Datasets:

hndc

/

AI4Hazard-small

like 0

54 PROCESS SAFETY IN UPSTREAM OIL & GAS ●Fundamentals of Drilling Engineering (SPE, 2011) ●Petroleum Engineering Handbook Vol 2: Drilling Engineering (SPE, 2007) Additional references from both IADC and SPE are available via their respective websites. ABB (2013) published a handbook that provides a simpler overview of the complete upstream production process, including details on the reservoir and well. Multiple recommended practices (RP) re lated to well construction and well integrity are available from API for onshore and offshore wells (regularly updated and summarized in API, 2015), Norsk O&G (2016) for offshore well integrity, and NORSOK (2013) for well designs achieving two barriers. In addition to these standards and RPs, most large companies produce their own company specific well construction manuals. These may exceed sta ndards and RP requirements based on company experience and in dustry good practices. In some ways a loss of well control is similar to an abnormal situation for a downstream processing plant. The event mu st be recognized and addressed before the situation escalates to something more serious. Generally, more time is available to deal with a kick event than, for example, many runaway reaction situations, but the indication may be less obvious. Some relevant incident descriptions are provided in this chapter to highlight examples and the possible application of RB PS. The first incident is the Deepwater Horizon loss of well control incident, named after the rig involved. It is also widely known as the Macondo incident, based on the prospect name. This book standardizes on Deepwater Horizon. 4.1.1 Drilling the Well: The Well Bore The most important geologic, reservoir, and geomechanics factors related to process safety and loss of containment are pore pressure and fracture gradient, which are unique to each well. A summary of pore pressure and fracture gradient terms follows, including partial reference to the Schlumberger Oilfield Glossary. Pore Pressure: The pressure of the subsurface formation fluids, commonly expressed as the density of fluid required in the wellbore to balance that pore pressure. In reservoir zo nes which have sufficient permeability to allow flow, this is the pressure of the hydrocarbons or other fluids trying to enter the wellbore. Safe well design balances the reservoir pressure with drilling muds of adequate density such that the mud hydr ostatic pressure at the reservoir is sufficient to prevent inflow. Fracture Gradient: The pressu re required to induce fractures in rock at a given depth. If the fracture gradient is exceed ed, then some of the dense drilling mud can be lost into the formation leading to a potential loss of hydrostatic head. If the pressure exerted by the hydrostatic h ead falls below the local pore pressure in the reservoir zone, then hydrocarbons can flow into the well.

146 GUIDELINES FOR MANAGING ABNORMAL SITUATIONS guide “ Crew Resource Management for Well Operations Teams ” (IOGP 2020) that conducted a literature search and survey then listed key NTS categories and elements for wells op erating personnel. These are also relevant to other hazardous in dustries and are as follows: Table 5.8 Non-Technical Skills , Categories and Elements Category Elements Situation Awareness Gathering information Understanding information and risk status Anticipating future state/developments Decision Making Identifying and assessing options Selecting an option/communicating it Implementing and reviewing decisions Communication Briefing and giving feedback Listening Asking questions Being assertive Team Work Understanding own role with the team Coordinating tasks with team members/other shift Considering and helping others Resolving conflicts Leadership Planning and directing Maintaining standards Supporting team members Performance shaping factors - stress and fatigue Identifying signs of stress and fatigue Coping with effects of stress and fatigue

F.2 Culture Assessment Protocol |361 146. Have the results of hazard/risk analyses been used to plan, organize, and execute the other elem ents of the PSMS? Exam ples: use of the causes and safeguard information developed during HIRAs/PHAs to determ ine which equipment should be included in the AI/M I program, the use of HIRA/PHA results to determ ine process safety related training for operators, m aintenance, and other personnel, the use of HIRA/PHA results to determ ine the contents of the emergency response plan? 147. Has the organization included analysis of inherently safer design (ISD) considerations in its process safety program ? Are appropriate ISD provisions implemented when feasible? 148. (U.S.A.-specific) Has the organization interpreted what constitutes offensive vs. defensive actions when trying to determ ine whether a U.S. facility is responding to emergency events at a level that would invoke the HAZWOPER (1910.120) regulations? This is affected to a large degree by the emergency response culture that has been established within the emergency response team at the facility, and by the philosophy that was used to develop and emergency provisions of the operating procedures. 149. Have key leading and lagging process safety m etrics been established and reported to m anagement on a periodic basis? Are the process safety m etrics defined in such a way as to artificially indicate a PSMS status that is not com pletely accurate. For exam ple, overdue ITPM m etrics that include only those from the m ain MI/m aintenance data base and do not include all of the process safety-relevant ITPM tasks being perform ed (e.g., fire protection equipment m anaged separately)? 150. Do process safety metrics never vary from very high/positive values? While this may seem satisfying, it usually does not com port with the reality of actual facility operations. 151. Are PSMS audits and process safety metrics m et by severe pushback?

Acknowledgements xxvi Initial manuscript peer reviewers (2018): Daniel Callahan Stepan Charles Foshee Chevron Jennifer Mize Eastman Al Morrison Chevron Although the peer reviewers provi ded comments and suggestions, they were not asked to endorse this guideline and did not review the final manuscript before its release. The book committee wishes to express their appreciation to Elena Prats and Kathy Anderson, ioMosaic Corporation, for their contributions in prepari ng the guideline’s draft manuscript. Sincere appreciation is extended to Dr. Bruc e K. Vaughen, PE, CCPSC, of CCPS for his contribution in restructur ing the book committee’s efforts, addressing the final comments from both the book’s committee and peer reviewers, and in creating the final, published manuscript. Before publication of the final man uscript for this guideline, an additional technical review of each of the chapter’s drafts and the restructured and enhanced manuscript provided additional insights that were incorporated into the final, published manuscript. Much appreciation is extended to the final reviewers for their time. Restructured draft and final manuscript reviewers (2020): Theresa Broussard, Chai r Chevron Corporation Dan Sliva CCPS Staff Consultant Jennifer Bitz CCPS, Project Manager Pete Lodal Eastman Dr. Anil Gokhale, PE CCPS Project Director Special appreciation is extended to Kiezha Ferrell for copy editing the draft of the final manuscript.

130 switchgear can be relocated outside of the classified electrical area, rather than be designed for it. This makes the installation inherently safer (completely removes the risk of ignition rather than reducing it), and less expensive (standard electric al enclosure vs. one designed for classified locations). (Ref 7.6 Kletz 2010). Relocating personnel whom could be potentially impacted from a fire, explosion, or toxic release is another moderation strategy which can be employed. One refinery moved their control building and plant personnel offices to a re mote location (across the street, well away from the toxic and blast/fire zones) and purchased property around the site to create a buffer zone. This approach is a common facility siting technique which does not remove th e chemical hazard but separates people from the hazard. 7.4 CONTAINMENT If it is not feasible to contain a r unaway reaction within the reactor, it may be possible to moderate the consequences, by piping the emergency device effluent to a separate pr essure vessel for containment and subsequent treatment. Quench drums, vapor-liquid separation vessels, vapor-liquid separators, and other simi lar devices can be used to contain the effluent from exothermic / runaway reactions (Ref 7.1 CCPS 1993). Adequate secondary containment for tanks, vessels, or for entire process units is also an application of the moderation IS strategy, as it prevents the spread of liquid rele ases and minimizes the surface area for evaporation. Common containment structures that surround multiple tanks should be avoided or minimized if possible. The ongoing integrity of containment structures is also an important issue, particularly for earthen berms that can settle, erode, or otherwise weaken or lose their design capaci ty. Penetrations through secondary containment walls should be avoided or sealed properly. Another example of applying the moderation IS strategy to a layer of protection is the use of blast walls, heat shields, and other barriers to absorb the energy from explosions and limit their radius of effect, or to absorb other potentially hazardous energy sources such as sound and thermal energy, as illustrated by the following:

247 Table 12.2. Checklist analysis overview Typically Used During Resource Requirements Type of Results Advantages and Disadvantages Conceptual design Pilot plant operation Detailed engineering Construction / startup Routine operation Decommissioning Expansion or modification During What-If or HAZOP studies to address facility siting, human factors, and other general issues Material, physical, and chemical data Basic process chemistry Process flow diagram Operating procedures Piping and Instrumentation Diagrams Response to pre- defined questions. Documentation of compliance. Can be used with less experienced personnel if the experience is captured in the checklist. Quality of the analysis is only as good as the quality of the checklist. Checklists that are too long or don’t relate specifically enough to the process being analyzed may have a tendency to be completed without thorough evaluation. What-If Analysis The What-If Analysis technique is a brainstormin g approach in which a multidisciplinary team of experienced people familiar with th e subject process ask questions such as: What if the wrong material is delivered? What if Pump A stops running during start-up?, and What if the operator opens valve B instead of valve A? The purpose of a What-If Analysis is to identify hazards, hazardous situations, or scenarios that could produce undesirable consequences. The team identifies possible causes, their consequences, and existing safeguard and docu ments this in a worksheet. In some What-If analyses, the consequences can be risk ra nked to facilitate prioritization of any recommendations. They then suggest reco mmendations for risk reduction where improvement opportunities are identified or wh ere safeguards are judged to be inadequate. The method can involve examination of possibl e deviations from the design, construction, modification, or operating intent. It requires a basic understanding of the process intention, along with the ability to mentally envision possible deviations from the design intent that could result in an incident. This is a powerful technique if the staff is experienced; however, an inexperienced team may overlook potential caus es and consequences. Table 12.3 provides an overview of What-If Analysis requirements and results. What-If Analysis is well suited for addressing “what can go wrong?” by identifying cause-consequence pairs. HAZARD IDENTIFICATION

EVIDEN CE ANALYSIS & CAUSAL FACTOR DETERM IN ATION 183 is opened to see what is inside, the cover cannot be replaced in exactly the same manner it was originally. The oxidation layers and adhesives used to seal the cover cannot be replaced exactly as they were. Once a pump is hand rotated, it cannot be di sassembled to see the position in which it came to rest following the failure. Consequently, investigators must be careful to think about the data that is needed an d what data could be altered or destroyed when certain actions are take n. Protocols are intended to help investigators think ahead. Protocols also serve to gain agreement from multiple parties on how, by whom, and when the test should be performed. Typically, protocols are designed to an swer one or more of the following questions: How does the part work? Did the part functi on as intended? How did the part fail? Why did the failure occur? Protocols should be developed before the analysis of physical data is started. Protocols help: Ensure complete collection of required data Ensure complete analysis of the data Prevent inadvertent destruction of data by the investigators Gain agreement from all parties involved in the investigation concerning the analysis processes and methods Ensure the test is worth doing before it is done Identify decision points in the analysis The protocol sh ould include: The objective of the investigation activity The methods for performing the activity Safety considerations for executing the protocol A description of the methods/procedure Names of the persons who will perform the tasks in the protocol Scheduled times and locations of the protocol How the protocol results will be recorded and reported Information on multiple tests of the same item Disposition of the test sp ecimens after the protocol The order in which the different steps of the protocol will be executed

Piping and Instrumentation Diagram Development 118 screwing its ports by fitting it between the flanges of the two sides of pipes (Table 7.17). Generally valves are installed between piping sides without any other fittings. However, one important exception is when installing control valves. Sometimes, the selected control valve has smaller body size that can be fitted with the pipe size. In such cases, a reducer on the inlet of a control valve and one enlarger on the outlet of the control valve may be needed. 7.11.1 Valv es in Series A manual valve can be used in series, one blocking type and one throttling type. If a stream needs to be adjusted manually, and sometimes the stream should be totally stopped and tight shutoff is important, it is a good idea to use a manual blocking valve and then manual throt ­ tling valve in series (Figure 7.14). This arrangement can be used in services like toxic fluids or high‐pressure streams. Two (or more) manual throttling valves or two (or more) manual blocking valves are rare, which sometimes is considered a bad practice in P&ID development. However, sometimes having two or more manual block ­ ing valves in series happens. Each piece of equipment needs isolation valves around it for ease of maintenance. However, when there are two pieces of equipment, one upstream and one downstream, and they are close to each other, two of their isolation valves sit close to each other in a series position. In such cases, one isolation valve can be eliminated (Figure 7.15). Valve arrangement in series can be used for control valves or regulators. These are for the cases that a large pressure drop is needed in a stream. A large dropping pressure may cause vibration, noise, and erosion in the valve [2].A rule of thumb helps to decide when two regulators in series may be needed: ●Where a pressure drop more than 100 psig i s needed (or maximum 150 psig). ●Where pressure should be dropped to a value less than 1/10th of upstream pressure. ●Where the pressure on downstream should be accu­rately regulated (e.g. less than few psig). 7.11.2 Valv es in Parallel The parallel arrangement of two manual valves may be used. There are some cases that a manual blocking valve needs to be placed on a stream that has high pres ­ sure. In such cases, placing one single blocking valve, for example, a gate valve, makes life hard for the opera­tor who will have to open a manual valve from a fully closed position under high pressure (e.g. more than 3000 KP a). To solve this problem, another smaller‐ sized manual blocking valve is installed in parallel with the main valve. When the operator wants to open the main valve, the small bypass valve is opened at the beginning to equalize the pressure in both sides of the main valve, and then the main valve can be easily opened (Figure 7.16). Parallel manual valves could be used for other reasons, such as providing a minimum flow in the pipe even when the main valve is closed or for start‐up. As was discussed in Chapter 5, the general method of starting up a piece of equipment involves gradually opening the valve of the inlet stream. If the operator does not want to open the valve suddenly, a parallel and smaller manual valve can be added to the main valve and the operator can open it Figure 7.16 Tw o manual blocking valves in parallel.Table 7.17 P&ID symbol for c onnecting valves. Valve size Connection type P&ID sketch Nominal size <2″ Weld Screw Nominal size >2″ Flange Eq1 Eq1Eq2 If there are no branches Eq2 Figure 7.15 Tw o manual blocking valves in series and saving opportunity. Blocking Throttling Figure 7.14 Manual v alves in series: blocking and throttling.

3.1 Definition of Process Safety Leadership |81 at least one of four cultural factors as a root cause alongside [regulatory compliance] failures. A [PSMS] m ust be accompanied by a strong culture that requires critical leadership behaviors. If process safety leadership were a job description, there would be four basic competencies essential to success.” Leaders having these four com petencies should: Have the conviction to lead safety, Understand how process safety works, Possess (and practice) great leadership skills, and B e able to influence people. Motivated by culture lessons-learned from the 2005 incidents in Buncefield, Hertfordshire, UK and Texas City, TX, USA, the UK HSE in 2006 established a partnership of industry and regulators called the Process Safety Leadership Group (PSLG). The PSLG’s goal is to drive high standards in process safety leadership in the UK and to implement recommendations m ade by the B uncefield Major Incident Investigation B oard. PSLG (Ref 3.15) endorsed the com petencies noted by Stricoff and recomm ended the following leadership actions: Address process safety leadership and culture at the B oard of Directors’ level, and include at least one Board mem ber who is fully conversant in process safety to support the board’s governance and strategic decisions, Engage the workforce in the developm ent, prom otion, and achievem ent of process safety goals, Provide sufficient resources at the operating and leadership level, all having the appropriate level of process safety experience; and Monitor process safety performance based on process safety leading and lagging indicators. • • • • • • • •

130 INVESTIGATING PROCESS SAFETY INCIDENTS Throughout the interview, the investigator should: • be friendly, respectful, and professional • listen attentively and reflectively • show compassion • avoid attitudes that destroy rapport • remain as neutral as possible • project a calm demeanor • use language/terms that the witness understands • observe body language/facial expressions During an interview the investigator should not: • act surprised when the witne ss provides new information • act happy or pleased when the witness confirms other witnesses’ testimony or a current theory of the causes of the occurrence • be overbearing, commanding, proud, overly confident, overeager, timid, or prejudiced • judge the informat ion that is be ing presented by the witness, even if it is incorrect • rush the witness, even if little new information is appearing • make promises to the witness Remember that the point of the interview is to obtain as much information from the witness as possib le, not to show the witness how smart the interviewer is. Instead, convey respect to the witnesses for their experience, knowledge, and the information that they can provide to help lead the investigation team to the correct conclusions. 7.3.4.10 Promoting an Uninterrupted N arrative Using open-ended questions (questions that require mor e than one word yes or no answers), ask the witness fo r an initial statement. Examples of open-ended questions are prov ided in Table 7.1. It is important during this portion of the interview to remain quiet. Allow the witness to talk. As long as the interviewer is talking, the witness will remain quiet. Do not interrupt with follow-up questions after asking an open-ended question. Try to avoid closed-ended questions (those that only require short answers) during the initial portion of the interview. Too many closed-ended questions at the beginning of the interview can conditio n the witness to give short answers.

3 • Normal Operations 38 Procedures for all of the transien t operating modes listed in Table 1.1 may need additional, specific st art-up or shut-down related steps, checklists, and decision aids to address potentially hazardous conditions that may occur during the transition, such as the following: Additional personal protective equipment (PPE) required during the transition Special handover protocols before and after scheduled projects or scheduled main tenance (Chapters 4 and 5) Special start-up protocols after curtailed operations (e.g., during reduced customer demand; Chapter 4) Special operations shut-down-related activities during weather extremes (i.e., reduce the potential for freezing when the process is not operating during the winter months; Chapter 4) Special start-up protocols after an emergency shut-down (especially if the end state for th e process is not at its normal, safe, idle, and at-rest condition; Chapter 8), and Special shut-down protocol s for mothballing or decommissioning equipment (Chapters 6 and 9). An operating phase checklist noting some typical procedural steps that may need to be considered for a transient operating mode, depending on the hazards, includes the following modes [24]: Normal Shut-down for a Turnaround Start-up after a Turnaround Normal Shut-down for Standby Mode Start-Up after a Warm Shut-Down (a system put in standby mode) Emergency Shut-down Start-up after an Emergen cy Shut-down (ESD), and Initial Start-up/Commissioning. These modes will be covered in more detail later in this Guideline. One incident which revealed we aknesses in a routine start-up procedure occurred when fatigu ed heat exchanger equipment

61 Limiting the amount of energy available, by reducing the temperature of heating media. The runaway reaction in Seveso, Italy in 1976, which resulted in dioxin contamination of the surrounding farmland, resulted from overheatin g the reactor vessel with a steam supply whose temperature was both far above that necessary for effective heat transfer, and far abov e the initiation temperature of the runaway reaction. 3.11 REFERENCES 3.1 Agreda, V. H., Partin , L.R. and Heise, W.H., High-purity methyl acetate via reactive distillation. Chemical Engineering Progress, 86 (2), 40-46, 1990. 3.2 Center for Chemical Process Safety (CCPS), Guidelines for Engineering Design for Process Safety . New York: American Institute of Chemical Engineers, 1993. 3.3 Committee on Inherently Safe r Chemical Processes: The Use of Methyl Isocyanate (MIC) at Bayer CropScience, National Academy of Sciences, 2012 3.4 Doherty, M., and Buzad, G. Reactive distillation by design. The Chemical Engineer, s17-s19, 27 August 1992. 3.5 Englund, S. M. “The design and operation of inherently safer chemical plants.” Presented at the American Institute of Chemical Engineers 1990 Summer National Meeting, August 20, 1990, San Diego, CA, Session 43. 3.6 Englund, S. M. Design and operate plants for inherent safety - Part 1. Chemical Engineering Progress, 87 (3), 85-91, 1991a. 3.7 Englund, S.M. Design and operate plants for inherent safety - Part 2. Chemical Engineering Progress, 87 (5), 79-86, 1991b. 3.8 Englund, S.M. Process and design options for inherently safer plants. In V. M. Fthenakis (ed.). Prevention and Control of Accidental Releases of Hazardous Gases (9-62) . New York: Van Nostrand Reinhold, 1993. 3.9 Hendershot, D.C., et al. Implementing inherently safer design in an existing plant. Process Safety Progress, 25 (1), 52-57, 2006.

80 Guidelines for Revalidating a Process Hazard Analysis If the answer to any of these questi ons is “yes,” then past operational experience may not be a reliable indicator of future expectations, and operational experience after the change becomes significantly more important. If the change has had (or is expected to have) minimal or no direct effect on the unit, then the Update approach is a viable option; however, if the change has had (or will likely have) a major impact in area s such as central control, maintenance backlogs, or emergency response, then the Redo approach will probably be the better choice. Staffing. A second area of inquiry should explore any changes in the number of front-line workers or their responsibiliti es. Staff changes may adversely affect human factors related to detecting or responding to process upsets. If staff numbers have been significantly changed si nce the previous PHA, issues such as the following should be considered: • Are workers subject to mandator y overtime requirements? Are there any limits on the total hours worked or the consecutive hours worked in a defined time period? • Are required practices for mana ging fatigue being maintained? • Are exceptions to the Fatigue Man agement policy being handled as before the staffing change? • What tasks are not being done, or are being done differently, due to the staff changes? • Is there evidence that field checks of equipment and local instruments are being done as required? • Are there enough operators to respond to alarms in a timely manner, particularly during major upsets? • In upset situations, must operators wait for assistance before responding? • Are there enough operators to walk down job sites before issuing work permits? • Are the backlogs of maintenance or engineering tasks increasing? • Has the facility transitioned fr om 24/7 operator presence to unoccupied remote monitoring on weekends or night shifts? • Are there chronic personnel shortages that affect operations on a particular shift or day?

172 | 5 Aligning Culture with PSMS Elements team may be found to have been too conservative, and a lesser solution is acceptable. Conversely, som etimes the recomm endation m ay be found to not fully address the risk, and stronger m easures are found to be needed. In the end, a recomm endation should be considered closed only when the risk that the recommendation addressed has been m anaged by implementing a suitable solution. In the Risk Based Process Safety approach as well as some corporate and regulatory approaches, the process risk may be used to guide the efficient and effective use of resources in carrying out the PSM S elements. From a culture perspective. tailoring level of effort to risk also helps empower employees to fulfill their process safety responsibilities by focusing their efforts productively. Table 5.2 provides examples of how higher and lower risks might be addressed in som e PSMS elem ents. While Table 5.2 shows actions in two categories of risk, companies m ay have three or more action categories. Table 5.2 Exam ple of Tailoring PSMS Actions to Risk PSMS Elem ent Higher Risk Lower Risk HIRA Deeper risk analysis, e.g. QRA Faster risk analysis, e.g. checklist Asset Integrity More rigorous inspection, testing, and preventative maintenance schedule Run to failure Managem ent of Change M ore rigorous evaluation; higher level sign-off Less rigorous; lower level sign-off Auditing Supplement required audits with more frequent inform al audits Required audits only Metrics, m anagem ent review Specific metrics, more frequent management review General metrics, less frequent review

Manual Valves and Automatic Valves 121 when the pipe diameter size is large enough to contain a decent amount of liquid if concentric reducer and enlarger are to be used. Therefore, for cases that deal with aggressive liquid and large pipe size (say, more than 3 in.), it is a wise decision to use eccentric reducer and enlarger with flat on bottom (FOB). The other provision for draining/venting the piece of pipe between two isolation valves is putting drain valve on the bottom of the pipe and vent valve on the top of the pipe. However, if the pipe size is small enough (say, less than 3 in.), po ssibly just one valve from the bottom of the pipe could be enough to be used as drain valve and vent valve. Now the question is whether the drain valves are needed in both sides of control valve or just one side of that. There is a heated debate among professionals regarding this question. If there is an available guideline in your company you need to follow, otherwise you need to make the judgment whether to put two drain valves in one side of the control valve or just one drain valve downstream or upstream of control valve. There are at least three different answers for this question. A very conservative approach says that we need to put the drain valve in each side of the control valve to make sure that each side of the control valve can be drained independently. This approach is the best for critical cases like when dealing with toxic, hazardous, or high‐pressure stream. This is also a good decision when a control valve is FC. In such cases where the control valve is FC, there are trapped liquids in each side of the control valve, one trapped liquid upstream of the con­trol valve and the other trapped liquid downstream of the control valve. Therefore, two drain valves help the operator to drain each trapped liquid easily from each side of the control valve. The other approach says that one drain valve for this piece of pipe is enough. Professionals who are in favor of this solution are faced with the following question: “what if the control valve failed or jammed in closed position?” They answer that the control valve could be opened by a jackhammer and again there will not be any two separate trapped liquid. Therefore one drain valve is enough. If we chose this approach, there are again two available options: if the single drain valve should be upstream or downstream of the control valve. Some people prefer a single drain valve upstream of  the control valve. They believe that a drain valve upstream of the control valve helps us to drain the higher pressure side of the control valve more safely and if it failed to open up a jammed closed control valve, the other side of the control valve has lower pres ­ sure and possibly does not need to be drained through the drain valve. The downstream of the control valve can be naturally drained after disassembling the control valve and removing it from the piping arrangement. Putting drain valve upstream of the control valve has some other advantages for the operators. They can use this valve for start‐up and for purposes of chemical cleaning. The other option is putting the single drain valve downstream of the control valve. This option also has some supporters. To summarize the discussion about the need for drain valves on the control station, some people believe that we need to consider the failed position of the control valve. However, some other people do not take this into consideration on putting drain valve or drain valves around the control valve. The last thing that should be decided is the type of bypass manual throttling valve. The workhorse of the industry for throttling valve is the globe valve. Therefore, wherever we want to put a bypass throttling valve for a control valve, a globe valve is selected. However, globe valves are not available (or are very expensive) in larger sizes, probably not larger than 4 or 6 in. Where t he con­ trol valve size is larger than 4″ or 6″, there are some options available. One available option is using butterfly valves. Butterfly valves are good throttling valves and are available and affordable for large sizes, for example, more than 4 or 6 in. Howe ver, not all companies and professionals are in favor of using butterfly valves. The conventional butterfly valves have some inherent drawbacks. They may have internal passing‐by that makes them unsuitable valves for high‐pressure systems and where the service fluid is an aggressive fluid. If the required size of throttling valve is more than 4 or 6 in. and butt erfly valve is not an acceptable option, there are some exotic designs available. One completely acceptable option is providing the required valve capacity of the control valve through several small (less than 4 or 6 in.) glob e valves. In this solution the bypass of the control valve could be an arrangement of two (or more) 4″ manual globe valves. This arrangement – shown in Figure 7.21 – is fully func ­ tional but is very expensive. If it is known that the movement of the control valve stem is only in a short and limited span, a manual globe valve can be replaced with a gate valve and a smaller globe valve in parallel (Figure 7.21 top schematic). However, it has been seen in cases that a company decided to put a gate valve as the bypass valve for a con­trol valve. As gate valve is NOT a throttling valve, it is not a good choice in this situation. However, a gate valve in parallel to a manual globe valve can be considered if it is discovered that the control valve mainly works on its extreme sides of it range. At the end it should be mentioned that some companies and professionals believe it is not a good idea to provide

P re face |xxv This book offers several definitions of process safety culture. Even though there may be some disagreem ent about a definition of process safety culture, when you visit a facility you very quickly get a sense how im portant a positive process safety culture is to the facility. You will know it when you see it. From the first m oment when you encounter a security guard or a receptionist to a tour of a control room you can quickly gauge the culture. Are process safety metrics displayed around the plant? Are operators com municating with each other in a professional m anner? Is the senior manager well versed in the hazards of the operation? As you read this book you will learn many aspects of how to develop a sound process safety culture. From my experience, a strong process safety culture must start with leadership. B y leadership I mean everyone in a leadership position from the chairm an of the board to the supervisor on the shop floor. They m ust set the exam ple. It starts with leadership being aware of the hazards in their processes and putting in place the organization and expertise to control those hazards. Just as im portant, the senior leadership m ust communicate his or her concerns about the need for an effective process safety program . These concerns should be an ongoing part of senior leadership’s communications with the organization. This is the way to ensure the establishment of a culture of process safety across the organization. I comm end CCPS on the publication of its latest book and I encourage readers to turn its lessons into actions in their day-to- day work of ensuring safety for em ployees, contractors and the surrounding com munity. As well as saving lives and preventing injuries it is vital for the financial success and reputation of the chem ical process industries. John S. B resland Shepherdstown, West Virginia

90 INVESTIGATING PROCESS SAFETY INCIDENTS iii. Loss of Production The loss of production may be used as a classification criterion and could be expressed in units of hours, da ys, or weeks of expected downtime. A further improvement is to es timate both the actual and potential severity of the impact of such incidents. Making such a determination is an imprecise effort, and organizations are best served when a decision is made quickly with the evidence at hand rather than waiting for more perfect data. 5.3 IN CIDENT NOTIFICATION Depending upon the severity and type of incident, various stakeholders may need to be notified that an incident has occurred. These stakeholders may be internal (e.g., corporat e executives, key departments) and external (e.g., regulatory agencies, pa rtners, local government, etc.). All external notifications should follow the compan y’s policy and procedures for external communications. Making initial notification in a timely manner can be challenging immediately following an incident. The form at and timing of all external notifications should be identified and incorporated in to the management system before an incident occurs. The corporate emergency response and/or the incident investigation management systems should address how to handle these communications, and how to coordinate with facility emergency response plans. A checklist with key contact names, titles, and phone numbers may be developed and ke pt up-to-date for this use. With this information readily at hand, the pr oper notifications ca n be made quickly and accurately when an incident occurs. 5.3.1 Corporate N otification Initial notifications to the company’s headquarters may al ert executives and key departments (e.g., EHS, Legal, etc.) that an incident has occurred. Some companies only require notification fo r more severe incidents, while lesser incidents are simply entered into th e company’s reporting database. For example, some companies only re quire executives and corporate departments to be informed of CCPS se verity level 1 and 2 incidents (see Appendix G). Such incidents may ha ve implications for the company’s reputation and its license to operate and may justify a more thorough investigation approach. Some companies require initial notification within a certain timeframe, typically 8 hours to 1 day.

INVESTIGATION M ETHODOLOGIES 39 rather applies the causal factors to each branch in turn an d identifies those branches that are relevant to the specific incident. Like checklists, the comprehensiveness of the various predefined trees varies. Some are very detailed with nu merous categories and subcategories, whereas others may not fully reach root causes. This is hardly surprising, as the predefined trees are essentia lly a graphical representation of numerous checklists, organized by subjec t matter, such as human error, equipment failure, or other topics. The more comprehensive techniques were developed from many years of incident experience and management system experience across the chemical and allied industries. The advantages of predefined trees ar e that they may bring expertise into the investigation that the team do es not have, and, by presenting all investigators with the same classifi cation system, greater consistency is encouraged among investigators. La rgely, the technique ensures a comprehensive analysis and simplifies statistical trend analysis of the collected data. A disadvantage of predefi ned trees, as with a checklist, may be a tendency to discourage lateral th inking if the incident involves novel factors not previously experienced by thos e who developed the original tree. The use of predefined trees, overall, requires fewer resources and less prior training than the non-prescripti ve techniques involving team- developed trees that are discussed below. Some organizations have taken a generic, predefined tree and st ructured it along the lines of the company’s management system. The effectiveness of a predefined tree is dependent on how well the tree models the data and syst em of dealing with the incident. When choosing a predefin ed tree, the user should confirm that the tree models the technology an d system of the user. 3.3.3 Team-Developed Logic Trees Logic tree analysis is a top-down, anal ysis in which an undesired state of a system (e.g., injury, fire, explosion, or toxic release) is analyzed using Boolean logic to combine a series of lower-leve l events. Logic trees can vary over a wide range from simple trees to comple x fault trees. Most start at the end occurrence (e.g., injury, fire, explosion, or toxic release) and work backward until a point is reached at which the team agrees it would be unproductive to go further. Logic trees are best developed using a multi-discipline team. Starting at the end event, the discussion is gu ided by asking “Why?” and recording the results in a tree format. The general ap proach encourages investigators to

CON TIN UOUS IM PROVEM EN T 333 Table 15.3. Example Categories for In cident Investigation Findings (cont.) Category Circle Defining Statements M aintenance Procedures T / F T / F T / F T / F The maintenance procedures were: • available • adequate • accurate • approved and enforced (The focus of this category is the actual maintenance tools, techniques, and standards for work that go beyond the traditional scope of normal inspection and preventive maintenance activities.) Training T / F T / F Training was: • available and timely • adequate and verified to be effective to achieve functional and compliance requirements Inspection and Preventive M aintenance T / F Inspection and preventive mainte nance were in accordance with applicable procedures, manufacturer’s or experience-based recommendations and governing standards, and were adequate for the service conditions. Equipment and M aterials T / F The equipment, parts, and materials as initially procured were as specified, were not defective, and met or exceeded the applicable specifications. Personnel Fitness T / F Personnel were “fit for duty.” (Includes physical/mental/ emotional states and addresses preexisting physical conditions, substance abuse, and other related concerns.) Human Actions T / F Personnel actions, activities, and de cisions were in accordance with procedures, training, and expected workplace standards. External T / F External items including weather and external third party actions/events were not creating out-of-design conditions. Other T / F The incident has been satisfactorily classified in one or more of the above categories. It is important to understand that the above approach is only used after the investigation has been concluded. It is not a technique to be used for the investigation itself; rather it is an aid to identify the br oad categories into which the findings of investigations are falling. An analysis of the data collected will provide management with information on root causes and causal factors that repeat, which could be indicative of an improvement opportunity for the incident investigation system or another management system.

206 | Appendix: Index of Publicly Evaluated Incidents Section 2: Culture Core Principles (Continued) Combat Normalization of Deviance—Secondary Findings A4, A6 C20, C21, C26, C27, C44, C68 D19, D25 J25, J27, J28, J42, J52, J65, J114, J167, J168, J173, J194, J195, J233, J236, J237 S2, S4 Section 3: Selected Causal Factors Consequence Analysis—Primary Findings A2, A5 C23, C24, C56, C69 D21 HA10 J119, J143, J146, J149, J154, J156, J164, J165, J173, J174, J181, J182, J195 S3, S16 Consequence Analysis—Secondary Findings A10 C17, C37, C39, C44, C48, C49, C70, C73 D42 J37, J80, J86, J98, J101, J129, J132, J147, J152, J153, J155, J171, J176, J180, J196, J233, J235, J239, J260 S4 Corrosion Under Insulation—Primary Findings J33, J207, J265 Corrosion Under Insulation—Secondary Findings C25 J262 Dust Explosion Hazards—Primary Findings C18, C37, C39, C63, C70, C75 J75, J79, J83, J95, J102, J128 Dust Explosion Hazards—Secondary Findings C20 J152 Facility Siting—Primary Findings C11, C45, C72, C73, C74 D7 J119 S1, S12, S16, S17

Utilities 375 In utility plants, generally the highest required steam is generated and then this stream can be converted to all other steams with lower pressure and temperature by injecting a suitable amount of water in the high tempera-ture/pressure steam. 17.14.3 Cooling W ater Circuit Cooling water is provided for plants for cooling of pro- cess streams and units. The source of cooling water could be ground water, surface water, sea water or even treated waste water. The treatment of cooling water depends on the water analysis and is beyond the scope of this book. However, it could be said that the treatment comprises injection of different chemicals into recirculated cooling water. Therefore P&IDs of a cooling water preparation system are generally several chemical injection systems. A schematic of a cooling water circuit is shown in Figure 17.20. A glycol heat transfer utility system could be consid- ered as an “upgraded” version of steam or cooling water systems. In some process plants a glycol heat transfer utility is needed rather than a steam system or even a cooling water system. In plants, heating glycol media and cooling glycol media circuits are mating systems (Figure 17.21). As the circuit is closed, expansion drums are needed to handle the expansion of liquid after increasing the temperature. 17.14.4 Natur al Gas Preparation System Natural gas can be supplied from the natural gas pipeline. In hydrocarbon industries natural gas may also be gen- erated within the plant and it can be supplied from the plant itself. If natural gas is produced within a plant as the main product or by‐product, it is more economical to use it, otherwise natural gas can be brought from a third party, mainly through a pipeline. If natural gas is produced in a plant with different qual- ities it is better to mix them together before using them as a utility to ensure the constant quality of the natural gas toward the utility system. In these cases there could be a “mixed gas drum” to mix all different natural gas streams within the plant. The usage of natural gas can be split into two main applications: for burning and other applications. The natural gas for burning is named “fuel gas” and the natural gas for other applications is named “utility gas. ” Fuel gas is uniquely used in gas burners.Utility gas can be used for blanketing, purging (e.g. flare header) and other applications. 17.15 Connection Between Distribution and C ollecting Networks It has been shown that there are some utilities that require a collection system. In such cases, the combined distribution and collection networks could be set up as seen in Figure 17.22. Raw waterMake-up water treatmentCooling towerCWR CWS Cooling water pumpFigure 17.20 Cooling w ater circuit. Air cooler PumpExpansion drumGlycol heater Figure 17.21 Glyc ol circuit pair.

Application of Control Architectures 273 the function of an FB control loop versus FF type. See the example below (Table 14.5) about warming up water in a tank using a coil of steam. Table 14.4 shows the two different modes of tempera- ture control of a liquid in a tank. The temperature of the liquid in the tank is held by a steam heating coil. The FB control system measures the temperature on the discharge line, compares it to the set point and then adjusts a control valve to regulate the flow on the steam line to the vessel. Around 90% of control loops in indus - try operate on FB. In FF control, the sensor is located on the feedline to the tank. The control engineer must use a mathemati-cal formula to include all process variables, predict and control the outlet temperature based on the fluctuating input and then adjust the control valve on the steam line accordingly. This is a true proactive approach, which leaves no room for error, and for this reason most people prefer to have the backup of FB control as well. So, you very rarely come across a pure FF control system. The combination of FF with FB control is far more popular. In the example, you can see the difference between FB and FF in the P&ID document: ●There is no difference in the location of control valves. ●If you want to control temperature, the sensor should be a temperature sensor in an FB control loop, but this is not the case in an FF control loop. The parameter being controlled is not always visible from an FF loop. ●In an FB loop, the sensor is located downstream of the fluctuation while in an FF loop, the sensor is located upstream of the fluctuation. Hint: this is different from what some people mistakenly say: “in an FB loop, the sensor is located downstream of the equipment, while in an FF loop the sensor is located upstream of the equipment. ” even though this interpretation could be true in plenty of cases. When reading a P&ID, the way to differentiate between an FF and an FB loop is to see where the sensor is located. If it is located on the fluctuating stream then it is an FF system. An FB loop will have the sensor located on the resultant stream. 14.6.2 Single‐ versus Multiple‐L oop Control Single‐loop control is the default mode for system con- trol. In other words, whenever you can, keep it simple and don’t overcomplicate the control system. As we have discussed, single‐loop control may be in the form of FB or FF control. Multiple‐loop control can have a number of different architectures: ●Cascade control ●FB + FF control ●Ratio control ●Selective control ●Override control ●Split range and parallel control It is worth mentioning that each control loop in a multi‐loop control architecture can be separated from the other control loop(s) and work as a single independ-ent control loop, if this is desired during the operation of a plant. Such capability is generally for the control room operators.Table 14.5 Schema tic of feedback versus feedforward control. Feedback control Feedforward control STMSPTC STMInput(s)TC Sensor on: resultant stream (generally downstream of equipment)Sensor on:fluctuating stream (generally upstream of equipment) Control valve on:stream affects the parameter that has sensor on itControl valve on:stream affects the parameter that has sensor on it The majority of control loops in process plants Not very common in industry and very rare as standalone control for a piece of equipment (if needed they are used in the combination form with feedback control)

112 PROCESS SAFETY FOR ENGINEERS: AN INTRODUCTION exposure limits (PEL), set by OSHA, are used to evaluate long-term worker exposure to chemicals. It should be noted that exposure levels me ntioned in the previous paragraphs are based on defined time exposure time periods. In a wo rkplace; however, it may be important to know at what level chemicals may have an immediate effect or at what levels no adverse effects are expected. A more recently developed system for classi fying chemicals was developed by the United Nations. (UN) The Globally Harmonized System of Classi fication and Labelling of Chemicals (GHS) was developed over decades with the suppo rt of many countries and stakeholder organizations with expertise from toxicology to fire protection. Th e intent is to have a single, globally harmonized system to address classifi cation of chemicals, labels, and safety data sheets. This harmonization supports hazard comm unication and facilitates international trade in chemicals. The GHS classifies materials by: physical hazards, including flammability and reactivity, health hazards, including toxicity and carcinogenicity, and environmental hazards, including to the aquatic environment. To make it more complex, multiple toxins ma y be present in a workplace and thus may be involved in a single exposure incident. Each one may have different concentration criteria. It is not appropriate to assume that the lowest conc entration criteria will apply to the chemical mixture. This may underestimate the hazard. The U.S. Department of Energy (Baskett 1999) and others have recommended an “additive” approach (which is similar to Le Chatelier’s rule). Chemicals can cause other health impacts. in a ddition to toxic properties, for example, by displacing air and thus reducing the oxygen level. Air is normally 21% o xygen. Effects of oxygen depletion below that level are listed in Table 6.2. Immediately Dangerous to Life or Health (IDLH) means any condition that would interfere with an individual's ability to escape unaided from a permit space and that poses a threat to life or that would cause irreversible adverse health effects. (NIOSH) IDLH values are based on a 30- minute exposure. Threshold Limit Value (TLV) – The exposure level of a chemical substance to which a worker can be exposed day after day for a working lifetime without adverse effects. (ACGIH)

392 PROCESS SAFETY FOR ENGINEERS: AN INTRODUCTION CCPS 2008, Incidents That Define Process Safety , Center for Chemical Process Safety, John Wiley & Sons, Hoboken, N.J. CCPS 2008, Management of Change for Process Safety , Center for Chemical Process Safety, John Wiley & Sons, Hoboken, N.J. CCPS 2013, Guidelines for Managing Process Safety Risks During Organizational Change , Center for Chemical Process Safety, Jo hn Wiley & Sons, Hoboken, N.J. HMSO 1975, The Flixborough Disaster – Report of the Court of Inquiry , Her Majesty’s Stationery Office. HSE, “Chemical Information Sheet No CHIS7, Organisational change and major accident hazards”, https://www.hse.gov.uk/pubns/chis7.pdf. OSHA 2009, https://www.osha.gov/laws-reg s/standardinterpretations/2009-03-31-0.

CONSEQUENCE ANALYSIS 279 Table 13.4. Input and output for pool spread models Input to determine spill rate Input for materials Input for physical characteristics Tank pressure Liquid height Hole diameter Discharge coefficient Density VLE data Heat capacity Heat of vaporization Liquid density Emissivity Viscosity Ground density and thermal conductivity Ambient temperature Wind speed Solar radiation Output: The radius or radial spread velocity of the pool from which the total pool area and depth is determined. Aerosol Models . The fraction of released liquid vaporized is a poor predictor of the total mass of material in the vapor cloud, because of the possible presence of entrained liquid as droplets (aerosol). Aerosol and rainout models provide estimates of the fractions of the liquid that remain suspended within the cloud and the fraction reaching the ground. Aerosols may form through two mechanisms: mechanical and thermal. The mechanical mechanism assumes that the liquid release occurs at high enough speeds to result in surface stress which causes the liquid phase to breakup into small drop lets. The thermal mechanism assumes that breakup is caused by the flashing of the liq uid to vapor. At low degrees of superheat, mechanical formation of aerosols dominates; at higher degrees of superheat, a flashing mechanism dominates. Several methods exist to calculate aerosol formation and rainout and ongoing research projects are studying these, but this is still an area of significant uncertainty. Aerosol entrainment has very significant effects on cloud dispersion that include the following. The cloud will have a larger total mass. There will be an aerosol component (contributing to a higher cloud density). Evaporating aerosol can reduce the temperature below the ambient atmospheric temperature (contributing to a higher cloud density). The colder cloud temperature may cause additional condensation of atmospheric moisture (contributing to a higher cloud density). Taken together, these effects tend to signific antly increase the actual density of vapor clouds formed from flashing rele ases. The prediction of these effects is necessary to properly initialize the dispersion models. Otherwise, the cloud's hazard potential may be grossly misrepresented. Several different approaches can be used to address rainout. One a pproach is based on the elevation and orientation of the release an d the jet velocity, the amount of rainout of aerosol and the resultant mass of material in the cloud can be estimated using the settling velocity of the droplets. The amount of moisture in the ambient air should be included in these considerations. All these steps were shown in Figure 13.4 in simplified form.

RISK ASSESSMENT 323 All potential outcomes for each weather condition for each leak size for each scenario There will be a frequency associated with each outcome for each weather condition for each leak size for each scenario. All these combinations lead to the complexi ty. A QRA can be simplified by selecting a smaller number of combinations. This is often done as a first step and followed with more detailed QRAs focusing on higher risks. The basic steps of risk analysis as defined in Chemical Process Quantitative Risk Assessment (CPQRA) are as follows. 1. Define the potential event sequences and potential incidents. 2. Evaluate the incident outcomes (consequences) using tools such as vapor dispersion modeling and fire and explosion modeling. 3. Estimate the potential incident frequencie s using databases, faul t trees, or event trees. 4. Estimate the incident impacts on people, environment, and property. 5. Estimate the risk by combining the potential consequence for each event with the event frequency and summing over all events. A QRA can be supported using spreadsheets. As discussed in Section 14.9, software tools are available to conduct QRAs. Th e risk estimate resulting from a QRA is presented in terms of individual risk or societal risk. A combination of the two provides a more complete picture of the risk. 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. (CCPS Glossary) Societal Risk - A measure of risk to a group of people. It is most often expressed in terms of the frequency distribution of multiple casualty events. (CCPS Glossary) Individual risk expresses the risk to a single person in a single location exposed to an incident or all the incidents. It is sometimes referre d to as location specific individual risk (LSIR) to highlight this point. For example, the total indi vidual risk to an individual working at a facility is the sum of the risks from all potentially ha rmful incidents considered separately, i.e., the sum of all risks due to fires, explosions, toxic ch emical exposures, etc., to which the individual might be exposed. Individual risk is typically expressed as the frequency of fatal injuries per year. Individual risk is graphically displayed as risk contours of 10-6, 10-7, an 10-8 per year on a plot plan as shown in Figure 14.8. Individ ual risk appears simple, but it can be complex to interpret. Assumptions on ignition likelihood can greatly affect contour size and criteria suggested in regulations (i.e. toxicity concentr ations) are suitable for emergency response but generally overstate fatality risk.

Piping and Instrumentation Diagram Development 48 engineer will ask the mechanical engineer of the manu- facturing company to design the equipment to withstand the high and low structural integrity levels. Therefore, these levels are named “design” values, too. However the word of “design” in this context only refers to the integ-rity of the equipment and not the operating features of the equipment. The high and low structural integrity leve ls are also known as “mechanical design parameters. ” When a parameter goes beyond the high or low struc - tural integrity level, there is a potential of immediate danger as a result of an explosion or collapse of the pro-cess item or instrument. Process parameters are arbitrarily split into four areas: normal operation, mild upset, severe upset, and immedi-ate danger. These area (bands) are shown in Figure 5.5, and their features are outlined in Table 5.1. By defining all process parameters for an item and propagating them for all parameter levels, a matrix will result that maps the operation of the item during lifetime of a plant. Table 5.2 shows an example of parameter defi-nition matrix for a warm lime softener. Process parameters on each level may have a specific name. They are not always named, for example, high‐high level of pressure. These names are shown in Figure 5.6. A naming system is defined for each parame-ter, and these parameters are discussed more fully in the following discussion. 5.3.2.1 Pressure Levels Pressure and temperature are the most important parameters for the structural integrity of all process items and instruments. Design pressure is the value the process engineer needs from the mechanical engineer for the design of the structure of the equipment. However, the mechanical engineer of the fabricating company cannot necessarily follow the process engineer’s request because of many limitations, including the standard thickness of the Immediate danger Immediate dangerSevere upset Severe upsetMild upset Mild upsetNormal operating bandHigh str uctural integ rity level High–high le vel High le vel Normal le vel Low le vel Low–low le vel Low structural integ rity level Figure 5.5 Main oper ation bands. Table 5.1 The f eatures of main operation bands. Range Process goal Equipment functionality Equipment integrity Normal operationHigh leveltolow levelProcess goals are met Equipment is fully functional and on its optimum window of operationEquipment is intact Mild upset High‐high leveltoHigh levelorlow‐low leveltolow levelProcess goals are not precisely met Equipment is fully functional but not on its optimum window of operationEquipment is intact Severe upset High structural integrity leveltoHigh‐high levelorLow structural integrity leveltolow‐low levelProcess goals are not precisely met; product could be off‐specification or may not have product at all; hazardous material may be producedEquipment is not fully functional; may not function at allEquipment is intact Immediate dangerBeyond structural integrityProcess goals are not precisely met; product could be off‐specification or may not have product at all; hazardous material may be producedEquipment is not fully functional; may not function at allEquipment explosion or collapsing, release of gas, vapors, or liquids to environment

EVIDEN CE IDEN TIFICATION , COLLECTION & M ANAGEM ENT 147 • A high-quality process safety information (PSI) package, including process hazards assessments (PHAs), is uniquely valuable to the investigation team. Unfortunately, the PSI package may have been partially damaged or ev en destroyed in the incident. It is good practice to maintain a backup duplicate package in a less vulnerable location. Alternatively, the information may be available on the company intranet. In some cases, the information may be more limited and the team will need to work with the data available. In most cases, it is best for the team to work with photocopies of paper documents (such as check sheets, permits, recorder charts and alarm printouts) to avoid damage, alteration, or loss of the originals. In addition to the data sources typi cally available within the facility or organization, other sources of information for the investigation team may include: • News media video footage • Video footage from nearby business security cameras • Social media content • Contacts with other manufacturers with similar processes • University research organizations • Proprietary databases such as those maintained by insurance carriers • Freedom of information document access to government records • Former employees of contract maintenance companies who have personal experience (but not necessarily any vested interest) in the unit of interest • Transcripts of police and other emergency service communications 8.2.2 Physical Evidence and Data Physical data can provide a source of valuable information for investigators. When examining physical data, typical items and matters of interest include: • Fractures, distortions, surface defects/marks, and other types of damage to tanks, vessels, valves, piping and other process equipment • Blast damage • Items suspected of internal failure or yielding

E.46 This is the Last Place I Thought We’d Have an Incident |339 No person or group took responsibility for sabotage, which norm ally occurs. Could the sabotage theory have been advanced to enable workers, m anagers, and the governm ent as an excuse for not fulfilling their safety responsibilities ? If the cause was not sabotage, then the pump head had clearly been short-bolted during a prior maintenance activity, perhaps accepting the short-cut rather than cleaning out and re-tapping the bolt holes in the valve body. Were there other exam ples of normalization of deviance in plant m aintenance activities? The plant circulated a survey asking em ployees whether they felt the incident was caused by sabotage or safety failure. Did employees feel compelled to select sabotage? The plant was bordered closely on all sides by residences and businesses. How did the plant interact with the community on safety issues? Maintain a Sense of Vulnerability, Understand and Act Upon Hazards/Risks, Combat the Normalization of Deviance. E.46 This is the Last Place I Thought We’d Have an Incident An inorganic powder used as an oxidation catalyst was being isolated for disposal. The powder had been filtered from the reaction mixture and washed with clean solvent, and the solvent was being removed by sweeping the filter with warm inert gas through a chilled water condenser. During the drying cycle, an exotherm ic reaction occurred in the filter that dam aged it. The mix of inorganic powder and organic solvent exited the filter and found an ignition source. The resulting overpressure caused some dam age, and the fire was quickly extinguished by the fire suppression system. Fortunately, no injuries resulted. The investigation team (Ref E.12) found that some years earlier, reactivity testing had identified a reaction between the Actual Case History

6.2 Assess the Organization’s Pr ocess Safety Culture |219 pressure for participants to not deviate from the group-think. Take care to avoid forming groups whose participants do not get along with each other. People in such groups may offer contrary opinions out of habit rather than expressing true feelings. Moderators should be alert to both possibilities and make the necessary adjustments. Moderators need to make all participants feel safe and not pressured to answer in any specific way. They should inform participants how their comments and the overall session results will be summ arized and reviewed. Moderators need to stress that no opinions expressed are wrong, and that all participants should respect others’ opinions, even if they disagree, while collecting the contrary opinions. A m oderator m ay say, “If you have a totally different experience or opinion than the rest of the group, I need to hear it. Your view represents others who are not here today to support your view. I hope you will have the strength to speak up.” The moderator should offer praise for the first contrary opinion with a comment like, “Thanks for sharing. I knew you all cannot be agreeing about this. Can we hear m ore?” Plan Focus Group Sessions. The number of focus groups needed depends on the size of the site and the number of functions and levels in the site’s organization. As noted above, focus groups should comprise participants of similar levels. Therefore, there will likely be one focus group of senior managers, two or more of m iddle m anagers, and increasing numbers of groups at lower levels. Each focus group should be designed around specific goals. Groups at the same level may have different goals. Most literature recommends 6 to 10 participants per focus group, plus one moderator, plus possibly a note-taker. In larger groups, more vocal participants can drown out the input of others. The potential for side conversations also increases. Each

114 | 4 Applying the Core Pr inciples of Process Safety Culture Definition of Ethics Ethics is defined as (Ref 4.5): The study of the general nature of morals and of the specific moral choices to be made by the individual in his relationship with others. The rules and standards governing the conduct of the m em bers of a profession. Any set of m oral principles or values. The moral quality of a course of action. For the purposes of this book process safety culture was defined in Chapter 1 as: “The pattern of shared written and unwritten attitudes and behavioral norms that positively influence how a facility or company collectively supports the successful execution and improvement of its PSMS, resulting in preventing process safety incidents.” Clearly, ethics and process safety culture are closely related concepts. They share an almost total reliance on how people feel about certain aspects of their jobs and how they behave. Each also shares a reliance on rules, standards, procedures, and m anagements systems. However, several things differentiate them . For exam ple, morality is the basis for ethical behavior, while process safety culture is based on the human value of preserving life and property. However, like ethics, a positive safety culture does include moral behaviors that are fair, honest, and open. An interesting question then emerges: does the process safety culture drive the ethical behavior of an organization or does the ethical behavior drive the culture? Clearly, good or bad ethical behavior of influential persons can affect the culture for good or bad. Likewise, good or bad culture can affect the ethics of an • • • •

199 Ship and use the intermediates ra ther than the raw materials. 8.10.3 Transportation Mode and Route Selection Select a transportation mode to mini mize risks to the extent practicable. Drums, ISO containers, tank trucks, ra il tank cars, barges, and pipelines offer tradeoffs in throughput/invento ry, container integrity, size of potential incidents, distance from supplier or customer, and the frequency of incidents. Barges ma y have fewer acci dents than tank trucks, but the environmental and economic consequences of a major release from a barge in a major wate rway, particularly one that supplies potable water to surrounding populati ons, may be severe enough to make the tank truck shipments a more attractive choice. The transportation mode used will affect the shipper’s options with regard to the selection of the shi pment routing. Using truck shipments instead of rail to ship drums, ISO co ntainers, and tank trailers, may allow the shipper to specify highway routes rather than rail to avoid high risks. Shipment via highway has more route options than rail which are fixed and has less options from a given pa iring of point of origin and final destination. The time of day and dura tion of travel is also easier to specify with truck shipments than with rail, as security escorting, if it deemed necessary, is usually easier with trucks than rail shipments. Railroads choose the routing of rail ta nk cars and shippers have little or no control to select routings th at represent lower risks. However, large amounts of hazardous chemical s can be shipped via rail through densely populated urban areas. In July 2013, a runaway train carrying flammable Bakken crude oil derailed in the center of the Quebec town of Lac-Megantic. The resulting fire and explosion killed 47 people. Alternative routings and improved tracking of rail shipments by the railroads has helped reduce hazards such as long-term storage of tank cars containing toxic or flammable materials on spurs adjacent to residential areas. The routing of barge shipments is e ssentially fixed by the location of the shipper and the receiver and the waterway(s) that connect them, and there is generally no choice of routin g via pipeline. Data on accident rates by mode and references are given by CCPS (Ref 8.14 CCPS 2008) and can be used to select the safest shipping mode.

106 Figure 6.1 – A traditional methyl acetate process using separate reaction and distillation steps (Ref 6.19 Siirola)

Piping and Instrumentation Diagram Development 4 P&IDs are used by operations personnel, control tech- nicians and engineers, maintenance personnel, and other stakeholders. One main use of P&IDs is for maintenance personnel to initiate lockout–tag out actions. This con-cept will be discussed in Chapter 8. Some individuals in the operation of a process plant may consider to not know about the development of P&IDs because it is “not their business. ” However, this approach is not completely correct for different reasons. For example, a considerable number of items on P&IDs are things inherited from the design and development stages of the P&ID; therefore, to have a good understand-ing of the P&ID, its development needs to be understood. 1.2 What Is a P&ID? A P&ID is the focal drawing in all process plants. P&IDs may be named differently by each company; however, P&ID is the most common. P&IDs can also be called engineering flow drawing (EFD) or mechanical flow di agram (MFD). A process plant can be an oil refinery, a gas processing plant, a food processing plant, mineral‐processing plant, pulp‐and‐paper plant, pharmaceutical or petrochemical complexes, or water and wastewater treatment plants. All the plants that make non‐discrete “products” use P&IDs to show their process. For example, in an automo-tive factory, they make discrete things (e.g. cars), so they do not use P&IDs. Some other industries that traditionally are not classi- fied as process industries have started to develop and use P&IDs. One such example is the HVAC industry. P&IDs can even be used to show the system of some machines that do some processing of some sort. P&ID is a type of engineering drawing that describes all the process steps of a process plant. It basically is a process plant on a paper. A P&ID is a schematic diagram of pipes, process equipment, and control systems by a set of predecided symbols with no scale and no geographical orientation. Equipment symbols are typically a side view of the real shape of the equipment, and if possible, are shown relative to their actual sizes. Different types of lines on the P&ID represent pipes and signals. However, the length of lines do not represent the real length of pipes or signal carriers (e.g. wires). There are, however, a set of P&IDs that are shown in plan view rather than in side view. They are generally drawings that only show piping. Drawings, such as util-ity distribution P&IDs, are shown as schematics but in plan view (Figure 1.3). Different types of P&IDs will be discussed in Chapter 4. 1.3 P&ID Media P&ID is handled in two different platforms: paper media and electronic media. P&IDs used to be outlined on paper. We are now in a transition state and moving to electronic P&IDs. Whether there will be paper P&IDs will still needs to be determined. P&IDs are published on paper. The paper size is differ - ent for each company because P&IDs are not drawn to scale. The only criteria in choosing a paper size for a P&ID are the ease of reading its content and the ease of handling.Loop diagram Alarm table Piping DS Piping model IsometricsP&ID Mechanical DSProcessCalculations Process MechanicalPipingI&C Figure 1.1 The P&ID is used b y other groups to prepare other project documents. ProcessPlant modeling PL&P P&IDs Operation I&C OthersAs a reference in plant during operation Loop diagramsI&C Mechanical Operation PL&P Electrical CSA Figure 1.2 The P&ID is a documen t consolidated and used by different groups. Sometimes an “abridged” version of a P&ID is created for the purposes of operation. Some people try to create an “Operational P&ID’s” because they claim that the P&IDs they receive (by the time the plant is in operation) has many features related to the design phase of project that are not relevant to the plant’s operation. The concept of an Operations P&ID is not accepted by all industry professionals.Do you mean that can I draw a P&ID for my washing machine, vacuum cleaner, or even coffee maker? Yes, you can, and I did it as practice. However, it is not helpful during the design stage of the “project” or for household repair specialists.

460 PROCESS SAFETY FOR ENGINEERS: AN INTRODUCTION expected to fulfill their individual proce ss safety responsibilities. For example, operators and mechanics fully follow and pr operly complete procedural checklists, engineers follow engineering practices, an d managers diligently consider process safety decisions and resources. People are pre-occupied to identify the next failure or deviation - Everyone is vigilant. The organization maintains a high awareness of process hazards and their potential consequences, maintains a sense of vulnerability, and is constantly vigilant for indications of system weaknesses that might foreshadow more significant safety events. Deviations are not tolerated, instea d they are investigated, and actions taken to address them. The organization values learning vs. blaming - The organization learns from smaller problems and views failures as opportunities to improve, not blame. Everyone wants to learn. All involved wa nt to improve their own and the overall facility performance. Expertise is sought and valued. Personnel attend training and have coaches to support on the job learning. Investigations and audits are viewed as opportunities to learn and improve. Employees feel comfortable to ‘speak up’, point out problems, allow for dissenting views - Open communication is encouraged. Healthy communication channels exist both vertically and horizo ntally within the organization. Vertical communications are two way – managers listen as well as speak. Horizontal communications ensure that all workers ha ve the information. The organization emphasizes promptly observing and report ing non-standard conditions to permit the timely detection of weak signals that might foretell safety issues. These examples of what a good process safety culture looks like are similar to the characteristics of High Reliability Organizations (HRO). HRO’s are organizations with strong safety performance in high- risk environments. Examples include U.S. Navy aircraft carriers, U.S. forest firefighters, Federal Aviation Adminis tration (FAA) traffic controllers, and also some private companies. They all possess the following characteristics. (Weick 2001) Preoccupation with failure – Personnel are always alert to early warning signs and envision where the next potential failure will occur. Reluctance to simplify- Personnel are not quick to acce pt simple explanations to anomalies, and probe for deeper understanding. Sensitivity to operations – The organization recognizes and understands how different elements of an organization in teract/impact the front line and others in safety critical roles. Commitment to resilience – The organization not only is able to prevent failures but also recover quickly from them. This requires strong and fast learning capabilities. Deference to expertise – senior decision makers recognize that lower level employees have relevant knowledge and ex pertise to address problems and make them feel comfortable for speaking up.

KEY RELEVANCE TO PROCESS PLANT OPERATIONS 77 3.4.2.6 Type of Operation/Steady-State vs. Transient Chemical processes may be operated in continuous, semi-continuous, or batch modes of operation. Batch operat ions, in particular, have a history of incidents precisely because: (a) th ey are frequently operated across a range of operating conditions within a single batch, and/or (b) reactions frequently have the potential to run away, as cited in Example Incident 3.17. Example Incident 3.17 – Polystyrene Reactor A polystyrene production facility ha d a history of runaway events that resulted in emergency dumping of reactor contents, and odorous fumes of unreacted styrene disp ersing through the surrounding residential area. The fumes were not life threatening, but they were objectionable enough and present often enough that the local authorities demanded a stop to the events. Given that the normal procedure for handling a runaway was to dump/vent the reactor contents intentionally to the atmosphere, a more comprehensive hazard management process was required. First, a HAZOP was conducted to identify all the (many) possible causes of a runaway, then a fault tree was used to quantify the relative importance of each. The site team developed a proposed fix to the problem (installing a pre-release vent pot to capture an d cool the vast bulk of the dump), but the fault tree indicated that in 80% of the routes to a runaway, the vent pot would be undersized. Ultimately, the solution was to rely on procedures and systems to prevent the runaway rather than manage it. Lessons learned in relation to abnormal situation management: Knowledge and Skills: The team recognized the situation, which in this case was routine and not abnormal. They then developed a proposed fix and conducted an analysis of their proposal. This was excellent teamwork. Procedures: The team created procedures to prevent the runaway, which is normally a better option than to have to manage the consequences.

THE UPSTREAM INDUSTRY 33 Barriers can fail and are not perfectly reliable and this is often explained using Reason’s swiss cheese analo gy (e.g., IOGP, 2018a) – a slice of cheese representing the barrier and the holes representing its pot ential failure on demand. In principle, therefore, multiple barriers are needed to give confidence that a threat pathway is always terminated or adequately mitigated. Bow ties can be used to identify and share important barriers during design and operation, and if an incident occurs, to explain which barriers failed. 2.8 OVERVIEW OF INTERNA TIONAL REGULATIONS Process safety regulations exist for most offshore regions. A trend has been to separate process safety and environmenta l regulators from government functions promoting the development of offshore resources. Example safety regulators (some with environmental responsibilities as well) include the following. ●BSEE (Bureau for Safety and Environmental Enforcement) in US ●HSE (Health and Safety Executive) in UK ●PSA (Petroleum Safety Authority) in Norway ●ANP (National Agency of Oil, Gas and Biofuels) in Brazil ●C-NLOPB (Canada-Newfoundland and Labrador Offshore Petroleum Board) in Newfoundland, Canada ●NOPSEMA (National Offshore Petroleum Safety and Environmental Management Authority) in Australia These and other regulators meet periodically and publish various technical studies and incident statistics at the International Regulators Forum website (irfoffshoresafety.com). There are two broad approaches to regula tion: prescriptive and goal-based. This differentiation, and their historical development, are discussed fully for downstream applications in CCPS (2009). Historically, all safety regulations were prescriptive, usually developed in response to an incident, and where the regulation specified the nature of the safety remedy required (Broadribb, 2017). An example of a predominately prescriptive regulatory appr oach for upstream is seen in the US, although elements of goal-b ased appear in the SEMS regulation it also spells out exactly what must appear in the management system and thus is partly prescriptive. Examples of the goal-based regulatory ap proach for upstream are given by the UK, Norway and Australia – all for offshore. Canada, in 2020, is in transition from a prescriptive to a goal-based approach. These approaches are not pure; goal-based regulations include some prescription a nd prescriptive regulations include some goal-based aspects. Major downstream incidents in the US (e.g., vapor cloud explosion in Pasadena, Texas in 1989) resulted in OSHA developing the Process Safety Management (PSM) Regulation (OSHA 1910.119) which was phased in over several years

152 Human Factors Handbook Some examples are: • A fitter must correctly fit a seal to a pump on three separate occasions. • Process control room supervisors ma y need to successfully manage a simulated emergency response in thr ee tests (randomly chosen out of a possible set of 10 scenarios), and disp lay appropriate skills such as task delegation, and effective communication. Assessment of competency is concerned with individual’s progression across proficiency levels (awareness, basic application, skillful application, mastery and expert), until the highest level of proficie ncy (appropriate for specific job role) is reached. Achieving a certain level of profic iency (including expert proficiency), does not mean that no further development and/or assessment is required. Individual competency should be kept up to date, and therefore frequently reassessed (see section 14.4 for more informatio n on competency re-assessment). 14.3.2 Select assessment methods Various methods can be used to aid competency assessment. The chosen method should be suitable for the assessment of the competency in question. For example: • Assessment of knowledge-based comp etency may use a series of “talk- through” questions or a multiple-choice quiz. • Assessment of skill-based competency may be assessed via simulation exercises or a “show me” technique. Examples of assessment methods [59], their suitability for different types of performance, advantages, disadvantages, an d issue to consider are listed in Table 14-1.

SUSTAINING PROCESS SAFETY PERFORMANCE 449 Figure 22.2. A shower of foam debris after the impact on Columbia’s left wing. (CAIB 2003) Over the previous decade, NASA was placed under severe pressure to reduce costs. The focus on measuring costs resulted in losing about 40% of its budget and workforce. Part of the response was for NASA to hand over much of its operational responsibilities to a single contractor, replacing its direct involvement in sa fety issues with a more indirect performance monitoring role. NASA managers continued to ta lk about the importance of safety, but their actions sent the opposite signal. Despite the cutbacks, personnel felt pressu re to keep the Space Shuttle program on schedule, particularly to complete the Internat ional Space Station (ISS). The uncertainty over the long-term future of the program resulted in reduced investment, with safety upgrades delayed or deferred. The CAIB found that the in frastructure had been allowed to deteriorate, and the program was operating too close to too many margins. Technically, the cause of the incident was the failure of the foam insulation at the bipod attachment. No non-destructive testing (NDT) of hand-applied foam was carried out other than visual inspection at the vehicle assembly building and at the space center, even though NDT techniques for foam adherence had been succe ssfully used elsewhere. The CAIB concluded that too little effort had gone into the understan ding of foam fabrication, adhesion, and failure modes. Culture also played a key role in the incident. In spite of cutbacks and deadline pressures, the organization continued to pride itself on its “can do” attitude, which had contributed to former successes. This enabled the phenomenon known as “Normalization of deviance”. The failure of the foam without significant conseq uences was observed so many times that it

264 INVESTIGATING PROCESS SAFETY INCIDENTS rotation and to open if the handle is turned counterclockwise. Deviating from normal convention, expected actions, and established habits can be an underlying cause of human error. 2. Over time, minor modifications an d changes can individually or collectively cause human performance problems. A fourth pump was added to a group of three existing pumps. In the field, the fourth was added in sequence alongside pump C. The arrangement was A- B- C- D. However, there was no room on the control board for the new switch to be added after the “C” switch, so it was added beside the “A” switch where there was space ( Figure 11.2 ). Consequently, in the control room the corresponding switches were configured in D- A- B- C sequence. In an emergency, the operator could easily mistakenly flip th e first switch (the new “ D ” s w i t c h ) thinking it is the fam iliar “A” switch in that position. This ergonomic trap proliferates as time goes on and changes are made without consideration for operator habits, tendencies, and normally expected actions. Figure 11.2 Example of Poor Pump and Switch Arrangement

26. Learning from error and human performance 343 Figure 26-2: The consequences of blame culture To get a fuller account of the incident and people’s actions, the following techniques can be used: • Encouraging people to “freely” desc ribe what happened. For example, why do they think this happened, what were they thinking or feeling, what could they see. • Refreshing people’s memory by: o Providing a short factual description (focused on technical issues) of what happened. o Asking them to show what they were doing or seeing, by taking them to their usual work environment. • Emphasizing that the focus is on learning and future error prevention, not on individual fault-finding. Blame CultureLack of accountability Lack of trust Impaired relationships Employee disengagementPrevents problem solvingUnderreporting

Process Safety Culture Learning Objectives The learning objective of this chapter is: Understand the concept of Process Safety Culture. Overview Process safety culture, put succinc tly, is “How we do things around here” or “How we behave when no one is watching.” Process Safety Culture - the common set of values, behaviors, and norms at all levels in a facility or in the wider organization that affect process safety. (CCPS Glossary) Process safety culture weaknesses have been id entified through investigations such as in the Space Shuttle Challenger and Columbia disast ers and the BP Texas City Refinery Explosion. As seen in these incidents, many of the indi vidual elements of process safety were weak. Process safety culture impacts and is impacted by the process safety management system elements as well as other business ma nagement systems, e.g. financial. It is common for organizations to perform cu lture surveys as a method to determine the current level of culture and then conduct subs equent surveys to monitor improvement based on action taken. This approach was taken fo llowing the BP Isomerization unit explosion in Texas City and the survey approach is docume nted in the Baker Panel report. (Baker 2007) It is not possible to write a policy requiring a good process safety culture or a procedure that tells someone how to achieve it. What woul d be the requirements? It is, however, possible to see a good process safety culture in action . The following are examples of what a good process safety culture looks like. Leadership sets the tone (‘tone at the top”) - Management demonstrates process safety is a priority. They do this throug h their own actions. They are personally involved in process safety. In other words, th ey walk the talk. A literal example of this is when management walks through the pl ant, discusses process safety concerns with operators and follows up on those concerns. Metrics, Organization an d Incentives support strong safety culture - Process safety is at the same level as other busi ness functions. Just as with finance, employee relations, and other functions, process safety is included in top level business management. This means that pers ons with process safety responsibility are included in the meetings and that process safety metrics are included in the discussions. Safety metrics promote st rong safety priorities/behaviors, and discourage excess risk taking. Conduct of operations is valued - The organization clearly defines safety-related responsibilities. Accordingly, employees are provided the resources needed and are

57 volume by 80%, and confining the MIC storage to one area of the plant, which simplified the internal logistics of MIC handling (Ref 3.3 Committee). An additional method to reduce r a w m a t e r i a l i n v e n t o r y i s t o manufacture the hazardous component in a just-in-time fashion, rather than purchasing it in bulk and storin g it. This potentially achieves similar inventory reductions as a just-in-ti me purchasing strategy, with the added benefit of eliminating the a dditional material handling (i.e., transportation container unloading). Consider a chlorination process which uses railcar quantities of purc hased chlorine on a daily basis. A chloralkali plant can generate low pr essure chlorine as needed for the process, with very little in-process in ventory, and whatever inventory is required can be at low pressure (ano ther ISD strategy, i.e., Moderation, discussed in Chapter 5). This has the added benefit of reducing the offsite consequences of a loss of containment (similar to the bromine example above) and provides a supply of caustic soda for either outside sale or reuse/recycle onsite. This example was considered for an actual organic chlorination process which generated hydrogen c h l o r i d e a s a low market value byproduct. The co -generated caustic soda would be used to neutralize and recycle the chlo rine essentially lost as byproduct to high-value-added chlorinated organi c. The worst-case scenario of a release incident for a co-generated facility is nearly an order of magnitude less severe than the original variation. 3.8 PROCESS PIPING All process equipment and units must be connected by piping systems, making the layout of piping within a plant a significant influence on the inventories of materials on-site. In turn, the layout and placement of units and equipment will influence the length of inter-unit and intra-unit process piping and transfer piping systems which link these units. Since the volume of piping is a function of the square of the piping diameter, each additional linear foot of piping length can represent a measurable increase in site material inventorie s. Each 100 feet (30 meters) of piping, with an inner diameter of 2 inches (50 millimeters), adds approximately 16 gallons (60 liters) of liquid to site inventory, and each 100 feet (30 meters) of 6-inch (150 mm) piping adds approximately 150 gallons (568 liters). For the same reason, process piping diameter should also be

5.2 Risk Management-Related Element Grouping |179 However, the investigation should not stop there, and instead continue until the root causes are identified, including the cause for why the illegal or anti-policy act had not been detected and prevented. Indeed, if such acts were comm itted, the trust that m anagement will properly address safety problem s can be broken. Auditing (Element 19) Like audits of any other business practice, PSMS audits serve critical roles in governance and risk m anagement. Process safety audits are independent reviews to determine if PSM Ss are functioning as intended to m anage process risks and to comply with regulations and corporate standards. Com panies and facilities with a strong process safety culture will also use audits to identify opportunities to improve the PSMS. Audits are typically conducted every 5 years, although high-risk facilities may be audited more frequently. Audits also provide a window into the process safety culture of the organization. It is possible, and indeed a good practice, to audit process safety culture specifically. Appendix F provides a list of sample questions that can be incorporated into a culture audit. Audit findings describe the non-conform ances with regulations and standards identified. Some companies ask auditors to recommend means to close conformance gaps, while others prefer auditors to focus only on auditing. The choice of approach depends partly on the company’s legal philosophy and partly on the strength of the culture. In general, if the company has a strong process safety culture, either approach can be successful. However, if the culture is not yet strong, the auditors should not offer recomm endations. This often leads to cosmetic solutions that aim to reduce the number of findings, but that do not fully close the gap. Facilities with strong process culture welcom e audits and encourage their personnel to cooperate fully with auditors.

20 | 1 Introduction safety culture have been shown to consistently have better financial perform ance. There is a strong business case for strengthening and sustaining process safety culture. 1.8 REFEREN CES 1.1 Sielski, M ., The Philadelphia Inquirer, October 25, 2014. 1.2 Schein, E.H., Organizational Culture and Leadership , 3rd Ed., Jossey- B ass, 2004. 1.3 CCPS, Guidelines for Risk Based Process Safety, American Institute of Chemical Engineers, New York, 2007. 1.4 International Atomic Energy Agency (IAEA), Safety Series No. 75 – INSAG-4, Safety Culture, 1991. 1.5 Mathis, T., Galloway, S., STEPS to Safety Culture ExcellenceSM, Wiley, 2013. 1.6 National Aeronautics and Space Administration, Columbia Accident Investigation Board Report , Washington, DC, August 2003. 1.7 Rogers, W.P. et al., Report of the Presidential Commission on the Space Shuttle Challenger Accident, Washington, DC , J une 6, 1986. 1.8 J ones, D., Kadri, S., Nurturing a Strong Process Safety Culture , Process Safety Progress, Vol. 25, No. 1, American Institute of Chemical Engineers, 2006. 1.9 CCPS, Process Safety Culture Tool Kit, American Institute of Chemical Engineers, New York, 2004. 1.10 Baker, J .A. et al., The Report of BP U.S. Refiner ies Independent Safety Review Panel , J anuary 2007. 1.11 McCavit, J , B erger, S., Grounds, C., Nara, L., A Call to Action - Next Steps for Vision 20/20 , CCPS 10th Global Congress on Process Safety, New Orleans, 2014. 1.12 Whiting, M. and B ennett, C., The Conference B oard, Driving Toward ‘0’: Best Practices in Corporate Safety and Health , Research Report No. R-1334-03-RR, 2003. 1.13 Hale, A.R., Culture’s Confusions , Safety Science, Vol. 34, No. 1-3 (2000). 1.14 Canadian National Energy B oard, Advancing Safety in The Oil and Gas Industry Statement on Safety Culture , from M earns, K., Flin, R.,

CASE STUDIES/LESSONS LEARNED 197 modifications be implemented. Th is would be a key element of a Management of Chan ge (MOC) system While not contributing significantly to the incident, the action taken by operators to drain the liquid level from the wet gas compressor interstage drum using two steam hoses coupled to the flare header should have been considered a te mporary modification, requiring a formal risk assessment. Again, this would form part of a formal MOC system. 7.2.8.6 Communications Due to the number and the magnitud e of operating problems, a large team was in the control room prov iding assistance. The report states that under such circumst ances, there is an even greater requirement for effective communication, to ensure that contradictory operations are avoided. 7.2.8.7 Training/ Knowledge & Skills The report recommends that st aff be trained to include: “An assessment of their knowledge and competence for their actual operational roles under high stress conditions; and Guidance on when to initiate controlled or emergency shutdowns and how to manage unplanned events including working effectively under the stress of an incident.” Regular training on incident and accident scenarios, particularly involving the use of simulators coup led with some “traps” of unreliable instrumentation, would be beneficial. 7.2.8.8 Learning from Experience The first two recommendations on the HSE report state that: The safety management systems should include a means of storing, retrieving, and reviewing incident information from the history of similar plants. Safety management systems should have a component that monitors their own effectiveness.

17. Error management in task pla nning, preparation and control 195 A successful example of barrier ownership is illustrated by Figure 17-3. Two Pressure Safety Valves (PSV- A and PSV- B) were installed in a gas header to a plant. These valves were lined up through two 3-way valves, one at the PSV inlet and other at the PSV outlet. Before start-up of plant, as part of safety valve checklist (barrier), a verification round was conducted by the supervisor. Th e supervisor observed that the inlet 3- way valve was lined up with PSV A but th e downstream 3-way valve was lined up with the other PSV B. Thus, both the PSVs were non-functional. The line-up of the valves was corrected, and a possible event was averted through barrier ownership. Figure 17-3: Barrier ownership prevented wrong valve line up Wrong line up Correct line up 17.5 Distractions and interruptions 17.5.1 Error-likely tasks and situations It is important to minimize distractions and interruptions when: • A task requires a high level of concentration. • It is important to do actio ns in a specific order. • It is necessary to remember info rmation from earlier in the task. • It is necessary to remember a signif icant amount of information, such as the status of multiple pieces of equipment. • It is a long duration task where it is possible to accidentally miss a task step due to a lapse of memory, especially when fatigued or tired.

26 PROCESS SAFETY FOR ENGINEERS: AN INTRODUCTION RBPS Element 10: Asset Integrity and Reliability Asset integrity and reliability is the systematic implementation of inspections, tests, and maintenance to ensure that equipment, and safety critical devices will be functional for their intended application throughout their life. This includes proper selection of materi als; inspection, testing, and preventive maintenance; and design for maintainability. Duri ng the design stage, potential asset integrity problems can be anticipated and significantly mitigated. Equipment and control systems can be affect ed by harsh environments. Some equipment can be hard to inspect, particularly in remote or offshore installations. Many existing facilities are operating beyond their intended design life and are managing aging issues which can impact asset integrity. RBPS Element 11: Contractor Management Contractor management is a system of controls to ensure that contracted services support both safe facility operations and the compan y’s process safety and occupational safety performance goals. This element includes the se lection, acquisition, use, and monitoring of such contracted services. These controls ensure that contract workers pe rform their jobs safely, and that contracted products and services do not add to or increase safety risks. Contractors are prominent in both operatio ns and maintenance activities. They have specialized knowledge and equipment to enable challenging tasks to be performed safely and e f f i c i e n t l y . I t i s n e c e s s a r y t o a l i g n t h e p r ocess safety program of the company with its contractors to ensure that all aspects are addressed and that everyone knows their responsibilities. RBPS Element 12: Training and Performance Assurance Training and performance . assurance involves practical instruction in job and task requirements and methods. Performance assu rance provides a means by which workers demonstrate that they have understood the traini ng and can apply it in practical situations. Training and performance assurance applie s to operators, maintenance workers, supervisors, engineers, leaders, and process safety professionals. Performance assurance verifies that the trained skills ar e being practiced proficiently. Work is challenging, and a high degree of sk ill is needed to perform tasks correctly. Many incidents identify weaknesses in training and jo b execution as underlyin g causes. Defining the training that is required to perform a task su ccessfully helps underpin a training program. This should include process safety hazards and how to participate in or interpret risk analysis studies, as appropriate. Formal testing of knowledge and skills is an im portant part of this element to assure that participants have understood the material. It includes on-the-job task verification. RBPS Element 13: Management of Change (MOC) MOC strives to ensure that changes to a process do not inadvertently introduce new hazards or unknowingly increase risks. This in cludes identification of a change, review of

204 Human Factors Handbook 17.8 Team briefings Tool Box Talks, Tailgate Meetings (a team briefing at the rear of a vehicle) and other forms of team briefings are a standard part of process operations. They are also an important part of error management. In particular, they can: • Communicate task expectations and objectives. • Communicate task-specific information and knowledge to people, helping to ensure they know what to do. • Provide a forum for people to: o Ask questions and check their understanding of a task. o Challenge the realism of plans and identify potential problems. o Reinforce safety instructions an d the importance of following the safety requirements. Features of a good team briefing are provided in Figure 17-6. This can include highlighting errors that could be caused by unfamiliar tasks, unreliable equipment, and misguided assumptions (e.g., assuming the cause of a fault without checking and without any evidence). Tool Box Talks and briefings communicate safety aspects related to the specific job.

3. Options for supporting human performance 31 Figure 3-3: Strategies for knowledge and rule-based human performance When a process is changed, knowledge can become outdated. Knowledge of process operations and hazards should be kept current by updating both process documentation and training. This should be ensured by a Management of Change procedure as noted in the CCPS “Guidelin es for Risk Based Process Safety” [5]. 3.4.2.2 Job aids Up-to-date procedures and job aids can show the circumstances and conditions where a sequence of actions should be used – it will also outline what these actions are. A logical step-by-step guide or list of clear instructions can help with understanding and carrying out these actions, especially when a person has had previous training and experience. These should be designed to be practical and meaningful to operators as noted in the CCPS “Guidelines for Risk Based Pr ocess Safety” as per the Operating Procedures element [5]. 3.4.2.3 Training and experience Training and operational experience can help people to remember and use their process and procedural knowledge. This is part of the ‘Training and Performance Assurance’ element of the CCPS “Guidelines for Risk Based Process Safety” [5]. See Chapters 10, 11, 12, 13 and 14 for more information on training and performance assessment. Diagnostic, communication & decision-making skills Education in process, system, faults & hazards Workload & fatigue management Information, schematics, decision-making aids & procedures Teamwork, shared situation awareness, co- ordination, clear roles & responsibilities Task & team design (distractions & interruptions) See Chapters 5, 1, 7, and 8 for more information on job aids.

128 7.1 OPERATING PROCEDURES Human intervention, consistent with documented operating procedures (including emergency procedures), is a key layer of protection for any process. Procedures that are not followed due to obsolescence, inaccuracy, unavailability, or difficul ty in implementing often present process safety risks. The Simplification and Minimization strategies can be applied when developing or revisi ng operating procedures. Applying inherently safer techniques to th e design of procedures requires consideration of the following (Ref 7.3 CCPS 2006): Completeness and accuracy : Does the procedure have enough information for the user to perfor m the task safely and correctly? Appropriate level of detail : Has the level of detail considered the experience and capabilities of the users, their training, and their responsibilities? Leaving detail out of a procedure in lieu of relying on user training is a crit ical aspect of writing procedures. Conciseness : Conciseness means elimin ating detail and language that does not contribute to work performance, safety, or quality. Conciseness also means segreg ating “need-to-know" from the "nice-to-know" information. There is a balance between conciseness and appropriate level of detail. Consistent presentation: This element ensure s that the procedure is readily comprehensible. It demands the use of: oA consistent terminology for naming components and operations, with corresponding labels in the field. oA standard, effective format and page layout. oA vocabulary and sentence structure suitable for the intended user. Writing documents at a targeted level of reading comprehension is possible. Administrative control: All procedures should be reviewed thoroughly before use and period ically thereafter. A “Job Cycle Check” is an effectiv e means followed in the industry to ensure personnel are periodically practi cing the procedures, and also helps get feedback on ease of operating with the procedures. CCPS (Ref 7.2 CCPS 2006) includes guidelines for when a procedure is required, as well as an example procedure checklist.

143 11 REAL MODEL SCENARIO: CULTURE REGRESSION “Knowing is not enough; We must apply. Willing is not enough; We must do.”—Bruce Lee, Martial Artist, Entertainer, and Leader The Lamington Oil Company operates in the East Timor Sea, about 500 km off the coast of northern Australia. The company’s offshore rig was built in 2005 in the rush to tap into the speculated USD 50 billion worth of oil and gas in the fields. The rig operates 24 hours a day, seven days a week. The employees on the rig worked 12-hour shifts with two short breaks plus a lunch break. They worked on a demanding 14/14 rotation (14 days on the rig followed by 14 days off the rig), but fatigue risk assessments were not carried out regularly. At the beginning, the rig crew was meticulous about following all written procedures. No short cuts were taken. Equipment maintenance was done as scheduled and all the needed work permits were filed properly. The original rig manager made personal and operational safety top priorities. He made sure that every shift started with a safety meeting where everyone was engaged. But when the rig manager left for an onshore position, his successor had a different perspective. The new rig manager said he was concerned about safety, but it was clear to the crew that he put production numbers as the top priority. Maximum production always was the new goal he constantly stressed to everyone on the rig. The new rig manager’s message may have had some unintended consequences. The rig’s safety officer saw an uptick in minor accidents such as slips and falls. But he didn’t feel compelled to report them. He wanted the safety record of the rig to remain stellar. He just thought he was doing what the boss wanted. The rig’s production and maintenance supervisor were under a lot of pressure to meet his numbers. He began to stray from the The individuals and company in this chapter are completely fictional. Driving Continuous Process Safety Improvement From Investigated Incidents By CCPS and EI © 2021 the American Institute of Chemical Engineers

Piping and Instrumentation Diagram Development 52 High structural integ rity levelLess promising action by operator Mechanical relief action Mechanical relief actionSlS action Alarm Alarm SlS actionBPCS actions More promising action by operat or Less promising action by operatorHigh–high le vel High le vel Normal le vel Low le vel Low–low le vel Low structural integ rity level Figure 5.15 Opera tors’ actions.there are enough provisions implanted in the system that we can consider the operation as a safe operation. However, when the parameters exceed either level, some hazards start to be involved in the operation. Here unsafe is the abridged version of the operation, which could be potentially harmful for the health and safety of people in and around the facility and also the environ-ment. There should be alarms to warn the operator that an unsafe operation imminent. When the parameters reach high‐high or low‐low level, the interlock system (SIS) is activated to prevent harm to the health and safety of the personnel and save the environment from contam-ination. Figure 5.14 shows the level of parameter versus the level of hazard and control. 5.3.5 Par ameter Levels versus Operator Role Why do we need plant operators when we have all of these layers of control? As sophisticated as the control system may be, it is not intelligent (putting aside the concept of artificial intelligence for now). An operator with human intelligence to deal with an out‐of‐control situation is needed to run the equipment. It is this operator who should take remedial action to bring the process back under con-trol and prevent the activation of the SIS. Remember, the SIS involves taking drastic and invasive action, which will interrupt the production process, with consequent loss in revenue for the company. However, it is vital to have the SIS layer built into the control system as a backup because an operator may not make the correct decision. The sole purpose of an alarm is to alert an operator to a process parameter that is out of control and that it cannot be rectified by the BPCS. When an alarm is activated, the operator is expected to take action. It is essential that control system designers afford the operator every oppor - tunity to respond; otherwise the alarm is pointless. However, when a process parameter deviates from its normal operating band, the operator may be stressed and not make the best decisions (Figure 5.15).Table 5.3 Temperature levels. Levels Design consideration Example High structural integritySafety valve set point 120 °C High‐high SIS action 100 °C High Alarm 85 °C Normal BPCS action band between low‐low and high‐high80 °C Low Alarm 75 °C Low‐low SIS action 60 °C Low structural integritySafety valve set point −29 °C Mechanical r elief action Mechanical relief actionSlS action SlS actionAlarm AlarmBPCS actionsHigh–high flowHSl le vel Safety actions Safety actionsControl actionsHigh flow Normal flow Low flow Low–low flow LSI le vel Figure 5.14 Saf ety actions for a flow parameter.

5.4 Worker-Related Element Gr ouping |195 Com puter-based training (CB T) is now common. CBT provides m any advantages in term s of efficiently getting training to those who need it, tracking training, and conducting testing as part of the perform ance assurance activities. Leaders should be aware of the drawbacks of CBT, most notably that if a trainee does not understand some part of the training, there is no instructor to ask for clarification. CBT is also less useful for training that needs to be conducted hands-on, such as performing physical tasks like m aintenance, inspection, and worksite evaluations. If CB T is used for such tasks, it should only be to provide basic familiarity, and be supplemented with in-person instruction and demonstrated proficiency. A recent advance in CB T for process facilities is the use of simulators. These can be particularly useful for training operators on the processes they run. Various deviations can be im posed on the sim ulation, and the operator can gain experience in how to handle them . Sim ulation can also help trainees develop a sense of vulnerability by being allowed to virtually blow up the plant. Whether training is in-person or CBT, it should try to incorporate hands-on elem ents. This could involve group or individual exercises, supervised work in the field, and sim ulators. Physically performing tasks helps people remember what they learned. In certain topics, training cannot cover every eventuality. While som e parts can be learned by rote, other parts require the trainee to develop understanding. For example, when training a supervisor how to prepare a safe work permit, the mechanics of filling out the perm it and filing it can be learned by rote. However, the ability to recognize hazards and determ ine the appropriate safeguards requires developing deeper understanding. The ultimate aim of training is proficiency. It is not acceptable for a mechanic or operator to perform their jobs correctly m ost of the time. Therefore, the target score is 100% for every training

4 • Process Shutdowns 47 Since there are essential project- related planning steps used to ensure that the process equipment is prepared and ready for handovers to the group or groups, this chapter provides a brief overview in Section 4.3 of some guidance for different types of engineering projects and how to effe ctively manage the process safety risks during the transition times. Th ere are several different stages in a project’s life cycle, as well, whi ch have different process safety- related risks associated with them as group handovers occur. Section 4.4 provides a brief overview of the project’s life cycle phases, focusing on the times when the facility is in the transient operating mode. In addition, a process shutdown requires several steps for effectively managing projects: 1. Planning for the projec ts in the shutdown, 2. Preparing the equipment for each project (if there is more than one project), 3. Executing the work safely on the isolated equipment, 4. Commissioning and confirming that the equipment is ready for the operations group, and 5. Safely starting the equipment an d the process unit back up. The two transient operating mode s for a process shutdown are: 1. The shut-down mode (steps 1 and 2), and 2. The start-up mode afterw ards (steps 4 and 5). The transient operating modes before and after the process and facility shutdowns are illustrated in Figure 4.1. Each of these steps is discussed briefly in this chapter, with the process shutdown and its associated shut-down discussed in Se ction 4.5. Safely starting up the process afterwards is discussed in Section 4.6.

EQUIPMENT FAILURE 183 The automatic tank gauging level detector had a history of failing due to sticking and this had not been corrected. The IHLS did not function because a test lever for the switch was not locked in the neutral position. The lever enabled testing of the high-level function, and/or the function of the low-level function (if the low-leve l function was installed) of the IHLS. Failure to lock the lever in the middle position allowed the lever to slip into the low-level test position, disabling the high-level function. Experts were surprised by the severe damage from the explosion, given the low level of congestion at the site. The extent of the damage was such that experts concluded that a DDT occurred. This surprise led to recommendations to do further study of the mechanism for the DDT. Figure 11.4. Breakup of liquid into drops spilling from tank top (Buncefield 2008) The following factors contributed to the DDT: Mist formation as the gasoline spilled over top of storage tank. Normally, a spill of a liquid from a storage tank would be modele d as evaporation from the pool created by the spill. As the gasoline spilled from th e top of Tank 912, liquid droplets formed, and this enabled the transport of air into the vapor cloud (Figure 11.4). (Mists can also increase the hazard of a flammabl e release because they can ignite at temperatures below their flashpoint, although that was not the case in this incident.) Low or no wind causing little dispersion and dilution of the flammable cloud. The lack of wind meant the cloud did not disperse. When dispersion occurs the concentration of vapor in the cloud is redu ced by entrainment of air. At Buncefield this lack of dispersion led to the large cloud with a large portion (or almost all) of it in the flammable zone. Strong ignition source from the pump house. The pump house was near Tank 912 and was completely submerged in the cloud. The ignition in the pump house led to an explosion inside the pump house itse lf, and this explosion created a strong ignition source that also created turbulen ce around the pump house, leading to a strong external explosion and the DDT.

Selecting an Appropriate PHA Revalidation Approach 93 Even if the core analysis can be Updated , the complementary and/or supplemental analyses may warrant the Redo approach. Revalidation team members should be cognizant of any do cumentation or analysis shortcomings in the prior PHA and strive to ensure the Redo remedies those problems. There may be circumstances where a Redo is preferable, not because of gaps or deficiencies in the prior PHA, bu t from a purely logistical standpoint. For example, if a process has experienced a large number of changes (or very significant changes) since the prior PHA, Redoing the PHA may be more time- or resource-effective than Updating the documentation to incorporate each MOC. In other words, there may be so many ch anges to a PHA that starting over from the beginning will take less time and result in a higher quality PHA than searching for each change and ensuring it is ad equately documented. A similar situation may arise if the company has Updated (as opposed to Redoing ) the PHA over several revalidation cycles. Document rete ntion and revalidation logistics (e.g., determining what was reviewed, what is still valid) become increasingly complex as more and more MOC documentatio n is appended to the PHA report. Redoing the PHA is a way to simplify the PH A documentation before it becomes unmanageable. Refer to Chapter 8 for more guidance on documentation issues. Example - Overlooked MOCs Company A has 10 process units at the same site. An MOC was completed four years ago as part of the project to install ambient detectors for toxic gas throughout the facility. That MOC was documented and labeled as a Unit 1 change because that unit had the highest concentration of the material. The Unit 4 revalidation team Updating its PHA might easily overlook this change and the unique aspects of the Unit 4 emergency response that were not considered in the MOC. However , if the revalidation team were Redoing the PHA, the discussion should include the appropriate response to toxic gas alarms, even if the MOC was not included in the list of changes.

5.2 Risk Management-Related Element Grouping |175 additions pose another challenge. New personnel must be brought into the culture and adopt it. Additional care should be taken that the new personnel do not bring negative cultural aspects from other places they have worked. In recent years, asset integrity efforts have experienced num ber of challenges that have led to incidents. These include: Inferior castings, bolts, and equipment that contain voids, stresses, or other m anufacturing defects but pass positive m aterial identification, Asset integrity database errors introduced during asset integrity database management, upgrades, and migration, Components that are not tagged and therefore not included in the asset integrity database; and Neglecting to improve design and m aintenance practices as they evolve in the industry, including useful inform ation from outside the industry sector. In a strong process safety culture, leaders em power the technical staff to study emerging issues that can improve the way they discharge their process safety responsibilities and defer to their expertise when they raise issues such as these. As in m ost other PSM S elements, asset integrity can be threatened by time pressures. This particularly can be a challenge with asset integrity tasks that need to be done during a turnaround. Keeping turnaround as short as possible has significant competitive advantages. Nonetheless, leaders should m aintain the imperative for process safety and defer to expertise before concluding the process can be restarted. However, hurrying to restart before critical asset integrity tasks have been com pleted, including rem oving blinds, replacing relief valves, and restoring bypassed interlocks, can be deadly. • • • •

95 Harris (Ref 5.8 Harris) provides an excellent set of guidelines for the design of storage facilities for lique fied gases that can minimize the potential for vapor clouds. Figure 5.2 shows a liquefied gas storage facility that incorporates many of these principles. •Minimize the wetted area of the substrate surface •Minimize pool surface open to atmosphere. •Reduce heat capacity and/or ther mal conductivity of substrate. •Prevent “slosh over” of co ntainment walls and dikes. •Avoid rainwater accumulation. •Keep liquid spills out of sewers. •Shield the pool surface from the wind. •Provide vapor removal system to a scrubber or another emission control device. •Provide a liquid recovery system for the contained volume to storage where possible. •Avoid direct sunshine on containment surfaces in hot climates. •Direct spills of flammable mate rials away from pressurized storage vessels to reduce the risk of a Boiling Liquid Expanding Vapor Explosion (BLEVE). •Provide sealed below-grade collection sumps directly below or adjacent to tanks or vessels havi ng volatile toxic materials that will rapidly collect and contain liquids and vapors that are released. The aforementioned Figu re 5.1 shows an example of a collection sump system for chlori ne releases (Ref 5.3 CCPS). •Provide below-grade collection sumps for flammable or combustible materials released fr om tanks or vessels that will allow the released materials to burn harmlessly without any effects on other equipment, cont ainment systems, or people. Figure 5.3 shows an example of a collection sump system, with a fire pit, for flammable/combustible liquid releases (Ref 5.3 CCPS).

146 and will also generate significant radiant heat which will represent a hazard beyond the boundaries of the fireball itself (Ref 8.17 CCPS 2010). The ignition of a flammable material in congested spaces, or the interior of process equi pment (vessels, as well as piping) can result in overpressures that are significantly greater than unconfined vapor cloud explosions. Vapor clouds do not need to be completely confined to result in these amplified effects. Outdoor releases which are partially confined by buildings, equipment, or have obst acles to the flam e front such as obstructions or topography have been shown to create conditions to enhance both the possibility of igni tion and the effects of resulting explosions (Ref 8.61 NFPA 2014); (Ref 8.57 Lewis). Pool Fires. Pool fires are of much long er duration than vapor cloud ignition events and the thermal radi ation intensity near the pool is usually high. In general, pool fires result in property damage, but do not usually result in signif icant numbers of casualties. However, one by- product of a pool fire is its poss ible effects on adjacent process equipment. The most severe effect from this external heat source is a possible Boiling Liquid Expanding Vapor Explosion (BLEVE). These events, which result in very large fireballs, occur when process vessels containing flammable materials with high vapor pressures (generally light hydrocarbons or chemicals with similar flammable properties, such as vinyl chloride monomer) are ex posed to significant amounts of external heat and fail rapidly and catastrophically (Ref 8.17 CCPS 2010). In general, the exposed vessel must be very close so that the flames from the pool fire impinge on the vessel or are exposed to the heat flux from the vessel for this hazard to be rea lized. Another by-product of pool fires is the potential for toxic combustion products to be released into the atmosphere. Jet Fires . Jet fires represent a special type of flammable hazard. A release of flammable liquid or gas under pressure creates a roughly conical tongue of flame that, like a pool fire or fireball, creates flame impingement and thermal radiation ha zards inside and outside of the cone boundaries (Ref 8.17 CCPS 2010). Runaway/Exothermic & Decomposition Reactions . A runaway reaction in a vessel or a physical overpressuriza tion of a vessel can cause it to lose

370 Human Factors Handbook [97] International Association of Oil an d Gas Producers (IOGP), “Report 423 – HSE management guidelines for working together in a contract environment,” International Association of Oil and Gas Producers, https://www.iogp.org, 2017. [98] U.S. Chemical Safety and Hazard I nvestigation Board (CSB), “E. I. DuPont De Nemours Co. Fatal Hotwork Explosion,” U.S. Chemical Safety and Hazard Investigation Board, www.csb.gov, 2012. [99] Energy Institute, “Managing major accident hazard risks (people, plant and environment) during organisational change,” Energy Institute, www.energyinstitute.org, 2020. [100] Center for Chemical Process Safety (CCP S), “Introduction to Op erational Readiness,” Center for Chemical Process Safety, https://www.aiche.org, Undated. [101] Center for Chemical Process Safety ( CCPS), “Process Safety Leading Indicators Industry Survey,” AIChE/CCPS, New York, NY USA, 2007. [102] Center for Chemical Process Safety (CCP S), Guidelines for Process Safety Metrics, Hoboken, NJ USA: John Wiley & Sons, 2009. [103] U.K. Health and Safety Executive, “HSE Management Stanadards,” HSE Books, https://www.hse.gov.uk, Undated. [104] Center for Chemical Proces Safety (CCPS), Driving Conti nuous Process Safety Improvement from Investigated Incidents, Hoboken, N.J., U.S.: John Wiley and Sons, 2021. [105] Center for Chemical Process Safety ( CCPS), Guidelines for Investigating Process Safety Incidents, New York: Wiley Inter Science, 2003. [106] T. E. Conklin, Th e 5 Principles of Human Performanc e: A contemporary update of the building blocks of Human Performance for the new view of safety, Santa Fe: PreAccident media, 2019. [107] British Petroleum (BP), “Deepwater Horizo n - Accident Investigat ion Report,” British Petroleum (BP), https: //www.bp.com, 2010. [108] D. Izon, E. P. Danenberger and M. Ma yes, “Absence of fatalities in blowouts encouraging in MMS study of OCS incidents 1992-2006,” Drilling contractor, vol. 63, no. 4, pp. 84-89, 2007. [109] Montara Commission, “Report of the Mo ntara Commission of Inquiry.,” Australian Government, www.industry.gov.au, 2010.

176 Guidelines for Revalidating a Process Hazard Analysis Q T E Application of Analysis Methodology (Section 3.2.1) Is the PHA method applied consistently throughout the prior study? Evidence includes: • A description of the PHA method used • Proper documentation of an Update , such as “No new causes discovered” or “No new issues” when the team could not identify additional unique causes for a deviation If the What-If method was used: • Were the analysis nodes small enough to identify hazards? • Was a checklist used to formulate questions? • Were all appropriate questions documented for each node? • Were most questions about haza rds for which the safeguards were deemed adequate? If the Failure Modes and Effects Analysis (FMEA) method was used: • Was each component boundary clearly defined? • Were concurrent safeguard failures analyzed? • Were reactive chemical hazards considered? • Were human errors considered? If the Hazard and Operability (HAZOP) Study method was used: • Were the analysis nodes small enough to identify hazards? • Were all applicable deviations in each node documented, even if there were no consequences of interest? • Were deviation meanings defined consistently? • Were loss of containment deviations considered? Is there evidence that required pr ocess safety information (PSI) was available and up to date (or recommendations were made to complete the PSI)? Such as: • Safety data sheets (SDSs) • Process chemistry • Maximum intended inventory • Operating limits • Equipment information (e.g., design temperatures/pressures) • Piping and instrumentation diagrams (P&IDs) • Electrical classification • Relief system design and sizing • Ventilation design • Codes and standards • Material and energy balances • Safety system design

E.33 Not Empowered to Fulfill Your Process Safety Responsibilities? |319 E.32 High Sense of Vulnerability to One Dangerous Material Overwhelm s the Sense of Vulnerability to Others A facility was restarting operations following a turnaround for replacem ent of a pressure vessel and a major control system upgrade. During start-up, a runaway chem ical reaction occurred inside the pressure vessel, causing the vessel to explode violently. Untreated residue and highly flam mable solvent sprayed from the vessel and immediately ignited, causing an intense fire that burned for more than 4 hours. The fire was contained inside the unit by the plant fire brigade with assistance from local volunteer and m unicipal fire departments. Shrapnel from the explosion flew in the direction of a day tank containing a highly toxic chemical, but was stopped by protective shielding placed for this purpose. Two em ployees who had been dispatched from the control room to investigate an unexpected pressure rise were near the residue treater when it ruptured. One died at the scene; the second 41 days later. Six volunteer firefighters and two contractors working at the facility were treated for possible toxic chem ical exposure. M ore than nearby 40,000 residents, including students at the adjacent university, were ordered to shelter-in- place for m ore than three hours as a precaution. The investigation team determ ined that the runaway chem ical reaction and loss of containment of the flam mable and toxic chem icals resulted from deviation from the written start-up procedures, including bypassing critical safety devices intended to prevent such a condition. Other contributing factors included an inadequate pre-startup safety review; inadequate operator training on the newly installed control system ; an unevaluated temporary change; and insufficient technical expertise available in the control room during the restart. Poor comm unications during the emergency Actual Case History

110 PROCESS SAFETY FOR ENGINEERS: AN INTRODUCTION Table 6.1. Example chemical exposure effects (CDC) Chemical Acute Effect Chronic Effect Aniline Impairs the blood’s ability to transport oxygen which causes skin to turn blue Asbestos Lung cancer and mesothelioma Benzene Irritating to skin, eyes, respiratory tract May cause central nervous system damage, anemia, and leukemia Hydrogen Cyanide Eye irritation , headache, vomiting at lower levels to profound cardiovascular and respiratory effects at higher level, potentially fatal Lead Memory loss, irritability, insomnia, depression, anorexia Nitrogen gas Lightheadedn ess at moderate levels, immediate asphyxiation fatality at high levels Phenol Nausea, sweating, arrhythmia. Coma and seizures can occur up to 18 hours after exposure Phosgene Irritant to skin, eyes , respiratory tract, pulmonary edema up to 24 hours after exposure, potentially fatal Exposure and concentration limits The impact a toxin is dependent on both the toxi c properties of the chemical and the duration of the exposure. A person can tolerate a certain amount of a toxin for a certain period of time and this varies for both the person and the toxin. For example, a small amount of alcoholic beverage can be tolerated with little effect; howe ver, a large amount in a short time may cause impairment or alcohol poisoning. The same am ount of alcohol consum ed by a young healthy person and an older adult with preexisting heal th conditions can cause different effects. The difference in response can be due to age, weight, diet, health, and other factors. Toxic hazards can be managed by identifying chemicals with toxic properties, understanding what concentration level can caus e impacts, and managing potential exposure. Toxicologists have conducted testing to dete rmine concentration levels at which health impacts occur. Exposure limits have been estab lished by several organizations. They define a concentration level above which a human will have defined health impacts.

Piping and Instrumentation Diagram Development 408 ●Guideline 2: Figure 19.3 shows the relative position of different control architectures in the framework of BPCS and SIS structures. A simple control may need a single loop control, or, depending on the situation, selective control, split‐range/parallel‐range control, or ratio control. If you are looking for “superior” control (tight control) you may choose feedback + feedforward control, or cascade control, or even feedback + feedforward + cascade control. When the control action starts to deviate from regula- tory control (or BPCS) and to go toward a more trip‐type control (or SIS type), override control, and then limit control, can be implemented (Figure 19.3). From a very simplistic point of view, it could be said that: “since the purpose of a unit operation/process unit is to convert one material into another – physically or chemically – the only required control loops are ‘composition control loops’ . ” This viewpoint is generally incorrect. In the majority of cases, we control unit operations and unit processes, not through composition loops, but through other loops. We mentioned before that for various reasons, we prefer not to use composi-tion control loops unless we really have to. Instead, we try to find some “underlying parameters” (among temperature, pressure, flow rate, and level) that are known to “direct” the composition, and then we control those simple parameters. The above statement is the golden rule in controlling unit operations and unit processes. One imaginary com-position loop may be broken into a few T/F/P/L loops. This could raise the issue of the loops interfering with each other. Beware the majority of units, either conversion units or separation units, that have a portion to store or hold fluids. This means the control of each unit most likely includes some aspects of vessel or tank control. Actually, container control is one common control scheme in the majority of unit operation and unit pro-cess controls. The below are some general rules that help: 1) The e quipment that we buy for a plant is not “custom built equipment” that we can then expect to operate exactly to our operating needs. Even for the case of custom made items, we usually expect equipment to operate in a pre‐determined “win-dow” of operation. The result is that almost all equipment in the plant should be “tamed” through a control system. Provide the required control (BPCS) to bring about the duty of the item. The item will be bought to do a task but usually the item will output a range of parameters. BPCS control will force the equipment to function in the required “window. ” Examples of tasks are flow rate and head for pumps, and heat duty for heat exchangers. 2) Che ck the required temperature, pressure for the item (inlet, outlet) 3) Che ck the required flow rate for the item. What is the minimum flow rate that can be handled without impact on process and what is the minimum flow before there is harm to the equipment? 4) Che ck the required composition for the item and care that should be taken. For example, a positive displacement pump is very prone to plugging if liquid has large suspended solids. In this case a strainer should be placed. 5) What ar e the required utilities and their tempera- ture and pressure? 6) What ar e the weak points of the item and the care that should be taken in designing a proper SIS for the item? 7) Which p arameters of the item need to be monitored by a rounding operator? (Think about those five parameters: temperature, pressure, level, flow rate, and composition.) 8) Which p ortions of the item need inspection and/or monitoring. 9) Is any hi story of item failures (frequency and time for maintenance, etc.) available? It can affect stages three and four of the item. For this step you may need to interview other users. 10) What if we lose the item? How can we minimize its impact on the rest of plant? Can we have a similar system as spare? (If the item is expensive other options should be considered.) Now we are going to apply our learning to develop a P&ID for two pieces of non‐general equipment: ●Example 1: gravity separator control (Figure 19.4) ●Example 2: flash drum control (Figure 19.5) -Override control-Simple control -Selective-Split/Parallel range-Ratio-Cascade control-Added feed forward“Superior control”BPCS -Limit control SIS Figure 19.3 Mo ving from a BPCS toward a SIS.

Table C-1 continued HF Competency Performance/ Knowledge Criteria Level 1 - Operator Level 2 - Supervisor* Level 3 - Manager** Non-technical skills Leadership Understands the concept and importance of leadership in emergency situations Can identify characteristics of effective leadership in emergency situations Can demonstrate effective leadership skills (e.g., centralizing communication, coordinating tasks, managing teams understanding of the situation etc.) in emergency situations Is able to assess effectiveness of leadership skills in abnormal situations Managing contractors Working with contractors Can effectively support contractors task performance Can implement methods for supporting contractors task performance Can identify potential risks of working with contractors and identify suitable controls Can review effectiveness of arrangements for working with contractors and lead their improvement

17. Error management in task pla nning, preparation and control 211 • Interlocks are defeated to allow someone to “get the job done”, with an assumption that other engineered protection will assure safety. • Interlocks are often left by- passed at the end of a calibration procedure. Routine and occasional defeating of interlocks is more likely if one or more of the following apply: • The interlock is easy to defeat. • Disabling interlocks has become an accepted practice. • The trips or automatic safety systems are easy deactivate, such as a toggle switch to turn off gas detection. • Process instrumentation is known to be faulty. • A high frequency of equipment faults or process upsets requires people to frequently shut down and fix faults. • Lack of local indication. Instead of going back to a control room to check indicators, the operator relies on their recollection of isolation and system status and defeats the interlock in the belief that the system is isolated. • No indication, alert, or alarm to notify control room operators that an interlock has been defeated. These conditions can make it easy to defeat an interlock, with limited opportunities for others to correct it. An example of a fatal accident where an interlock was defeated is in Table 17-6. Highly competent and experienced people defeat interlocks. They believe that they understand the system and can safely bypass or defeat the interlock, to help “get the job done”.

Chapter No.: 1 Title Name: Toghraei c04.indd Comp. by: ISAKIAMMAL Date: 25 Feb 2019 Time: 12:16:32 PM Stage: Printer WorkFlow: <WORKFLOW> Page Number: 21 21 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 This chapter discusses what is shown in the main body of a P&ID sheet, which is followed by a discussion of the different types and names of P&IDs based on their content. 4.1 Items on P&IDs Anything related to the process or anything needed to present the journey of raw materials into becoming final products should be shown on a P&ID. The above mentioned fact can answer many questions such as “Do we need to show the HVAC system of an indoor process plant?” In some cases, a portion or the whole plant could be indoors. In indoor plants, there can be an HVAC system in the building(s) to create a more suitable atmosphere for operators and equipment. As a general rule, very few details of an HVAC system are shown in such plants. However, in HVAC industries, the P&IDs can be drawn with their main purpose, that is, adjusting the air parameters. There are basically four different items that can be shown on P&IDs: 1) Pipe s and other flow conductors. 2) Equipment . 3) Instrumen ts. 4) (Instrumen t and control) signals. 4.1.1 Pipes or O ther Flow Conductors Pipes and other flow conductors such as pipes, trenches, channels, and so on direct and transfer fluid from one equipment to another. The general rule is that the flow conductors of the main process fluids should be shown in the P&IDs along with the pipes. In a water treatment plant, the water flows in channels, so the channels are shown, too. One important exception are tubes, which are not generally shown in P&IDs. However, there can be some “footprints” of tubes that can be seen on P&IDs. This will be discussed in chapters 13 and 18. The items for transferring bulk materials are generally categorized as “equipment” rather than “flow conductors. ” When it comes to showing pipe fittings, there is one rule: No pipe fittings are shown except tees, reducers, pro- cess flanges, and cap, plug, and blind flanges (Figure 4.1). A straight piece of pipe on a P&ID could be a pipe circuit in‐field with a bunch of elbows. A straight piece of pipe in‐field can be represented as a line with several directional changes on a P&ID. 4.1.2 E quipment The main players in processes are the equipment such as pumps, compressors, heat exchangers, and reactors. Containers can arguably be classified as equipment, too. Tanks and vessels are for process and/or storage purposes. All equipment should be shown on the P&IDs. If the equipment, however, are purely associated with mechanical details that are not related to the process, they may not be shown on the P&IDs. Examples are a gear box associated with a mixer, small built‐in lubrica-tion systems, and power hydraulic systems. In large compressors, the lubrication system can be large and a separate system. In such cases, the lubrication systems are shown, too. Equipment can be metallic, fiberglass, concrete, and so on, and in all cases, these should be presented. 4.1.3 I nstruments To implement every process, two requirements should be met: the process element (i.e. equipment) should be designed and tailored for a certain process and the control system should be formed to ensure implementa-tion. If one of these is not followed, it is most likely that the process goal will be only on paper. Instruments are the hardware that implement the control strategies in the plant. Industry practices with 4 General Rules in Drawing of P&IDs

24 | 2 Core Principles of Process Safety Figure 2.1 Overview of the Core Principles of Process Safety Culture 1. Establish an Imperative for Process SafetyProduction not possible without process safety 2. Provide Strong LeadershipLeaders inspire others to process safety excellence and Walk the Talk 3. Foster Mutual Trust Everyone does what they say and says what they mean 4. Ensure Open and Frank CommunicationsCommunication channels are open and encouraged and messenger not blamed 5. Maintain a Sense of VulnerabilityHealthy level of respect for hazards and risk of facility and company 6. Understand and Act Upon Hazards/RisksHazards and risks analyzed, controlled with appropriate safeguards, and managed 7. Empower Individuals to Successfully Fulfill their Safety ResponsibilitiesWorkers have authority and resources to performed assigned process safety roles 8. Defer to Expertise Technical knowledge related to process safety valued and technical opinions accepted 9. Combat the Normalization of DevianceDeviance from approved rules and standards never tolerated. 10. Learn to Assess and Advance the CultureCulture lessons-learned sought internally and externally. Learnings used to maintain and enhance culture.

General Rules in Drawing of P&IDs 37 ●Systems that are not directly related to the main pro- cess of a plant. ●Systems whose deletion from the main P&IDs does not hinder the understanding of the process (also cuts down on the crowded look) and eases readability. ●Systems that appear on several P&ID sheets and are exactly the same. If a detail P&ID is referred to by several main P&IDs, the detail P&ID is named “typical detail P&ID” , but if a detail P&ID is referred to by one main P&ID, the detail P&ID is named “nontypical detail P&ID” . Examples of typical detail P&IDs are pump seal flush, sampling sys - tem P&ID, safety shower and eye‐washer P&IDs, utility station P&ID, special control P&IDs (remotely operated valves, electric motors), piping detail P&ID, and HVAC equipment P&IDs. Examples of nontypical detail P&IDs are rotary machine lubrication and fire or gas detection and deluge system.A referred detail of a P&ID can be written down near the system in the main P&ID, or the detail P&ID can simply be mentioned as one of several “reference” P&IDs on the main P&ID. Some companies show the content of detail P&IDs in the guidelines and not on a P&ID. For example, a com-pany may not like developing special control P&IDs and display relevant information in the I&C design guide-lines. Piping detail P&IDs, which may have hook‐up piping detail of sensors, steam traps, and instrument air manifolds, may not be provided in a P&ID and instead be mentioned in the Piping design guidelines. Some of the types of detail P&IDs are explained below. ●HVAC drawings: These represent the HVAC system for industrial buildings. A sample HVAC P&ID is shown in Figure 4.25. ●Sampling system drawings: Sometimes the plant has many different sampling systems for the sampling of REV DESCRIPTION OF REVISION DATE BY ENG. APPR.HOT WATER DISTRIBUTION 0 PD-300-1003From wash water recycle pumpWash water To XXXX Wash waterWAT - AA -8/uni2033 -3018WAT - AA -4/uni2033 -3019 Wash water To XXXXWAT - AA -6/uni2033 -3020 Wash water To XXXXWAT - AA -4/uni2033 -3021 Wash water To XXXXWAT - AA -2/uni2033 -3022 Wash water To XXXXWAT - AA -2/uni2033 -3023Intermittent Figure 4.22 A Utilit y Distribution P&ID.

HUMAN FACTORS 359 Human Factors - A discipline concerned with designing machines, operations, and work environmen ts so that they match human capabilities, limitations, and needs. This includes any technical work (engineering, procedure writing, worker training, worker selection, etc.) related to the human factor in operator-machine systems. (CCPS Glossary) Many terms are used in the area of human factors. They are related but are not synonymous. The term “ergonomics” is typi cally used to describe the relationship between the human and the physical work environment such as locating valve handles within easy reach. “Human factors” includes ergonomics and is broader than the human’s relationship with the physical environment. It also includes the way we perceive, process information, and respond – the way our brains work. “Human performance” is another term ofte n used. Human performance is the result. Humans can perform a task successfully, or poor ly. By understanding human factors, we can strive to support the human to perform their task correctly. This support of successful human performance then supports good process safety performance. Human factors considers the hum an as part of a system. Human Factors Methods for Improving Performance in the Process Industries features the 3-part system shown in Figure 16.4. (CCPS 2007) The three parts are 1) people, 2) th e facilities & equipment that the human works with, and 3) the management syst ems that the human works within. Figure 16.4. Model for human factors (CCPS 2007)

DETERM INING ROOT CAUSES 213 Figure 10.3 Structured Root Cause M ethods Described in This Chapter While some methods use checklists as the logic analysis step, an understanding of the logic tree approa ch is still helpful because checklists are often developed from logic trees. Ch ecklists are especially helpful for incidents involving human factors. The approaches show n here also present tools to test logic, determine if the root causes identified go deep enough, help discern what to do if a team gets stuck, and aid in decision-makin g. These tools work with any logic analysis methodology. It is not the intention of the CCPS to endorse one particular method, but to present guidance on the various options and applications available. Structured methodologies that seek out multiple underlying systems-related causes of an incident and provide the mechanisms for determining and correcting system faults are genera lly found to be the most effective.

284 PROCESS SAFETY FOR ENGINEERS: AN INTRODUCTION by the wind momentum. If the jet has positive buoyancy (buoyant jet), the upward momentum will increase, and the initial momentum will be come negligible compared to the momentum gained due to the buoyancy. Then, the jet will be have like a plume. For a dense or negatively buoyant jet, upward momentum will decrease as it travels. Finally, it will reach a maximum height where the upward momentum disappear s and then will start to descend. This descending phase is like an inverted plume. No simple rule defines a particular combination of stability and wind speed as worst case. Often F2 gives the longest plume, but it is al so the narrowest. If a plume passes over a small densely populated area the width of the plume ca n be more important than its length. This is why most analysts run several combinations of stability and wind speed. Figure 13.9. Effect of initial acceleration and buoyancy on a dense gas release (CCPS 1999) Dense Gas Dispersion Many materials of concern in process safety are denser than air and require dense gas modeling that factor in density mixing effects. A dense gas is defined as any gas whose density is greater than the density of the ambient air th rough which it is being dispersed. This result can be due to a gas with a molecular weight greater than that of air, or a gas with a low temperature due to auto-refrigeration during re lease, or other processes. The behavior of dense gas dispersion differs markedly from ne utrally buoyant clouds. The major mechanisms include gravity slumping, air entrainment, an d thermodynamic processes. When dense gases are initially released, these gases slump towa rd the ground and move both upwind and downwind as initially the dense spreading effect upwind may exceed the wind speed effect downwind. Dense gases spread more widely an d often not as far as in neutrally buoyant dispersion and thus are not well modeled by gaussian simplifications. (CCPS 1996) Distinct modeling approaches have been attempted for dense gas dispersion: mathematical and physical.

90 | 7 Keeping Learning Fresh where R is the fraction of memory retained, t is time, and s represents how stable the memory has become. Many subsequent researchers have shown that memory can be made more stable with (1) frequent repetition of the lesson soon after the first learning, followed by (2) less frequent, but regular repetition over time. Both are critical. In the Throness study, employees clearly were receiving the first part of the process of memory retention. After any incident in a facility, the lessons learned will initially be repeated frequently, as the plant goes through the process of rebuilding and restarting operation. Similarly, in the Michel-Kirjan study, the experience of living through Hurricane Katrina and then rebuilding provided the initially strong reminders. But then, the reminders stopped as things returned to “normal,” and the lessons were forgotten. Then, what about employees at other company plants, or people who live in hurricane-prone areas away from where Katrina hit? They may not have received the initial strong reminders. For many them, the lessons-learned may be fleeting. Therefore, in implementing the REAL Model, special attention must be paid in the Embed and Refresh step to communicate the lessons-learned frequently at first, and then follow up with regular reminders. In Section 5.2.1, we discussed the theories of multiple intelligences and learning styles. We concluded that, because everyone can expect to see a broad range of learning styles among members of their workforce, we needed create tools for learning from incidents that are suited to many of these styles. In the following sections, we will discuss ways that companies and other organizations can create institutional knowledge using the REAL Model—and keep it fresh by communicating in ways that consider Gardner’s theory of multiple intelligences. We describe ways to use different communication styles, providing both hypothetical and real-world examples from committee members’ experiences at a range of companies. We will also show how communicating in multiple ways can reinforce one important process safety message that is often forgotten in the heat of the moment: Resist the very human urge to rescue a colleague who has collapsed (CSB 2008) and call for trained responders instead. We know intellectually that,

23. Working with contractors 303 It was also concluded that: “The contractors … were allowed to complete the hot work permit and begin hot work without getting approval from any DuPont employee knowledgeable about the process.” p11 CSB [98]. 23.2.3 A Human Factors perspective Working with contractors creates some Human Factors risks. These include: • A need for communication between personnel of the client’s organization and the contractor’s organization. As with all communication, this poses a potential source of error. • Procedural discrepancies between the contractor and the client not being recognized or agreed upon. • The contractor may be reluctant to challenge their client. • Some contractors may only work occasionally on a site. They may not be familiar with the site-specific hazards, safety management arrangements, or procedures. • As with all team and inter-team workin g, ensure clear allocation of roles, responsibilities, and accountabilities. • Contractual deadlines may create a perceived pressure for working long hours and/or rushing work. • Contractors often perform specialist inspection, testing, commissioning, and maintenance work. These tasks may be complex. • Contractors work on many sites with various terms or different procedures. • Contractors may perform the same task elsewhere with different safety management system. The client organization personnel should recognize the risks, and proactively offer support to help contractors perform tasks successfully. It is important to verify the activities and stop points with contractors. The host employer (Company) should ensure that the work is inspecte d and the work plan is being followed. “…the potential lack of familiarity th at contractor personnel may have with facility hazards and operations, pose uni que challenges for the safe utilization of contract services”. CCPS, [24].

B.2 Advancing Safety in the Oil and Gas Industry – Statement on Safety Culture |267 The organization understands that a decrease in or lack of reporting does not mean that culture is strong, or perform ance is improving. High quality and timely feedback is provided to staff following receipt of a report/concern. Employees know and believe that they will be treated fairly if they are involved in a near-miss or incident. Disciplinary policies are based on an agreed distinction between acceptable and unacceptable behavior. Mistakes, errors, lapses are treated as an opportunity to learn rather than find fault or blame. Incident investigation aims to identify the failed system defenses and improve them. Incidents are thoroughly reviewed at top-level m eetings. Lessons learned are implem ented as global reforms rather than local repairs and communicated effectively to employees.Lessons are learned from incidents that occur across the industry and in other high hazard industries. Lessons learned from internal data collection are shared with others across the industry. Leadership seeks to exceed the minimum established regulatory expectations with regards to safety. Leadership owns process safety standards and perform ance and does not rely on regulatory interventions to manage operational risk. Understand and Act Upon Hazards/Risks Attributes and descriptors for vigilance that promote a greater understanding and action on hazards and risks include: Process safety leading and lagging metrics are collected, evaluated and acted upon. Data gathering includes third parties, such as contractors.• • • • • • • • • • • • •

Preparing for PHA Revalidation Meetings 111 of the unit because of evolutionary changes in the unit itself (MOCs) or new PHA requirements (e.g., LOPA or damage mechanism review). If the Redo uses some salvageable content from the prior PHA as the starting point (rather than completely blank worksheets), the overall revalidation effort should be reduced. Estimating the time required for a revalidation using the Update approach is much more difficult. Revalidation of a simple system with relatively few changes will still require a significant fraction of the original PHA effort, even if the team chooses to read the prior document and reaffirm its accuracy. Units with more numerous and/or complex changes may require as much (or more) time as the original effort. If some portion of the prior PHA must be Redone to repair specific defects, additional time and resources w ill be required. All the issues discussed in Chapters 2, 3, and 4 should be considered, such as: • The number and complexity of new PHA requirements • The quality and completeness of the prior PHA • The quality and completeness of the PSI • The rigor of MOC/PSSR documentation • The number of undocumented changes • The learnings from operational experience Any estimates of revalidation meeting time usually assume that all the modifications to the process (and an y associated documentation) will be reviewed prior to the meetings by the revalidation leader or other experienced personnel. This ensures that pers onnel who are very familiar with the modifications can explain the rationale for each change and facilitate evaluation by the full revalidation team. The best strategy for scheduling is to begin with the end in mind. In some jurisdictions, authorities may impose s evere penalties for missing revalidation deadlines. When stating the allowed inte rval between PHA revalidations, most regulations and company policies refer to “completion” of the prior PHA, but some do not clearly define that term. When is a PHA complete ? Many companies define the completion date as the calendar date on which a final authorizing signature is affixed to an Approvals , Authorization , or Acceptance page of the report. Other definitions of the completion date include: (1) the last m eeting day of the core PHA team; (2) the day the PHA facilitator assembled the co mplete PHA report (including core, complementary, and supplemental analyses ); (3) the day the PHA report and/or its findings were transmitted or pres ented to management; or (4) the day management formally accepted th e report, usually accompanied by

160 Guidelines for Revalidating a Process Hazard Analysis 8.2 REPORT AND ITS CONTENTS Some companies have a specified format for the documentation of initial PHAs and subsequent revalidations. Absent that, the following is a possible table of contents for the revalidation report: Report Body. • Executive summary of the revalidation activities, the elevated risks identified, and the recommendations to reduce them • Purpose and introduction • Description of the scope of the revalidation (e.g., the process unit/sections studied and applicable requirements for inclusion in the PHA). Note: For clarity, this may mention out-of-scope items, particularly those analyzed in other studies. • Study dates and member attendance • Table of team members indicating their job function (e.g., operator, engineer) and indicating those team members who meet specific requirements for PHA team expertise • Description of revalidation approach , including justification of how specific requirements were met • Summary of recommendations • Summary of loss scenarios with elevated risks deemed as low as reasonably practicable (ALARP) Information Included in Appendices. • List of P&IDs, including revision number or date of version used • List of study sections or nodes • Copy of P&IDs, showing study nodes • List of PSI referenced, with copies of anything not version- controlled • Summary of assumptions • List of MOCs considered • List of previous incidents considered Study Worksheets and Checklists. • Core analysis (e.g., HAZOP, What-If) • Supplemental risk assessments (e.g., LOPA, bow tie) • Change Summary Worksheet (See Appendix C for example)

7.10 References | 103 7.12 Michel-Kerjan, E., Lemoyne de Forges, S. and Kunreuther, H. (2012). Policy tenure under the U.S. National Flood Insurance Program (NFIP). Risk Analysis 32 (4): 644–658. 7.13 Murre, J.M.J. and Dros, J. (2015). Replication and Analysis of Ebbinghaus’ Forgetting Curve. PLOS One, 6 July 2015. 7.14 Petrobras (2015). Acidente da P-36 - Explosão e Naufrágio [Video]. www.youtube.com/watch?v=Oz10Rsw_bJc&t=5s (accessed May 2020). 7.15 Price, A. (Arrangement) (1964). House of the Rising Sun [Song]. From the album “The Animals” by The Animals. 7.16 Reynosa (2012). Gas Plant Explosion Mexico [video]. www.youtube.com/ watch?v=6jhCKp2LHro (accessed April 2020). 7.17 Throness, B. (2013). Keeping the memory alive, preventing memory loss that contributes to process safety events. Proceedings of the Global Congress on Process Safety, San Antonio, TX (28 April–2 May 2013). New York: AIChE.

EQUIPMENT FAILURE 215 contained a solution of sodium nitrite. Sodium nitrite reacts with Chemfos 700 to produce nitric oxide and nitrogen dioxide, both toxic ga ses. Minutes after unloading began, an orange cloud was seen near the storage tank (Figure 11.30). Unloading was stopped immediately, but gas continued to be released. 2,400 people were evacuated, and 600 residents were told to shelter in place. (CCPS d) Figure 11.29 1 - Pipe connections in panel 2 and Chemfos 700; 2 - Liq. Add lines (CCPS f) Figure 11.30. Cloud of nitric oxide and nitrogen dioxide (CCPS f) Example 5 . During painting, a tank’s vacuum relie f valve was covered with plastic to prevent potential contamination of the conten ts. When liquid was pumped out the covering prevented air/nitrogen from replacing the liqui d volume. A vacuum developed, which led to the partial collapse of the tank, as shown in Figure 11.31. (CCPS 2002)

CASE STUDIES/LESSONS LEARNED 187 Figure 7.4 FCCU Separation Section

332 INVESTIGATING PROCESS SAFETY INCIDENTS Instructions: Review each classification statement to determine if it is TRUE or FALSE for the incident investigation finding in question. Any statement that is answered with FALSE presents an associated management system improvement opportunity. Table 15.3 Example Categories for Incident Investigation Findings Category Circle Defining Statements Design T / F The current design used the correct specifications and was built such that it was adequate for the intended service. (This includes design logic, hardware, installation accuracy, ar rangement, and ergonomic factors.) Process Controls T / F The control system(s) for the equipment or activity in question performed in accordance with the design logic, programming, or other instructions. (This addresses the actual control operation or execution. It would not include control logic that is in the “design” category.) Administrative Procedures T / F T / F T / F T / F The administrative procedures were: • available • adequate • accurate • approved and enforced These are the procedures covering broad organizational needs such as management of change, design and in stallation expectations (including avoiding low piping that someone could hit their head on and providing logical labeling), procurement (including approving substitutions and vendor equivalents), implementation (including defining training requirements and administrative support systems), safety (including specifying appropriate protective gear), environmental compliance, housekeeping standards, and emergency response. Operation Procedures T / F T / F T / F T / F The operational procedures were: • available • adequate • accurate • approved and enforced M aintenance Procedures T / F T / F T / F T / F The maintenance procedures were: • available • adequate • accurate • approved and enforced (The focus of this category is the actual maintenance tools, techniques, and standards for work that go beyond the traditional scope of normal inspection and preventive maintenance activities.)

THE UPSTREAM INDUSTRY 19 Figure 2-6. Example FPSO Figure 2-7. Example FPU (on dry tow showing parts normally submerged)

Fundamentals of Instrumentation and Control 249 when the process parameter of interest is not a simple parameter. In other words, it involves some computational aspect of the parameter, e.g. ratio, total or differential. For instance, “PD” would indicate the measurement of a pressure differential between two points in the process. The use of PD is of great benefit wherever flow goes through an obstructed route like in filters, strainers or even ion exchangers. The PD is used to ascertain when the porous medium has become plugged and needs backwashing. The modifier of “D” is widely used for pressure but it is not common for other parameters. Another example is the use of “Q. ” “F” means flow rate, but “FQ” means volume (total of flow rates means volume!). The modifier of “Q” is used arguable only for flow rate as it is not generally meaningful for other parameters. The total flow rate could be important for some utility streams, on some intermittent flow pipes, or other cases. The use of the ratio descriptor is very useful for flow, and the acronym for this would read “FF. ” This is of particular importance in dosing systems, where you need to control the dosing rate according to the flow rate of the process stream. The third letter is called the function letter, which is what we want to do with the parameter obtained and mentioned in the parameter of choice stated in the first letter. If the instrument is sensor, we use the letter “E”(element); if it is a controller, we use “C, ” etc. If the instrument is an indicator, we use the letter “I” but usage of “I” is not as common as in older days. The reason will be discussed in Section 13.9.3. The fourth letter in the acronym is, again, an optional descriptor. The fourth letter is a modifier that is most often used for alarm purposes or in SIS actions to indi-cate the action point of a loop, for example, high, low or low‐low. Table  13.6 gives more examples of instrument acronyms.Figure 13.10 gives one complete example of a control loop. 13.9.2 Divider T ypes The balloon dividers generally specify the “location” of instruments. What is important from an I&C practition-er’s viewpoint regarding “location” is if the instrument is in the field, in the control room, or in the field cabinets. Different divider shapes are shown in Table 13.7.In Table 13.7, I have shown the symbol with an irreg- ular shape. This is so that you focus on the divider and not the shape. The shape of the symbol will be discussed in Section 13.9.3. For the divider, there are five different cases: 1) No divider . This means that the instrument is out - doors, in the field. Examples are a flow element or sensor, or a level switch or gauge. They are connected to the control system, but they are not encased in a control room or in an auxiliary control cabinet. The majority of sensors are located outdoors and their tag doesn’t have any divider. 2) Single s olid line . This shows that the instrument is situated inside the main control room. It also indi-cates that the instrument is accessible and visible to the operator. Table 13.6 Examples of instrument acr onyms. No. ExamplesWhat is the parameter we are looking for?What we want to do with the parameter? MeaningParameter of interestParameter modifier FunctionFunction modifier Example 1 FQI F Q C Volume of fluid is controlled Example 2 LT L T Liquid level is transmitted Example 3 TC T C Temperature is controlled Example 4 PDC P D I Pressure difference is shown Example 5 AT A T An analyte is transmitted Example 6 LEHH L E HH Level sensor alarms on “HH” levelLC 504FT 504 Loop No.Loop No. Loop No.Loop No. FE 504LV 504Flow Indicator Controller Flow Control V alveFlow Transmitter Flow Element Figure 13.10 Instrumen t acronyms shown on a P&ID.

APPLICATION OF PROCESS SAFETY TO WELLS 85 Some universities recently have begun to include process safety in their curricula. The upstream industry has addressed process safety in detail in its own training courses since Piper Alpha. 4.3.7 Management System Audits and Safety Culture Surveys Onshore and offshore facilities apply safety and environmental management systems to control their own and their contractor activities. Audits are an essential aspect of these management systems to ensu re that what is specified actually occurs. These management systems apply to all aspects of upstream operations, not only well construction. In the US, larger on shore facilities usually follow PSM (OSHA 1910.119), while offshore, BSEE mandates SEMS which is based on API RP 75 with several additional requirements. Bo th PSM and SEMS require audits. BSEE requires periodic third-party independent audits. The Center for Offshore Safety has developed SEMS audit requirements (COS, 2014) and an audit service provider accreditation system that help to ensu re effective and consistent audits. Auditing and Management Review are elements in the RBPS pillar Learn from Experience . Audit results are considered in the Management Review, which also considers safety and environmental performance, incident investigation results and learnings, and mechanical integrity statistics relating to important barriers, to decide whether any changes to the current management system are warranted. The process of measuring current results (whether by metrics or by audit) and addressing these with changes if performance falls be low target demonstrates continual improvement. Safety culture is a recognized issue in both the downstream (Baker, 2007) and upstream industries (Deepwater Horizon Commission, 2011). Process Safety Culture is the first element of RBPS. The ap plication of PSM or SEMS helps create a positive safety culture. Howe ver, this is not a guarant ee that a company will reach the high level of process safety culture that it desires. The Baker report differentiates between a general safety culture (which focuses on more frequent occupational safety risks) and a process safety culture (which addresses rarer major incident risks). It is possible to be excellent in the former but weaker in th e latter, as was seen in the Deepwater Horizon incident. The introduction of SEMS in the US and safety case elsewhere have the intent to improve safety culture offshore. The usual tool for assessing culture is a survey. This can be a written questionnaire or a series of focus group meetings, both have advantages and disadvantages. The questionnaire has the advantage of anonymity and may be completed by everyone. But since it is generic, its responses tend to be general (e.g., “my supervisor values production more than safety”). Focus groups are not anonymous and may be impractical to apply to all personnel, but they do typically identify specific instan ces that are addressed more easily (e.g., “at the last shutdown, contractors started working without fully following permit requirements”). The Baker Panel (2007) provides an example survey questionnaire and shows how this is scored.

4.4 External Influences on Cultur e |125 Some of the more common external parties that can influence culture include: Contractors, Labor unions, Vendors/suppliers, Industrial and residential neighbors, External emergency responders, Law enforcement, Regulators and elected officials, Trade and professional organizations, The media: and Financial institutions. While not strictly external, corporate staff who work outside the facility and the com pany B oard of Directors may also have cultures that differ from the facility. The following pages discuss the potential influences of these external parties and how facility m anagers can manage those influences. Some of this m aterial has been provided courtesy of Hoffm an (ref 4.10). Contractors Contractors perform a wide variety of services for facilities, including operators, m aintenance, construction, and professional services. When contractors arrive at a facility, they bring with them the culture of their company as well as cultural influences from other facilities they serve. Leaders should be aware of the degree the contractors’ cultures differ from that of the facility. With that knowledge, leaders should then manage the business relationship to align the way contractors work and act with the facility’s culture. The economic forces that drive facilities to use contractors, and the cost- and time-competitive nature of contractors’ • • • • • • • • • •

389 a toxic gas supplied in cylinders. The re action is carried out at 200 ºF and at slight positive pressure. After production of C, the intermed iate is batch distilled under full vacuum and the purified C is colle cted, re-inhibited with MEHQ, and stored at ambient conditions under air. C is a reactive monomer with a flash point above 200 ºF. Distillation bottoms are drummed for disposal as a reactive waste. Final Product Production: C + D = Z Intermediate C is polymerized with raw material D in a batch reaction producing final product Z, which is diluted in solvent E. Current production is in a batch reactor with all materials including initiator and solvent in the initial charge. The re action is conducted at atmospheric pressure, and cooling is achieved by solvent reflux and supplemented by a reactor jacket. Available pl ant cooling water is used. Raw material D is a reactive, corr osive (to human tissue) monomer, and Solvent E is flammable and cons idered toxic. Both materials are supplied and stored in bulk. The in itiator is a peroxide type which requires refrigerated storage. The final product Z is a polymer which, by itself, is non-toxic and nonreactive. However, in the current solvent, the product is flammable and toxic (see Figure 15.1).

390 Human Factors Handbook supply, leading to a range of conseque nces including a fire on the crude distillation unit, and various effects on the vacuum distillation unit, the alkylation unit, the fluidized catalytic cracking unit (FCCU), and the Butamer units. These conditions led to a plant upset but were not the cause of the explosion that happened five hours later. • Hydrocarbon flow was lost to the deethanizer, a vessel in the FCCU recovery section. As a result, the liquid was emptied into the next vessel along the debutanizer. The system was set up to prevent loss of liquid in vessels. This caused valve FV 404 to close, preventing hydrocarbon from leaving the vessel. This had a knock-on effect on outlet valve FV 436 , causing it to close. The hydrocarbon in the debutanizer was now blocked. However, the trapped liquid was still subject to heat. As a result, the liquid vaporized and the debutanizer pressure rose, which caused the pressure relief valves to open. It also caused the debutanizer to vent into the flare knock-out drum and on to the flare. • Shortly after this event, the liquid level in the deethanizer was restored, valve FV 404 reopened, and flow to the debutanizer restored. This should have caused valve FV 436 to open and allow hydrocarbon out of the debutanizer into the naphtha splitter, but this did not occur. The operators in the control room received a signal incorrectly indicating that valve FV 436 did open. The debutanizer continued to fill with liquid, while the naphtha splitter emptied. • Operators’ control systems did not a llow overview of the whole process. The process was broken down into disc rete sections, which could be seen on separate screens. The operators focused on problems around the deethanizer and debutanizer. • The operator opened another valve (valve HCV 439), with the intention of relieving the pressure on the debutanizer system. Opening valve HCV 439 did not prevent the debutanizer becoming full of liquid, and it vented to flare via the knock-out drum (for the second time). • Opening of valve HCV 439 caused the liquid levels in the interstage drum to rise, so that it flooded into the dry end and caused the compressor to trip (shut-down). A large volume of ga s had nowhere to go and had to be vented to the flare stack to be burned off. • There were high liquid levels in the flare knock-out drum, which were increased by an operator’s next acti ons. The operators tried to remove the flooding from the dry end of the interstage drum by draining the liquid directly to the flare line via an im promptu modification that employed steam hoses. The operators’ actions resulted in the gas compressor restarting, which increased flow through the unit and caused an increase in pressure in the debutanizer, which vented to flare (for the third time). • Operators had decided to alleviate the pressure in the debutanizer by opening valve HCV 439 to allow hydrocarbon to move from the

Piping and Instrumentation Diagram Development 156 However, this solution has some shortcomings. In this solution there is always the chance of reverse flow in one or more tying pipes. If there is a chance of intermittent flow in one of the tying streams, or the chance of drop-ping pressure in any of these streams, a check valve should be installed on the intermittent or low pressure stream. The other disadvantages of having more than enough nozzles are that each nozzle potentially is a source of leakage or source of corrosion. However, by shifting a nozzle from shell site to roof site the chance of liquid leakage is eliminated, but this is not always doable. The other solution to minimizing the number of noz- zles on containers is using shared nozzles rather than the dedicated nozzles. For example, some companies use one nozzle on top of the tank shell for the dual purpose of venting and overflowing. This is important to consider that two nozzles can be merged to one shared nozzle if the functionality of each of them is not overlapping. It means the need for each nozzle should be only in one phase of operation. In the previous example where we merged overflow nozzle and vent nozzle it was doable because the vent nozzle needed to be functional during the normal operation of a tank while the overflow nozzle only needed to be functional during upset operation. Therefore, these two nozzles could be merged to a single shared nozzle because they are not operating in the same phase. The other examples for using shared nozzles are using one nozzle for flushing and draining a container. The other example is using manway nozzles for the pur - pose of creation of draft to dry up the tank internally for maintenance purposes. One example is the nozzles that have the capability of being a PSV and a manway at the same time, or nozzles that are a PSV and “thief hatch” at the same time. 9.9.6 Nozzle I nternal Assemblies Nozzles on containers can be installed without any internal connected assembly and this is in the majority of cases. However, there are some cases that a nozzle should be connected to an assembly from the other side and inside of the container. The most common types of nozzle internal assemblies are vortex breakers, down commerce, risers, and extensions, which are shown in Figure 9.19.Vortex breaker: Vortex breakers are the type of nozzle that is sometimes installed on some outlet nozzles of liq-uid containers. The main purpose of the vortex breaker is to prevent the creation of a vortex in the outlet of a container. You may have seen the creation of vortex when you try to empty your bathtub. The vortex that we may see in our bathtub is not very harmful but in an industrial context it should be avoided. The reason we do not want a vortex in the outlet nozzle of some liquid containers is  that it causes some erosion inside the container and also the atmosphere inside the tank may be entrapped in the outgoing liquid, which may not be a good thing. Prevention of vortexes is more important when a liquid container sends liquid to the suction of a pump. If pre-vention of vortexes fails the liquid going in to the pump suction may have some gas bubbles, which is detrimental for the pump operation. Therefore, it is important to know that vortex breakers may be needed only if it is a liquid container and on an outgoing nozzle. One impor - tant thing is that not all outgoing liquid nozzles need vor - tex breakers. There are some formulas that show whether we need to install a vortex breaker for a nozzle or not, but generally speaking vortex breakers are not needed for huge tanks and may be needed for a small vessels only. Down comers: in some cases we need to install down comers if we need to take liquid from a specific zone Vs. Figure 9.18 Decision on the number of pr ocess nozzles. Vortex breaker Down comer Riser Extended Figure 9.19 Nozzles in ternals.

418 PROCESS SAFETY FOR ENGINEERS: AN INTRODUCTION A weakness in the OSHA standard is that it allows the use of wood “protected against impregnation by ammonium nitrate” for the w a l l s o f t h e b i n ( t h e f l o o r m u s t b e n o n - combustible) (OSHA 1998). NFPA 400 was updated in 2016 and now requires buildings be of non-combustible construction, automatic sprinkle rs, and fire detection systems, the last two being retroactive requirements. The fact that the OSHA standard covers AN is not well known in the fertilizer industry as reported by the industry itself (CSB 2013). OSHA did not have a history of citing fertilizer facilities under the Blasting and Explosive Agents standard, contributing to this lack of knowledge. This contributed to a lack of proc ess safety knowledge in the industry, which in turn led to inadequate hazard identifi cation and emergency response planning. AN is not covered by OSHA PSM, or EPA RMP. This means that facilities handling AN do not need a process safety management program. The lack of a PSM program led to several safety management gaps. Stakeholder Outreach . WFC shared little information with emergency responders and the community. The lack of process safety kn owledge on WFC’s part contributed to this. Without an understanding of the potential haza rds of ammonium nitrate at the WFC facility, they had no motivation to prevent the comm unity from building up near the facility. Process Knowledge Management. Since AN was not on the PSM or RMP highly hazardous chemical list, and because the fertiliz er industry was not familiar with the OSHA Blasting and Explosives Agents standard, neither the WFC management and employees, nor the emergency responders, were familiar with the AN hazard. The Emergency Responders did not know that AN could detonate. Process safety knowledge includes collecting and disseminating information and learnings from incidents with similar technologies and chemicals from throughout the industry. AN producers and handlers should learn from the long history of AN related incidents. In Texas in 2009, a fire occurred at another facility that stored and handled AN. The firefighters decided not to fight the fire but to evacuate the area . About 80,000 people were evacuated. A review of that emergency response was conducted, an d an after-action report was issued that emphasized the need for emergency responders to “reflect on protection, response and recovery activities” that occurre d in the 2009 fire (CSB 2013). This report apparently was not known by the West Fire Department. Emergency Management. The absence of AN from the PSM and RMP rules led to no emergency planning, which also would have b een required by these regulations. When responding, the fire department initially tried to fight the fire, but only the fire engines internal tanks could be used until a hose could be conne cted to the hydrant, which was 490 m (1,600 ft) away. They did not have enough hose to re ach the fire. Developing an emergency response plan should have exposed these problems an d allowed them to be addressed before an incident occurred. Introduction to Emergency Management Emergency management is a necessary element of process safety. Despite all the effort put into preventing and mitigating potential process safety incidents, they still occur. Be prepared.

168 GUIDELINES FOR MANAGING ABNORMAL SITUATIONS In both industries, automation is also gradually replacing many routine tasks, which can provide many benefits as well as risks when these systems fail. Major aviation incidents that are associated with failures of such automated systems include: October 19, 2018, Lion Air flight LNI 043, Boeing 737-8 (MAX) crashed into the Java Sea some 12 minutes after take-off from Jakarta (Republic of Indonesia, 2018 Preliminary Report) leading to 189 fatalities. An automated sy stem called “Maneuvering Characteristics Augmentation System” (MCAS) that was supposed to counter a “nose-up” tendency und er certain conditions operated erroneously and repeatedly pointed the nose of the aircraft down. The MCAS system relied on angle-of -attack (AOA) data from a single AOS sensor that was not functioning correctly. The design of the system was flawed, and the pilots were unable to understand and manage the problem. March 10, 2019, Ethiopian Airlines flight 302, Boeing 737-8 (MAX) crashed into the ground some 16 minutes after takeoff from Addis Ababa, Ethiopia (Ethiopian Mini stry of Transpor t 2019 Preliminary Report) leading to 157 fatalities. The erroneous operation of the MCAS system was a key factor again, although on this occasion the Pilot followed a procedure and switched off the MCAS/ stabilizer trim system. However, by the time he had done this, the aircraft was in a “mistrim” situation, and the pilots could not physically turn the trim wheels to correct the situation. Finally, they switched the trim system back on, likely attempting to use the automatic trim, but the MCAS cut in again and the plane nose-dived into the ground. January 9, 2021, Sriwijaya Air flig ht SJ 182, Boeing 737-500 crashed into the Java Sea just 4½ minutes after take-off from Jakarta (KNKT Preliminary Report, 2021) leading to 62 fatalities. The Flight Data Recorder showed an anomaly that indicated a gradual reduction in engine thrust from the left engine that was under control of the auto-throttle. The right thrust level remained unchanged. A disengagement of the autopilot occurred at 10,900 ft, which would have been compensating for the significant difference in thrust, followed by the sudden roll of the plane to the left to more than 45 degrees of bank before it crashed into the Java Sea. Prior to this

196 | Appendix: Index of Publicly Evaluated Incidents given incident, another person might have identified a different set of primary and secondary findings. Readers should therefore use the index only as a tool to identify incidents with potentially useful findings and not look to it for statistical information. While the printed version of the index only allows readers to search on single root causes, culture core principles or causal factors, an electronic spreadsheet version of the index with additional capabilities is available to download. The spreadsheet version provides capability to: • search for incidents involving multiple root causes, culture core principles, and causal factors • search for incidents by reporting source, industry sector, and equipment type. Additionally, the spreadsheet contains any explanatory comments made the CCPS subcommittee member who indexed it. Download the spreadsheet from the CCPS website at: www.aiche.org/ccps/learning-incidents On opening the spreadsheet, enter the password: CCPSLearning CCPS will update the electronic version periodically to add incident investigations reported after 1 January 2020 as well as investigations from other sources. Readers are invited to visit the index webpage periodically to check for updates. CCPS also invites readers to help index future incidents. The indexing form may be downloaded from the same webpage. Once the form is complete, please submit to CCPS by email:CCPS@AIChE.org. A.2 How to Use this Index 1. Use Section 6.1 to identify the RBPS elements, culture factors, and causal factor area at your company or unit where improvements are needed. 2. Look up these elements, culture factors, and causal factors in the index, Section A.3. Note the relevant index numbers. The listed primary findings

142 PROCESS SAFETY FOR ENGINEERS: AN INTRODUCTION Many refineries and chemical plants are located on the Gulf Coast and in the Los Angeles basin of the United States. In addi tion to the hazards associated with the chemicals they handle, what othe r hazards should they address? Fertilizers are used to increase crop yield. The primary components in fertilizers are nitrogen, phosphorus, and potassium. Ammonia can be used as the nitrogen source. It is made from natural gas and air. Ammonia and nitric acid are used to make ammonium nitrate which is used as the fe rtilizer component. Phosphorus comes from phosphate rock which is surface mined and th en treated with sulfuric acid and nitric acid. Potassium comes from potash which is also mined and includes a large tailings dam. (Madehow) What hazards should be addressed for the manufacture of these fertilizers? Covid 19 had an impact on the way nearly everyone worked. How do you think it impacted process safety? What hazard was involved in the Space Shuttle Columbia disaster? Describe the accident scenario. Long pipelines are used to transport materi als across regions, states, even countries. What hazards might threaten the pipelines? Food processing plants such as ice crea m factories and meat packing plants use refrigeration involving ammonia, propane, and nitrogen. What hazards and what risks might this pose? Figure 8.6 shows a refinery during historic river flooding. What impacts might this have on process safety in the facility? Figure 8.6. Coffeyville Refinery 2007 flood (KDA)

268 INVESTIGATING PROCESS SAFETY INCIDENTS 11.2.1 H uman Factors Before and During the Incident Leadership sets the tone on the importance of incident investigation and learning from incidents. Leaders and investigators are not out to assign blame. Actions taken to “blame and shame” are not constructive a n d generally do little to prev ent similar incidents from occurring. Therefore, it is necessary to foster an open and trusting environment where people feel free to discuss the evolut ion of an incident without fear of reprisal. Without such a supportive envi ronment, involved individuals may be reluctant to cooperate in a full disclosure of occurrences leading to an incident (Rothblum, 2002) and the incident investig ation may be concluded prematurely with the root causes left uncovered. Example: “An incident involved a control board operator, who was an introvert and had few, if any, friends at the workplace. Other members of the crew apparently played jokes at his expense. One day, the board operator closed a valve in erro r, whereas another crew member monitoring the process understood the error but intentionally delayed communication of the error to the board operator. By the time the crew member rudely informed the bo ard operator of his error, it was too late to prevent the incident. It was also found that the board operator spent considerable time on non- work- related telephone calls while the process was out of control.” (Broadribb, 2012) Operational discipline is a very important topic in human factors and process safety. Operational discipline is not about punishing a worker who may have made an error. Instead, it is about enabling people to perform every task correctly every time. (CCPS, 2011) This is done by clearly defining how processes will be managed providing needed resources and establishing clear expectations for fo llowing the procedures. This operating discipline is supported by leadersh ip, organization, communication, teamwork, resourcing, and documentation. These topics may underlie why a human has behaved in a certain way. Topics relating to operational discipline are included in Table 11.1 lis ting potential huma n factors issues. Human factors issues can also impact the performance of the investigation team itself. They may be subject to human biases that will lead them to assume they know what happe ned or to rely on judgements already established about the persons involved in the incident. (IOGP, 2018) It is

326 In the batch process, the correct am ount of Reactant A dissolved in Solvent S is fed to the reactor from a weigh tank sized to hold exactly the correct charge for one batch. The ap paratus features three-way valves which do not allow flow directly from the storage tank to the reactor and will return any excess material fed to each charge tank to a tank overflow to ensure the correct charge. Catalyst C is then added, and an exotherm created by the heat of mixing confirms that the catalyst charge has been made. Reactant B is then fed at a co ntrolled rate to maintain the desired reaction temperature. The feed tank is sized for the correct charge for one batch. Even if the entire char ge of Reactant B is somehow fed without reacting, the size is such that it is not possible to reach the critical concentration of Reactant B unless the operators filled and emptied the charge tank several times. The safety advantage of the batch process is that it is very difficult to reac h a hazardous condition. But, the disadvantage is that the reactor is very large, so the consequence of the potentially explosive reaction would al so be large in the event that the critical concentration was somehow reached. The continuous reactor, on th e other hand, can be sized much smaller–perhaps 1/10 the size of the batch reactor for the same production volume. Therefore, th e consequence of the potentially explosive reaction is much smaller in case the cr itical concentration of Reactant B is reached. However, to keep the process running, the continuous process must be connected to a large feed inventory of Reactant A/Solvent S mixture, Catalyst C, and Reactant B. It is possible, with various feed flow rate ratios, to produce almost any concentration of Reactant B in the reactor. The continuous process relies on instrumentation, such as flow me ters, ratio controllers, and control valves, to ensure that the material s are fed in the proper ratio. The control instrumentation, logic, and other hardware could fail in a way that results in a concentration of Re actant B in the reactor in excess of the critical concentration. While th e control instrumentation could be made highly redundant and reliable, the continuous process still relies on instrumentation to prevent th e Reactant B concentration from exceeding the critical value and should therefore be considered inherently less safe. In addition, wi th the continuous process operating at steady state and the reactor te mperature controlled by cooling the reactor jacket, there is no positive feed back signal to verify that Catalyst

10.5 Internalize | 137 be resistant to each individual material they carried but incompatible with mixtures? The oleum piping failure was happening from the outside-in. Could their hoses be failing for a similar reason? Antônio didn’t think Anonymous 3 applied to their situation. The plant instrument techs were religious about inspecting and testing grounds, cables, and continuity. But more importantly, none of the hose failures appeared to be caused by burning, even localized burning. Perhaps hoses rubbed across cables, though. He’d have to check that. 10.5 Internalize A week later, Antônio met again with João, Juliana, and Francisco to compare what they’d learned. Francisco reported that in every case, the proper hoses had been specified. João reported, with obvious relief, that operators were double verifying each time they made a hose connection and no exceptions had been found. Juliana shared the inspection and maintenance records. The inspection and replacement intervals hadn’t changed in five years. In the past year, 99.5% of hose inspections and tests were completed on time. While not perfect, the delay for the other 0.5% was far too short to explain the observed problem. Antônio reported that he found no duct tape in the control room and locker room, and that in fact the plant had not purchased duct tape in several years. Antonio also shared what he learned from the external incident review, and the four colleagues developed a plan for further study. They met again a week later to discuss their findings, summarized in Table 10.1. Table 10.1 Results of Study Based on External Incidents Study plan action Feijoada plant finding Evaluate how hoses might be subjected to rubbing, snagging, and impact. None found in the process of connecting, using, and disconnecting hoses. Positively identify hose material of construction. All materials of construction verified correct. Check if new vendors’ hoses look the same as older hoses. The new hoses looked slightly different but were marked more clearly than the old ones. Operators felt the new hoses presented a smaller chance of mix-up.

334 INVESTIGATING PROCESS SAFETY INCIDENTS 15.4 REVIEW OF NEAR-M ISS EVENTS As discussed in Chapter 5 (Initial Noti fication, Classifying and Investigating Process Safety Incidents), the reporting and investigation of near-miss events is an essential part of the safety management process. While the scale of the investigation for a near-miss may be significantly lower than that for a larger event, the learning can be just as relevant. Further benefits of investigating near-misses include: - More frequent investigations and learning. - Greater involvement of staff with the investigation and learning process. - Improvements in proc ess safety culture. Encouraging the reporting and investigation of near-misses can often lead, in the short term, to an apparent increase in the number of “incidents” albeit at a lower level of classification. This pattern is a useful indicator that the message about the importance of conducting investigations, whatever the scale of the incident was received by the workforce. In the longer term, the number of near-misses may start to decrease, although, more importantly, there should be a reduction in the number of the larger incidents. A review of the causes and recomm endations arising out of near-miss events should be conducted on a period ic basis to identify common factors that may be targets for improvement. This proc ess could be included the Recommendations Review shown below in 15.5, or part of a separate process. 15.5 RECOM M ENDATIONS REVIEW To effectively address the findings of an investigation, appropriate recommendations should be drafted and acted upon within the agreed timescale. Recommendations should ac curately translate the investigation findings into actions that are “SMART” (Specific, Measurable, Agreed/ Attainable, and Realistic/ Relevant, with Timescales; see 12.2.2). They should clearly define what is to be done so that the impl ementer understands not only what to do, but why . A well-written recommendation will also identify the consequences that are being avoided or abated, an d/or the likelihood of a reduction of consequences or occurren ce. Periodic checks or audits of

28 Guidelines for Revalidating a Process Hazard Analysis are recommended by the general NFPA 652 standard, along with several material-specific standards: • NFPA 61. “Standard for the Prevention of Fires and Dust Explosions in Agricultural and Food Processing Facilities” • NFPA 484. “Standard for Combustible Metals” • NFPA 654. “Standard for the Prevention of Fire and Dust Explosions from the Manufacturing, Processing, and Handling of Combustible Particulate Solids” • NFPA 655. “Standard for Prevention of Sulfur Fires and Explosions” • NFPA 664. “Standard for the Prevention of Fires and Explosions in Wood Processing and Woodworking Facilities” If dust hazards may be present in a pr ocess, dust hazard analyses based on such standards are often performed as complementary analyses in the PHA and are subject to revalidation. API-RP-752, “Management of Hazards Associated with Location of Process Plant Permanent Buildings” [5], API-RP-753, “Management of Hazards Associated with Location of Process Plant Portable Buildings” [6], and API-RP-756, “Management of Hazards Associated with Location of Process Plant Tents [7]. Most PHAs require special consideration of facilit y siting issues. Many companies use the guidance in these RAGAGEPs, and to perf orm analyses of buildings intended for occupancy, and the results of such studies are considered when addressing facility siting in related PHA(s) and making related risk judgments. In these cases, a revalidation team should have such study results available and consider whether (1) any changes to the process affect the potential for vapor cloud explosions, fires, and toxic gas release scenarios or (2) the use or occupancy of buildings near the process have been adeq uately managed with respect to such study results. 2.2 INTERNAL COMPANY POLICY REQUIREMENTS Many companies that perform PHAs have written company policies governing PHAs and related activities such as reva lidations. Such internal PHA policies can be broadly categorized as compliance-driv en; environmental, health, and safety (EHS)-driven; or value-driven policies, an d they often correlate with the maturity of the organization’s process safety culture.

86 Guidelines for Revalidating a Process Hazard Analysis • Documenting MOC hazard reviews thoroughly and so they can be easily addressed and Updated into the unit PHA. For example: o Using the same software and format to document MOC hazard reviews and PHAs o Ensuring personnel/contractors who perform and approve MOC hazard reviews are familiar with core PHA methods and the PHA requirements of the facility o Prior to determining the reva lidation approach, reviewing the MOC change log with experienced operations personnel to ensure that all significant changes have been included in the revalidation scope. In addition, auditing the P&IDs in the field with experienced operations personnel to see if any changes can be found in the field without proper documentation in the MOC system o Keeping a working copy of the prior PHA updated with approved MOCs over the revalidation cycle • Considering changes in interfacing systems (e.g., upstream units, downstream units, utilities) that could affect loss scenarios in the PHA being revalidated • Searching incident databases main tained by industry groups and other similar facilities within the organization to help broaden the collection of incidents for the revalidation • Reviewing metrics relevant to risk judgments during the revalidations, such as alarm syst em performance, overdue ITPM tasks, temporary impairments, dr awing revision backlog, overdue training, and incident trends • Reviewing any recent PSM (or similar) audits to identify any specific findings related to the PHA being revalidated or general findings related to key elements (PHA, PSI, MOC, mechanical integrity [MI], etc.) relevant to the revalidation • Addressing the PSM issues that caused deficiencies in the operating experience records outside of the revalidation process Obstacles to Success: • Trying to force a revalidation into an Update approach when a Redo would be the more efficient way to incorporate many complex changes/incidents • Trying to force a revalidation into an Update approach when many of the changes were not implemented as approved by the MOC reviewers