 years since the TCPA was enacted, the numbe r of TCPA-regulated facilities, as well as the quantity of EHSs regist ered in the state, have dropped significantly. This has had the benefit of reducing risk as well as avoiding regulatory requirements. It should be noted, however, that many TCPA covered facilities voluntarily reduce d their EHS inventories before New Jersey issued the Prescriptive Order or adopted the IS provisions in the TCPA.

7 • Unscheduled Shutdowns 136 Relevant RBPS Elements Process Safety Culture Process Knowledge Management Hazards Identification and Risk Analysis 7.6.4.2 Meteorological event incidents C7.6.4.2 -1 – Hurricane Georges flooding incident [36, p. 19] Incide nt Year : 1998 Cause of the facility shut -down : Hurricane Georges flooded a refinery on the shore of the Gulf of Mexico. The Category 2 hurricane storm surge overtopped the dikes built to protect the refinery, leaving the entire facility submerged under more than 1.2 m (4 feet) of salt water, and due to its slow movement, subjected the refinery to 17 hours of high winds and rain. Incident impact : Salt water damage occurred to approximately 2,100 motors; 1,900 pumps; 8,000 instrument components; 280 turbines; and 200 miscellaneous machinery items; resulting in replacement or extensive rebuilding of the damaged equipment. Risk management system weaknesses: LL1) The older areas of the refinery suffered from the flooding, causing significant property damage due to the equipment’s layout. Although most of the refinery had suffered significant damage, the newer control buildings and electrical substations sustained little or no damage as they had been built with their ground floors elevated approximately 1.5 m ( 5 feet) above grade. Relevant RBPS Elements Process Knowledge Management Hazard Identification and Risk Analysis Risk management system strengths: The incident occurred in 1998. The newer buildings, electrical substations, and equipment located above grade showed how

Utilities 373 The type of compressor depends on the consumption and could be a centrifugal type air compressor or of a posi- tive displacement type. Generally, the required pressure for instrument air users is a pressure between 6 to 15 psig . We usually provide instrument air for instrument air users at a pressure around 45 psig . Because of that, the instrument/ plant air system should provide a pressure more than that to be able to overcome the resistance in a route between the instrument/plant air generator and the final instrument air user. This pressure could be a pressure around 50 psi.Make-up utility Make-up utility Used utility 2Utility 1 preparationUtility preparation UsersUsed utilityWasted used utility Users of utility 1 Users of utility 2Utility 2 preparationUsed utility 1 Figure 17.15 BFD of once‐thr ough and mating utility systems. Ambient air Ambient airAir treatment Air treatmentInstrument Air (IA) IA usersEnvironment Environment PA usersPlant Air (PA) Figure 17.16 Instrumen t air and plant air route. Air blowerCooler Air receiverCartridge filterDessicantInstrument air Cartridge filter Instrument air receiver Utility air Figure 17.17 Air cir cuit BFD.

APPLICATION OF PROCESS SAFETY TO WELLS 71 For onshore facilities where there is si multaneous drilling and production, it is common to carry out facility siting studies. This is particularly the case if the inventory of hydrocarbons exceeds the nom inated threshold limit and OSHA PSM (1910-119) applies. Facility siting studi es are not common for temporary well construction facilities with no associated permanent production. Refer to Section 5.3.2 where the topic of facility siting is discussed in detail. 4.2.10 Surface Process Equipment at Well Construction Facilities Risks There are multiple potential so urces of flammable gas or liquid releases from surface kick handling and mud process equipment at well construction sites when the drilling reaches hydrocarbon zones. Th ese are mud degassers, solids control equipment, and mud tanks and are common at all drill sites. There also can be gas and other hydrocarbon handling equipment at drill sites that conduct well flow backs or underbalance drilling. Liquids storage, if any, is usually in atmospheric storage tanks that store hydrocarbon liquids (generally mixtures of C5 and heavier). This material is normally transported by truck, especially for remote areas as well as exploration wells that do not have the export infrastructure of production facilities. However, this is generally associated with limited well flowbacks or underbalance drilling which are not common in most drilling outside of shale developments. Wells undergoing workovers or interventions ar e usually close to production facilities and thus have SIMOPS risks. Offshore FPSO and FLNG facilities also store hydrocarbon liquids. Storage risks are discussed in Chapters 5 and 6. Key Process Safety Measure(s) Compliance with Standards : Process equipment should be designed in accordance with industry standards and company practi ces with the intent to provide integrity and minimize release potential. Hazard Identification and Risk Analysis : Potential leak scenarios should be identified, and safeguards reviewed for adequacy, with potential hazard zones estimated to establish risks to personnel, other process equipment, or to the affected public. CCPS (1999) pr ovides suggested means for ho w to predict hazard zones. 4.2.11 Harsh Weather Risks Harsh weather, both such as tornadoes for onshore and storm winds and waves for offshore, can lead to loss of containment events. Some harsh weather offshore events which a reader might wish to examine in clude the following (multiple references on-line). ●Alexander Kielland, Norway 1980, 123 fatalities ●Ocean Ranger, Canada 1982, 84 fatalities

EQUIPMENT FAILURE 181 The storage tank involved was Tank 912. Tank 912 was a 6,000 m3 (1.6 million gal) floating roof tank with an automatic tank gauging (A TG) system that was monitored in the control room. Operators could operate the appropriate valves to shut off flow and/or divert it to other tanks. The tank had an alarm for high and high-high level that could be set by the supervisors. Tank 912 also had an independent high-level switch (IHLS) that would stop incoming flow at a high-high level by closing the inlet valves and pr ovide an audible and visual alarm in the control room. The tank started receiving about 550 m3/hr (145,294 gal/hr) of ga soline (containing 10% isobutane) at about 7:00 PM on Saturday evenin g. The isobutane had been recently added to make a winter blend making the gasoline more vo latile. At 3:00 AM Sunday the tank was about 2/3 full, but the level gauge stopped recording any further increase in level. The independent high-level switch (IHLS) shutdown did not work. At about 5:20 AM the tank began to overflow, but flow into the tank continued, even increasing in rate to about 890 m3/hr (235,113 gal/hr). As fuel continued to overflow from Tank 912, a dense vapor cloud up to 2 m (6.6 ft) tall and covering an area of about 500 x 350 m (1640 x 1148 ft) formed, engulfing a large portion of the facility (Figure 11.3) (HSE 2017). The final extent of vapor cloud explosion is marked in yellow. The first explosion occurred at 6:01 AM. Initially the ignition source was hard to determine, candidates include; a pump house, heaters in the emergency generator building, and car engines (witnesses stated their cars began to run erratically, (i.e. surging due to drawing in fugitive gasoline vapors). Subsequent analysis (see below) has settled on the pump house as the initial site of ignition. Further explosions occurred and the entire facility was engulfed in fire.

1 • Introduction 11 associated process and facility shutdowns, covers the normal operations mode; Chapter 3 discus ses the normal operations; Chapter 4, process shutdowns; and Chapter 5, facility shutdowns. The transient operating modes—the shut-downs and start-ups—associated with normal operations, process shutdowns, and facility shutdowns are covered in each of these chapters, as well. These modes of operation will be defined in more detail in Chapter 2. Table 1.2 Chapter framework for this guideline.

3.8 For equipment containing materials that become unstable at elevated temperature or freeze at low temperature, is it possible to use heating/cooling media which limit the maximum and minimum temperatures attainable (i.e., self -limiting electric heat tracing or hot water at atmospheric pressure)? 3.9 Can process conditions be chan ged to avoid handling flammable liquids above their flash points? 3.10 Is equipment designed to totally contain the materials that might be present inside at ambient temperature or the maximum attainable process temperature (i.e., higher maximum allowable working temperature to accommodate loss of cooling, simplified reliance on external systems like refrigeration to control temperature such that vapor pressure is less than equipment design pressure)? 3.11 For processes handling flammable materials, is it possible to design the layout to minimize the number and size of confined areas and to limit the potential for serious overpressure in the event of a loss of containment and subsequent ignition? 3.12 Can process units (for hazardous ma terials) be designed to limit the magnitude of process deviations? • Selecting pumps with maximum capacity lower than safe rate of addition for the process • For gravity-fed systems, limiting maximum feed rate to be within safe limits by pipe size or fixed orifice • Minimum flow recirculation line for pumps/compressors (with orifice to control flow) to ensure minimum flow in event of deadheading 3.13 Can hazardous material liquid spills be prevented from entering drainage system/sewer (if potentia l for fire or hazardous reaction exists, e.g., water reactive material)? 3.14 For flammable materials, can spills be directed away from the storage vessel to reduce the risk of a boiling liquid expanding vapor explosion (BLEVE) in the event of a fire? 3.15 Can passive designs, such as the following, be implemented? • Secondary containment (e.g., dikes, curbing, buildings, enclosures) 448

4.3 Process Safety Culture and Ethics |113 to process safety should make it undesirable for a m anger to ignore process safety in favor of other business areas. Should There be a Process Safety Incentive at All? Should the organization forego all process safety-related incentives because it is simply the right thing to do? Choudhry (Ref 4.3) argues that working without injury should be a strong incentive by itself, as it provides workers with the long-term term benefit of being able to provide earnings for the company and them selves and their families. However, money is a very strong hum an motivator, and if used with care can help change behavior. This decision m ay be influenced by where a com pany is on its culture improvement journey. Nonetheless, good process safety perform ance should be rewarded at the very least by a heartfelt thank you from the leadership team . Sum mary Incentives can be particularly useful in underlining the core principles establish an imperative for safety and combat the normalization of deviance . They have the potential to influence process safety performance and process safety culture, both positively and negatively. Ultimately, metrics and incentive approaches should treat process safety on par with other business priorities, discourage managers from prioritizing production over process safety, and drive the desired results and behaviors. Leaders should examine incentives schem es to m ake sure they do not drive the opposite or negative behaviors. 4.3 PROCESS SAFETY CULTURE AN D ETHICS Krause (Ref 4.4) links process safety and ethics closely: “(Process) Safety appeals to the ethical ideals that motivate a company’s best leaders at every level of responsibility.”

172 Human Factors Handbook 15.3.3 Recognizing fatigue Good fatigue risk management includes being able to recognize the signs of fatigue in colleagues. Typical symptoms of fatigue are shown in Figure 15-6, adapted from the Canadian Center for Occupational Health and Safety [63]. Team leaders, supervisors and colleagues can recognize fatigue in themselves and others, creating the opportunity to take action. Rest breaks A very short rest break of a few minutes can stop fatigue increasing. A long break can reduce fatigue to its starting level. Naps Short naps can reduce fatigue by half, especially when working night shifts. Exercise Walking around during breaks, and moderate exercise before work can increase alertness. Bright ambient lighting Bright white lighting improves alertness. Talking Talking with colleagues can boost alertness. Some tips on main taining alertness Stimulating tasks Doing more interesting work can help reduce drowsiness. Non-technical skills • Recognition of fatigue in self and in others. • Ability to ask for help and challenge others’ fatigue.

36 INVESTIGATING PROCESS SAFETY INCIDENTS experiments, or other relevant data are then used to construct or test hypotheses that purport to solve it. Emphasis on the use of a scientific approach in investigations has increa sed due to court ru lings in the United States that require experts to have a scientific basis for their opinions. NFPA 921, the Guide for Fire and Explosion Investigations , has incorporated the scientific method as the key approach for fire and explosion investigations (NFPA 921, 2017). As an overview, the scientific meth od involves developing hypotheses based on investigation data including witness accounts, observations, measurements, recorded data and analys es. Hypotheses ar e then tested to determine if the hypothesis is true or not. Multiple hypotheses are considered. The process is often iterat ive with findings from one hypothesis suggesting an alternative hypothesis. The process is complete when all hypotheses have been te sted and either proved or disproved. The final hypotheses provide the basis for iden tifying causal factors. Chapter 9 describes the scientific method in detail as it is used most extensively with evidenced analysis. The scientific method does not replac e the use of timelines or sequence diagrams. Rather, the scientific method is complementary to the use of timelines and sequence diagrams. 3.2.4 Causal Factor Identification When using a predefined tree methodology for root cause an alysis, once the evidence has been collected and a timeline or sequence diagram developed, the next phase of the investig ation involves identifying the causal factors. These causal factors ar e the occurrences and actions that made a major contribution to the incident. Causal factors can involve human errors, equipment failures, undesirable conditions , and failed barriers that led to the incident. Causal factors point to the key ar eas that need to be examined to determine what caused that factor to exist. There are a number of tools, such as Barrier Analysis (Dew, 1991; Trost, 1985) and Change Analysis (Kepner, 1976), that can assist with the identification of causal factors. The concepts of incident causation encompassed in these tools are fundam ental to most of the investigation methodologies. The simplest approach involves reviewing each unplanned, unintended, or adverse item (negative event or undesirable condition) on the timeline and asking, “W ould the incident have been prevented or mitigated if

290 PROCESS SAFETY FOR ENGINEERS: AN INTRODUCTION that direction. Thus, a site with many ignition sources on or around it would tend to prevent clouds from reaching their full hazard extent, as most such clouds would find an ignition source before this occurs. Early ignition, befo re the cloud becomes fully formed, might result in a flash fire or an explosion of smaller size. La te ignition could result in an explosion of the maximum possible effect. The main consequence of a VCE is blast over pressure.. The blast effects produced depend on whether a deflagration (flame front less than sonic velocity) or detonation (flame front greater than sonic velocity) results. (Refer to Chapter 4 for definitions.) Thermal expansion occurs as the fuel is combusted and this drives flame acceleration. Flame acceleration is influenced by congestion within the fac ility and confinement of the vapor cloud. A deflagration event requires congestion and only the flammable material in the congested space contributes to the explosion. Once the flame passes through the congested space, its velocity drops and it becomes a flash fi re event. A detonation event, once initiated, is self-sustaining and the entire flammable ma ss contributes, regardless of whether the whole cloud is in congested space or not. Important parameters in explosion analysis ar e the properties of the material: lower and upper flammable limits (LFL and UFL), flash point, autoignition temperature, heat of combustion, molecular weight, and combusti on stoichiometry. The upper and lower flammable limits are used to determine the flammable mass. Some analysts use the mass between the upper and lower flammable limits, some use the mass between half the lower flammable limit and the upper flammable limit. Usin g half of the lower limit, as opposed to the full lower limit, is conservative as it includes more material in the flammable range. The impulse, (the area under the explosion pressure-t ime curve) is necessary to determine the dynamic loading effects on a structure. The three common VCEs models are explained in greater detail in Vapor Cloud Explosion, Pressure Vessel Burst, BLEVE and Flash Fire Hazards . (CCPS 2010) They are TNT equivalency model TNO multi-energy model Baker-Strehlow-Tang model TNT Equivalency Model . The TNT equivalency model represents a VCE as a TNT detonation of equivalent energy. This was one of the first models developed; however, case studies and experimental data have shown that th e results are not representative of a VCE. It is now understood that VCE blast effects are dete rmined not only by the explosion energy, but more importantly, by the combustion rate. Ther efore, use of the TNT equivalency model is no longer recommended for vapo r cloud explosion analysis. TNO Multi-Energy Method . The multi-energy model recognizes that only the confined portion of the cloud, not the entire volume of a vapor cloud, contributes to the blast effects. This method uses blast curves plotting overpr essure . vs. distance and impulse vs. distance. The initial blast strength is represented in a series of 10 curves, representing levels of congestion. The curve used significantly impacts the results. Limited guidance is available on which curve to use although most analysts use curves 5, 6, or 7. The TNO method is based on interpretations of actual VCE incidents.

5.3 Corporate Change Models | 63 5.3.3 Kotter John P. Kotter, the Harvard professor emeritus who is sometimes described as the world’s foremost authority on leadership and change, developed a model that focuses on how leaders drive change at a high level rather than getting into the details of what to change (Kotter 2012). Kotter guides leaders to: • create a sense of urgency • build a core coalition • form a strategic vision • get everyone on board • remove obstacles and reduce friction • generate short-term wins • sustain acceleration • set the changes in stone. Related to the desired learning model, Kotter’s model tends to address boxes I (Plan) and IV (Act) of Figure 5.1. Clearly, Kotter’s model applies for companies and plants that need a major overhaul of their PSMS, standards, policies, and even organizational structure. Sometimes, however, companies need only minor changes, for which Kotter’s model would be overkill. Either way, the learning process we seek must be part of a broader top-down process safety effort. If such an effort has not yet been implemented, the Kotter model would be a highly effective way to deploy it. 5.3.4 ADKAR® The ADKAR® model developed by the change management company Prosci is designed to drive change from the bottom up (Hiatt 2006). The ADKAR® acronym stands for the first letters of: • awareness of the need to change • desire to participate and support the change • knowledge on how to change • ability to implement required skills and behaviors • reinforcement to sustain the change. Related to the desired learning model, ADKAR® tends to address boxes III and IV (Check and Act) of Figure 5.1.

250 PROCESS SAFETY FOR ENGINEERS: AN INTRODUCTION Figure 12.4. HAZOP analysis method flowchart (CCPS 2008 b) Select a process section or operating step Explain design intention of the process section or operating step Repeat for all process sections or operating steps Repeat for all process variables or tasks Select a process variable or task Apply guide word to process variable or task to develop meaningful d Repeat for all guide words Develop action items Examine consequences associated with deviation (assuming all safeguards fail) List possible causes of deviation Identify existing safeguards Assess adequacy of existing safeguards based on judgment or scenario risk

xxiv | Driving Continuous Process Safety Improvement from Investigated Incidents investigations. Readers can, and should, also use the REAL Model to enhance their learning from internal incidents and to strengthen their recommendations and ongoing communication efforts. The eight steps of the REAL Model—and where they fit in the traditional Plan-Do-Check-Act improvement cycle—are summarized in Figure FM.1. Individuals Company Gather facts 2. Seek learnings 3. Understand 4. Drilldown Do 1. Focus Plan Interpret and act 5. Internalize 6. Prepare Check 8. Embed and refresh 7. Implement Act Figure FM.1 Recalling Experiences and Applied Learning (REAL) Model The book starts by highlighting the importance of driving permanent change based on learning from incidents. It then lays the foundation for the REAL Model, which is then introduced, followed by a discussion of learning styles and how to leverage them to effectively communicate what has been learned and keep the learning fresh. The latter part of the book discusses landmark incidents and features 6 hypothetical scenarios that are based on real-world situations readers may encounter. The book concludes with a call- to-action to drive continuous improvement. A brief description of each chapter follows. • Chapter 1 explains why it is important to translate the findings from incidents into lessons learned that become a part of the corporate culture. • Chapter 2 summarizes learning opportunities that are often overlooked and lists valuable resources including databases, publications, and more. • Chapter 3 evaluates obstacles to learning and describes a general philosophy for overcoming those obstacles. • Chapter 4 provides examples of incidents where companies failed to learn from previous incidents, whether in their own companies or externally. • Chapter 5 examines literature about learning, considering both how individuals and companies learn and change. Based on this literature, the

Utilities 371 is the ground where there are concrete pads and equip- ment installed on them. The form 2 is soil ground. It could be assumed that rainwater that comes on pad areas is contaminated storm water because process equipment is installed on those paths and there is a chance of chemi-cal spillage. However the rain on the soil area could be considered as non‐contaminated storm water. The “footprint” of a surface drainage system is very small on the main P&ID. It could be limited to notes near discharge points as “to drain system” with or without a symbol representing it. The end of the system, or “sumps” could be shown on the main P&ID or auxiliary P&IDs. The collection network can be handled in different ways. In some plants they don’t show it at all on P&IDs with the logic that “they are civil engineering considerations and we don’t show them. ” In some other plants they show them on auxiliary P&IDs. Such a surface drainage collection network is shown on a P&ID as shown in Figure 17.13. Surface wastewater is generally collected through a network of tranches. However, there are cases that it is done through pipes too. The pipes could be laid down in trench or buried as underground pipes. The minimum size of pipes for surface wastewater col- lection could be considered as 2″. Figure 17.14 shows a surface drainage collection net - work with pipes. In some cases dumping the liquid on the open‐top trenches involves hazard risks. For example if the liquid is very flammable or toxic. In such cases the drain should be hard‐piped toward a closed sump system. The surface drainage is generated by two types of sources: a point source and a surface source. An exam-ple of a point source is the draining nozzle of a vessel. An example for a surface source is washing water. It is fairly easy to direct the water from a point source to the trench, it just needs to pipe the point source to the trench. For surface sources, however, the floor should be sloped toward the collection network. Sometimes there is a need for several sumps for a specific area just because sloping is not possible to be implemented in the area. In some plants, there is need for a separate surface drainage collection system and separate sumps named Sewer water To Sump Catch basin Manhole Clean out Figure 17.13 Surfac e drainage collection network. FD-10FD-11FD-13 FD-12 Figure 17.14 Piping net work as a surface drainage collection system.

TOOLS AND METHODS FOR MANAGING ABNORMAL SITUATIONS 153 Example Incident 5.5 – Caribbean Petroleum Refining Tank Farm Explosion and Fire – ( cont.) Lessons learned in relation to abnormal situation management: Management of Change: No MOC was conducted to manage the loss of a critical safety barrier when the level gauge on the tank stopped functioning. As a result, they did not consider the aspect of human performance issues in managing the level in the tanks. Process control monitoring: The lack of the level measurement in the control room placed the control room operators in a position of operating blindly. Abnormal situation management recognition: Conducting the gasoline transfer under these co nditions was not recognized as an abnormal practice. This resulted in an underestimation of the potential consequences of an overflow. 5.9.2 Management of Organizational Change For an organization to perform consis tently at a high level, the roles within the organization must be well defined and communicated. For example, the responsibili ties of the control panel operator versus the field operator must be distinc t, documented, communicated, and validated. Their roles must not conflict, but rather complement one another. Once the organization ha s been well defined and the roles are clearly established, all changes to th e organization should be carefully considered and reviewed. CCPS has published a helpful resource entitled Guidelines for Managing Process S afety Risks During Organizational Change , (CCPS 2013). The book addresses many aspects of organization changes, from staffing to hierarch y changes. It also includes many examples that illustrate where an orga nizational change was one of the contributing factors that led to an incident. Additionally, the CCPS book provides several checklists and activity mapping forms that can be applied when organizational changes are being considered. These checklists and forms are tools that can be appropriately applied when revi ewing organization changes.

Application of Control Architectures 281 streams” in a process plant, except parallel streams to identical parallel process units. You may ask: “why would one install more than one sensor for one specific param-eter?” The short answer is unpredictability. If, for what - ever reason, you are not sure where on a stream you can find your “representative” value of the parameter of interest, you may use selective control. Let’s look at an example of a reactor, shown in Figure 14.19. The feed stream comes into the reactor from the top and the product leaves from the bottom. The operation involves an exothermic reaction and the reactor is jack - eted so that we can cool it down. The coolant enters the reactor at the bottom and leaves at the top. Because we are unable to predict where in the reactor the temperature will be very high, we place a number of temperature elements at various places on the reactor with temperature transmitters that send signals to a high selector, TY. The selector then selects the highest temperature (worst case), and sends a signal to the tem-perature controller, which in turn adjusts a control valve on the coolant feed line. Figure 14.20 shows another example of selective con- trol on two operating centrifugal pumps. Here we need to protect the centrifugal pumps against low flows, i.e. flows that are less than the “minimum flow” reported by the pump manufacturer. With centrifugal pumps, if the flow drops below the minimum specified by the manu-facturer, the pump will start to vibrate.In this case, we have two pumps operating in parallel (This is not a case of one pump being on standby; they both operate at the same time). In order to protect the pumps from minimum flow, sensors from both pump discharge lines send signals to a low selector, FY. This in turn sends a signal to the flow controller, FC, which adjusts a control valve on the recirculation line to ensure that the flow into the pumps stays above the minimum. 14.8.3 Ov erride and Limit Control Override and limit are two different types of control; however, both of them use high and/or low functions to operate. Therefore, it is a good idea to compare them side‐by‐side before we look at each one in more depth. First of all, we need to discuss the difference between selective control (discussed previously) and override control. Override control uses the same operators as selective control: a high selecting (>) or a low selecting (<) operator. However, the main difference between them is that selective control is part of the normal opera-tion of the control system acting on primary signals from sensors, whereas override and limit control only kick in on controller signals when the process drifts outside of its normal band of operation. At the schematic level, in selective control the opera- tors “sit” on top of the sensor signal while in override control, the operators sit on top of the controller output. Now it is time to talk about the difference between override control and limit control. Table 14.6 highlights the differences between override and limit control. TT TY TC Coolant> ProductTTTTFeed Figure 14.19 Selec tive control.FEFC FY< FE Figure 14.20 Selec tive control example.

21. Fostering situation awareness and agile thinking 271 Table 21-4 provides symptoms of observable behavior demonstrating group- think. Group Think can be very common in risk assessments and understanding what is happening in a process upset or incident, where individual(s) have strong preferences toward the likelihood of certain scenarios. With limited data or information to support otherwise, it ca n be very easy to succumb to the peer pressure of the group and agree with the consensus view. This highlights the importance of having good information and decision-making methods that remove subjectivity of the group and skewed opinions. By understanding group-think and recognizing the symptoms (Table 21-4), group-think can be avoided or mitigated.

INCIDENT IN VESTIGATION TEAM 101 Other participants can be involved in a full- or part-time consulting role, depending on the nature of the incident. It is impo rtant to include people who know what happens in the field—not just those who know what is supposed to happen. The team select ion should involve the appropriate competencies and roles to be credibl e with other stak eholders such as employees, departments, union representatives, community groups, regulatory agencies and legal departments. Positions to consider should be ba sed on the nature and scale of the incident and may include: • Emergency response perso nnel such as fire chief • Fire investigator—for expertise to help determine fire origin and cause • Explosion investigator—for expertise in understanding the ignition source and physics involving explosion • Process control (electrical/instru mentation) engineer / designer • Computer software specialist • Data recovery/ forensic data specialist • Instrument technicians, inspection technicians, and maintenance technicians • Maintenance engineer • Civil or structural engineer • Construction department • Contractor participant • Purchasing or st ores department • Original Equipment Manufacturer (O EM) representative—a factory or team services engineer • Materials/ corrosion /metallurgis t / failure analysis engineer • Rotating equipment specialist • Industrial hygienist • Environmental scientist or specialist • Chemist/ specialist testing lab services • Quality assurance specialist • Research technical personnel • Human factors specialist • Other technical consultant or equipment specialist • Human Resources representative • Recently retired employee with pertinent knowledge, skill, or experience • Collective bargaining unit participant

Initiating causes for a pressure surge leading to a release include: loss of flow in the circulation loop upset in the ratio control loss of coolant to the heat exchanger high temperature of coolant in the heat exchanger upsets in the ratio control During plant operation, several pressure surges from these causes opened the relief valve. The result ing cloud required evacuation of adjacent process units. The emergency response was to use a fire monitor to knock down the cloud while the unit corrected the control or heat transfer problem and waited fo r the tank pressure to come down. To prevent releases, an SIS was impl emented that isolated the make-up ammonia when there was high temperature, high pressure, improper feed ratio, high level, or loss of the circulating pump. The SIS had to be designed to have a fast response time, since temperature and pressure rose quickly during an upset. A new tank of 50 psig (65 psia, 4.4 atm absolute) design pressure was installed to make the plant less sensitive to upset. The concentration would have to reach about 34% at a temperature of 140 ºF (60 ºC) to cause a release, shown by the heavy black border in the lower right corner of Table 15.3. The new tank e ssentially eliminated releases from the relief valve. The SIS did not have to act as quickly, and, perhaps, some shutdown initiators could be eliminated. This example illustrates the fact that the cheapest equipment—even if it is free—may not always be the most cost-effective option, especially when the economics of the consequenc es of releases and the cost of SISs are considered. 407

417 Piping and Instrumentation Diagram Development, First Edition. Moe Toghraei. © 2019 John Wiley & Sons, Inc. Published 2019 by John Wiley & Sons, Inc.Companion website: www.wiley.com/go/Toghraei_PID 20 Examples

215 compromise the government’s ability to deliver essential services during an emergency. 5.Critical Relationship to Corporate or National Economy . Chemicals, materials or facilities that, if una vailable, could create significant adverse consequences for the corporation’s existence or economic well-being, or the nati onal or regional governmental economy. The first three issues (i.e., release, theft/diversion and sabotage/contamination) relate to th e properties of the chemicals and the potential for adverse human heal th effects. The last two issues– government mission and the economy–re flect the critical uses of some chemicals that may or may not be hazardous. Inherent Safety (minimization, moderation, or substitution) can affect consequences by either reducing or eliminating the hazard, or by moderating the hazard. Simplificatio n may reduce the opportunities for an event to escalate to a larger degree of consequence. Threat can be defined as any indication, circumstance, or event with the potential to cause loss of, or damage to, an asset. It is also the intention and capability of an advers ary to undertake actions that would be detrimental to valued assets. Threats are manifestations of an adversary’s malevolent intent directed at the chemical asset or use of the chemical asset as a means to at tack a different target, such as stealing a chemical for use in produc ing an improvised explosive device. The magnitude of a threat is influe nced by the intention and capability of an adversary to undertake actions that would be detrimental to valued assets. Sources of threats may be categorized as terrorists (international or domestic); current or former disgruntled employees or contractors; activists, pressure grou ps, single-issue zealots; or criminals (i.e., white collar, cyber hack er, organized, opportunists). Adversaries may be categorized as “insiders” (internal threats), “outsiders” (external threats) or a combination of both insiders and outsiders (internal-external collusi on). Government law enforcement and security agencies may provide ad ditional information on potential adversaries, with regard to motives and tactics. Companies may also choose to acquire such information through other channels, including commercial intelligence services.

108 safeguards related to pressure detection, control, and relief. Emergency relief devices, such as rupture disks or relief valves, may still be required by regulations and Recognized and Generally Accepted Good Engineering Practices (RAGAGEPs), but their size and relief capacity, as well as hazards associated with the opening of relief devices, may be reduced or eliminated. It may also be possible to eliminate catch tanks, quench systems, scrubber s, flare stacks, or other devices designed to safely discard the effluent from emergency relief systems. If external fire is a consideration, an inherently safer design would require that the design temperature of equipment be high enough to withstand the temperature and resultant pressure generated by the fire. In general, the temperatu res generated by fires will exceed the design temperatures of most materials used in the fabrication of process equipment, and therefore it is very di fficult to design process equipment to be inherently robust with respect to a fire. In this instance, fire-proof insulation could be used as passi ve safeguard protection against external fire.. However, the potent ial for corrosion-under-insulation (CUI) isis an issue to consider and address, as well as understanding that the integrity of the insula tion system is vital to the protection it can provide against fire impingement. This same concept also applies to internal fires in equipment. For example, if it is possible to structurally design a column containing reactive or pyrophoric packing materials to withstand the temperatures and/or pressures from an internal fire, then the active safeguards required to maintain an oxygen-free environment would not be as critical. In this case, it would still not be advisable to allow an internal fire in a column to develop because of the other personnel and equipment hazards associated with fires. However, if the high temperatures of a fire would not affect the basic integrity of the column itself , the severity of an internal fire would be less, and th e criticality of the safeguards would be lower. Process equipment containing liquid levels should be designed to withstand the maximum hydrostatic lo ad that could be imposed on the equipment if it were completely filled. This is particularly true of tall equipment, such as columns, towers, la rge reactors, etc. If such a design, under all feasible circumstances, e liminates the possibility of a loss of

186 | 5 Aligning Culture with PSMS Elements understanding and acting on hazards and risks . As part of accomplishing this, the MOC procedure should consider several points: Changes should not be made without M OC. Replacements-in-kind should be carefully evaluated to ensure they are not actually changes. The requester should include MOC in the project timeline. The requester should provide a complete description of the change using information from the process knowledge m anagement system . The level of MOC evaluation should be based on process risk, with higher risk processes subject to more rigorous MOC reviews. At no time should MOC be rushed or treated as a check- the-box exercise. Similarly, m anagement and personnel should avoid pressuring reviewers to approve an MOC before a sufficiently thorough evaluation has been done. Reviewers should resist such pressure and provide the appropriate open and frank communication if they feel undue pressure. The level of MOC approval should be based on process risk, with higher risk processes requiring approval by higher levels in the organization. Conflict of interest should be avoided. The persons requesting the MOC and sponsoring the change should not be approvers. All action items identified in the MOC should be closed-out and verified in the field. After the change has been implemented, the requester should update the inform ation in the process knowledge m anagement system . In recent years, m any facilities have sought to create MOC efficiencies. Electronic MOC system s (e-MOC) have becom e com mon. These can help address the document routing and document management needs and can help expedite the • • • • • • • • • • •

94 GUIDELINES FOR MANAGING ABNORMAL SITUATIONS Example Incident 4.2 - Ch ernobyl Disaster, April 1986 (cont.) In addition to the design issues , lessons learned in relation to abnormal situation management are: Abnormal Situational Awareness: The cooling system of this particular reactor system was not designed to mitigate an inherent risk of runaway core heating. The conditions that could lead to this heating situation should have been understood by operating personnel. Procedures: Written procedur es for safely managing and preventing abnormal situations involving the cooling process should be considered safety critical procedures. Knowledge and skills: Front-line personnel should be fully trained on all procedures. Simple design features can, particular ly under stressful conditions, lead to an incorrect response by an operat or. Such incidents may initially appear to be caused by “human error”, although a thorough root cause investigation often reveals underlying design or management system issues. Current Example Incident 4.3 (Syed 2015) shows this type of error.

108 GUIDELINES FOR MANAGING ABNORMAL SITUATIONS 5.1 TOOLS AND METHODS FOR CONTROL OF ABNORMAL SITUATIONS Much of the available literature on the management of abnormal situations focuses on Human Machin e Interface (HMI and procedural issues, and to some extent hazard identification (HAZID)) techniques to identify those scenarios that shou ld be considered in an HMI or procedure analysis. Journal articles by Errington, Bullemer, and Ostrowski describe the importance of each of these topics (Errington et al 2005; Bullemer et al 2010a; Ostrowski & Keim 2010). However, as illustrated by the wide range of real -world example incidents introduced in Chapter 3, abnormal situations are clearly not limited to only HMI and HAZID-related events. Therefore, this chapter takes a more holistic approach, considering a broader number of subject areas, arranged into these eight areas: Predictive Hazard Identification Process Control Systems Policies and Administrative Procedures Operating Procedures Training and Drills. Ergonomics and Other Human Factors Learning from previous Abnormal Situation Incidents Change Management Most of these subject areas are similar to the ASM® Consortium research areas described in Chapter 3, Section 3.1.1. Each subject area is discussed separately, along with associated tools/methods and links to some of the example incidents fr om Chapter 3 noted as applicable. Table 5.1 provides a summary of this chapter, along with references to applicable example incidents from this book, to enable the reader to find examples of each of the areas easily.

2 • Defining the Transition Times 23 2.4 A start-up incident A start-up incident, shown in the illu stration in Figure 2.1, resulted in a significant release of an ammonia cloud that drifted across a river [18]. This incident occurred when a roof-mounted pipe failed upon restart of the facility’s refrige ration system after a seven-hour power outage. More than 150 people reported exposure to the released ammonia, with thirty-two people being admitted to the hospital and four being placed in intensive ca re. What happened? Which transient operating mode applies? How can systems be implemented to prevent incidents like this from happening again? The following chapters will explore this incident—and many more—to help understand “what went wrong” and how to prevent in cidents from occurring during the transient operating mode. As shown in Figure 2.2, this incident occurred during the start-up after an unscheduled shutdown. This figure provides an overall timeline for helping identify the time of the incident relative to normal, abnormal , and emergency operations and will be discussed in more detail th roughout the rest of this guideline. It will be used to determine which transient operating mode applied at the time of the incident. Additional details on this ammonia release incident are described in Chapter 7, Case 7.6.3-1.

236 INVESTIGATING PROCESS SAFETY INCIDENTS and a contractor injury. Emergency response was impaired because the firewater pumps were inoperable, which contributed to the severity of the consequences. The fire spread to the vertical catalyst storage tank. A subsequent explosion of an adjacent catalyst storage tank resulted in the injury of four firefighters. The local fire department and plant fire brigade extinguished the fire at 12:10 PM. For this example, the first event will be considered. The top portion of the tree for the operator fatality is developed in Figure 10.18. Figure 10.18 Operator Fatality Branch The pool fire branch is further developed in Figure 10.19. Figure 10.19 Fire Branch

230 INVESTIGATING PROCESS SAFETY INCIDENTS the bottom of the filter. In addition, there wa s no way for the employee to tell if the pressure was still on the filter, since the pressure gauge could become plugged as well. In this case, the investigation team recommended that a pressure indicator an d a separate vent valve be added to the filter. Figure 10.17 Expanded Logic Tree Sample, Employee Burn [Note – Tree Top: the Tree Bottom is on the following page.]

397 5.In the revised system, Reactant B is charged to the feed tank through a three-way valve to th e bottom of the feed tank. The three-way valve allows flow either from the storage tank to the Reactant B feed tank, or from the feed tank to the reactor. It is not possible to pump Reactant B directly from the storage tank to the reactor. This system ma kes it much more difficult to overcharge Reactant B. Figure 15.4: Modified, inherently safer batch reactor system

Evaluating the Prior PHA 45 policies may specify consequence thresh olds, such as “serious danger to employees,” “environmental damage requir ing remediation,” or “loss of market share.” It is usually easy to verify whether the analytical scope of a PHA meets current regulatory and policy requirements by comparing the stated scope with those requirements. If the “consequences of interest” are not explicitly stated in the prior PHA, they can usually be infe rred from the consequences listed in the core analysis worksheets or from the cons equence categories of any risk matrix that was used. However, if the consequenc es of interest are not clearly stated, the reviewer should be alert for other indications that the prior PHA was poorly documented. As discussed in Section 3.1. 4, clear documentation of the hazards, risk controls, and consequences of their failure is essential for the Update approach to be a viable option. Beyond specifying the types and thresholds of consequences to be considered, the analytical scope must in clude all the required topics. PHAs are usually required to address human fa ctors and facility siting issues. The analytical scope might also include other topics, such as loss of essential utilities and services, corrosion damage mechanisms , or the hierarchy of risk controls. Facility Siting. Specific facility siting issues usually arise during a PHA team’s core analysis. In addition, facilit y siting issues may arise during consideration of external events, such as floods, earthquakes, or accidents (e.g., fires, explosions, or toxic chemic al releases) at neighboring facilities. However, facility siting issues may be more comprehensively addressed in complementary analyses in a variety of ways. The CCPS has consistently recommended that PHA teams perform a qualitative evaluation of facility siting issues, such as the example checklists provided in Appendix E. The CCPS has also published a monograph on assessment of natural hazards [32]. Some jurisdictions or compan y policies may require complementary analyses to identify plausible scen arios and perform detailed dispersion, fire, flood, seismic, and/or explosion modeling to identify facility siting issues. Human Factors. Like facility siting issues, some human performance gaps, and in many cases specific mistakes (wit h triggers), are identified in the PHA core analysis. These are often listed as specific causes or factors that can contribute to hazardous events. However , the core analysis techniques do not focus on the underlying causes of human error. For example, teams may list specific actions taken by operators and others to prevent or mitigate the effects of process upsets without considering how robust the human response safeguards are. In such cases, as discussed in Section 4.2.5, the