Incorporating Future Climate Scenarios in Oil Industry’s Risk Assessment: A Greek Refinery Case Study

Abstract

The impacts of climate change are anticipated to become stronger in the future, leading to higher costs and more severe accidents in the oil industry’s facilities and surrounding communities. Motivated by this, the main objective of this paper is to develop, for the oil industry, a risk assessment methodology that considers future climate projections. In the context of an action research effort, carried out in a refinery in Greece, we adapted the organization’s extant risk management approach based on the Risk Assessment Matrix (RAM) and suggested a risk quantification process that incorporates future climate projections. The Climate Risk Assessment Matrix (CRAM) was developed to be used to assess the exposure of the facility’s assets, including human resources, to future climate risks. To evaluate CRAM, a comparison with RAM for the specific organization for the period 1980–2004 was made. Next, the application of CRAM for the period 2025–2049 indicated that, even though the resilience of the operations of the company to extreme conditions seems adequate at present, increased attention should be paid in the future to the resilience of refinery processes, the cooling system, and human resources. Beyond the specific case, the paper provides lessons for similar organizations and infrastructures located elsewhere.

1. Introduction

The European Commission defines as Critical Infrastructure (CI) an asset, or system, which is essential for the maintenance of vital societal functions [1]. As part of Critical Infrastructures (CIs), the oil industry has been providing communities and the broader economy with indispensable goods and services for quite a long time [2]. Recent studies indicate that certain activities of the oil industry and related infrastructure are highly exposed to extreme weather events due to climate change [3,4,5], with severe consequences for the industry’s structural integrity, its services, and the occupational safety and health (OSH) of its employees [2,6,7], (fires, explosions, and toxic releases, the so-called Natech accidents [8]).

As climate change increases the risk of such disasters in the future [8], more effective prevention and mitigation measures are required [9,10]. Already several measures have been applied for better risk management [2]. The protection of the oil industry is an important aspect of national security. It is highly linked to the European Programme on CI Protection (EPCIP) [11] and to the effective risk assessment approaches for strong CI Protection plans by identifying threats, assessing vulnerabilities, and evaluating the consequences of extreme weather events [12]. Moreover, Natech risks are regulated by the Seveso Directive on the management of major-accident hazards [13]. In the USA, the Department of Homeland Security [14], along with the American Petroleum Institute (API), the American Chemistry Council, the National Association of Manufacturers, International Petroleum Industry Environmental Conservation Association (IPIECA), Oil and Gas Climate Initiative (OGCI), the Climate Leadership Council, as well as major oil companies and the Oil & Gas Preparer Forum [15] suggested methods to develop climate change risk assessments for the oil industry. In parallel, Australia [16], Canada [17], the United Kingdom (UK) [18], and Norway [19] have already taken the first steps to assess risks and implement climate adaptation measures into their broad risk frameworks.

Representative methodologies for the analysis of vulnerabilities and risks in CI, including those of the oil sector, include Athena [20], DUTCH (National Risk Assessment) NRA [21], Failure mode, effects, and criticality analysis (FMEA/FMECA) [22], Hazard and operability study (HAZOP) [23], and the Risk Assessment Matrix (RAM) [24]. These risk assessment methodologies are either qualitative, quantitative with probabilistic analysis, or semiquantitative approaches with the use of risk matrices [25]. Recently, several climate quantification approaches (QRA) have been developed for different sectors, or with respect to different climate hazards [26,27], while evaluation of risks through the use of risk matrices has been accomplished [28,29,30]. Recently, an integrated risk assessment methodology was established by Ebad Sichani et al. [31], who studied the climate change impacts of hurricane events to the oil industry and estimated the risk in terms of monetary losses.

The Risk matrix has been one of the most popular methods for demonstrating and disseminating the results of risk assessment and for expressing the size of the risks and mapping the combinations of likelihood and consequences of risks [32]. However, in general, extant risk approaches in the oil sector do not consider explicitly future risks under the context of climate change. They consider only risks based on the current estimations or past observations of climate extremes.

The oil industry’s infrastructures have been designed under the assumption of stationary climate conditions [33], based solely on historic observations (e.g., Eurocodes CEN, 1990 [34]). However, the characteristics (frequency and intensity), as well as the distribution of the extreme values of climatic parameters might change due to future climate conditions [35]. Thus, it becomes evident that more elaborate approaches that take into account expected climate change extremes need to be employed [36]. The approach proposed in this paper aims to conduct a climate change risk assessment for the oil industry by adjusting the risk assessment methodology formed around the well-known Risk Assessment Matrix (RAM), or Probability and Impact Matrix [24,37], which has been widely used in the industry for some time. Given the limitations of “traditional” RAM stemming from using estimations of the likelihood of the consequences of hazards solely based on data of the past, evidence, or practical knowledge of the past, it is of great importance to expand the approach and consider the likelihood of future climate hazards on the basis of future climate analysis based on established models. Nonstationary processes in climate change have already imposed new challenges to CIs, due to more frequent and intense extreme events with alarming prospects, we propose a risk quantification process based on the Joint Research Center (JRC) risk assessment approach for CI protection [38], which takes into account educated estimations of future climate conditions. The process can be used to identify potential hazards, demonstrate their consequences, and assess the risks imposed on technological assets, operations, and human resources’ OSH.

The proposed process/methodology is the result of action research carried out in a specific oil refinery installation in Greece. For the estimation of the likelihood of extreme weather events, the climate risk assessment relied on the availability of accurate information on climate dynamics, as provided by the downscaled Weather Research and Forecasting Model (WRF), a very high resolution climate model [39,40], which can be easily modified appropriately for different geographical contexts. In this context, the 4.5 and 8.5 Representative Concentration Pathway (RCP) of the Intergovernmental Panel on Climate Change (IPCC) future climate scenarios were developed for the specific region where the facilities of the focal organization are located, over the mid-term period up to 2050. These are pathways for greenhouse gas and aerosol concentrations, together with land use change. Each RCP provides only one of many possible pathways to that level of radiative forcing. RCP 4.5 is an intermediate pathway for stabilization of radiative forcing by 2100 at approximately 4.5 W/m² [41], while RCP 8.5 is the high-emissions scenario that is frequently referred to as “business as usual” and represents an average expected future and a worst-case scenario [42]. The period up to 2050 corresponds to the remaining lifetime of the majority of process equipment, before it has to be restored or even shutdown due to environmental legislation [43]. The new estimated likelihood approach, which takes into account climate change to calculate the initial probability of a climate risk scenario, was incorporated into the existing risk assessment methodology to increase the resilience of operations, explicitly taking into account future risks of the infrastructure [30].

In line with the European Green Deal initiatives to obtain climate-neutrality by 2050, the suggested modified risk assessment methodology could support oil companies, operators, stakeholders, as well as governments and local authorities in the assessment and management of climate change risk, towards a more sustainable, efficient, and resilient energy sector.

Following in the paper, we present the generic methodological approach, as it was developed through action research in a specific organizational setting. We briefly present the background concepts related to climate-related risk mapping and we end with the calculation of risks for each company asset. Then, we construct risk matrices using both the conventional empirical risk approach RAM and the proposed CRAM approach based on climate scenarios. We compare results over past (1980–2004) and future periods (2025–2049), exploring at a first level how the vulnerability of petroleum facilities are going to change in a future climate. Lastly, we suggest that urgent action and adaptation measures are needed for the specific company and for the industry as a whole, and, the conclusions of the research are provided in the final section.

2. The Overall Methodology: Structure and Development

The process/methodology developed builds upon the common understanding of risk management, as in the six working steps of the National Infrastructure Protection Plan Risk Management Framework (NIPP) of the U.S. Department of Homeland Security [14]. It is compatible with several National Risk Assessments, the EPCIP program [11], the IPCC report [44], the Sendai Framework with Disaster Risk Reduction [45], and with related International standards, e.g., International Organization for Standardization (ISO) 31000 Risk Management [46]. The NIPP framework comprises the following six steps:

• Setting of security goals and risk management strategy;
• Identification of assets, systems, networks, and functions;
• Assessment and evaluation of risks;
• Prioritization of risks for elimination, mitigation, management, etc.;
• Selection and implementation of protective programs; and
• Measurement of effectiveness.

From the above tasks, the overall climate risk assessment process/methodology developed pays particular attention to the identification of assets, systems, networks, and functions for the oil industry, to the assessment and evaluation of risks stemming from climate change, for the specific company for the particular region, and lists potential adaptive measures to be considered by the company and oil industry as a whole. The process was developed through action research using the method of Checkland and Holwell (1998), based on the Failure Modes Analysis (FMA) framework (Framework of ideas, Methodology, Area of concern) [47]. The area of concern was to provide a means to record and assess future climate risks for oil industry installations using a more educated, not purely practical, method; the methodology used the development of a case study; and the framework of ideas, beyond risk identification and assessment, came from climate scenario development and analysis.

The overall process followed is depicted in Figure 1. First, for every asset, the identification of the hazards’ quantitative critical values (design thresholds) for the asset was carried out. These thresholds are certain values, above or below which, the likelihood of climate impact is considered sufficient to render the asset, or operation, vulnerable [48]. They are mainly linked to the asset and operational attributes and are based on historical operation data, or design/structural operating parameters.

Then, the estimation of the likelihood of a hazard, including climate hazards, occurring followed, and its probability of occurrence affecting the assets, employees, and operations of the organization was determined. As was already mentioned, high resolution climate models were used to determine the future hazard, its probability of exceedance, and/or the return period of a hazard (likelihood of occurrence).

Then, the estimation of consequences/impacts focused on the impacts of the climate hazards on the assets of the organization at risk, and may concern injuries, deaths, losses, and damage to equipment and structures. To calculate risks, information concerning the consequences of the climate events was collected and analyzed on the basis of the assets at risk. This information was obtained from the literature and discussed and validated at a corporate-wide level in interviews and personal discussions with managerial and technical personnel. Risks for exposed assets such as structures, operations, and human resources expected to be affected by extreme weather, as a result of climate change were calculated, and the corresponding assets were placed in the appropriate risk-signifying cells of the CRAM.

Risks were quantified following the standard ISO 31000 approach, according to which risks are defined as the combination of the consequences of a hazard and the associated likelihood of its occurrence [49].

$$Risk=Likelihood \times Consequences$$

(1)

To develop and test the methodology, three risk matrices were constructed. One was similar to the method used by the company (conventional Likelihood-Impact or RAM), one using climate scenarios (CRAM) for the same (past) period as for RAM to compare the two approaches, and one taking into account future climate scenarios using the different climate models. In all three cases, the same constructs were employed, and the same methodological steps were followed.

In the context of action research, a validation process of the final matrices constructed based on questionnaires, interviews, and technical meetings, was undertaken. Researchers from two academic institutions, as well as managers, operators, and technical personnel of the company, and external stakeholders were actively involved. Feedback was related to important aspects of the company’s processes and provided information on how the organization deals at present with the impacts of extreme weather.

3. Background: The CRAM Description

3.1. Likelihood of Occurrence of Hazards Stemming from Extreme Weather

Likelihood refers to the initial probability for a climate scenario to occur and is usually defined as the inverse of the frequency of one or more incidents happening at various time scales (as defined by several European countries in their National Risk Assessments (NRAs), or as the probability of occurrence within a single year [33]. The levels of likelihood were defined by internationally accepted descriptive terms, classified into a set of five categories from the NRAs and IPCC, and correspond to numerical values (

Table 1. Classifications of likelihood considering climate change [50] and under RAM approach [24].

Classifications of Likelihood
	Very Low	Low	Medium	High	Very High
Likelihood/CLASS	1	2	3	4	5
CRAM Return Period	Occurs less than once in 100 years	Occurs once in 50–100 years	Occurs once in 10–50 years	Occurs once in 1–10 years	Occurs more than once in 1 year
or
Probability of occurrence	Probability within 1 year: 0.005% to 1%	Probability within 1 year: 1% to 2%	Probability within 1 year: 2% to 10%	Probability within 1 year: 10% to 100%	Probability within 1 year: 100% +
	1-A	2-B	3-C	4-D	5-E
RAM	Never heard of in the industry	Heard of in the industry	Has happened in corporate level or more than once a year in the industry	Has happened at the installation or more than once a year in corporate level	Has happened more than once a year at the installation

). In our case, the likelihood/class was estimated through the calculation of the probability of occurrence, which is related to the estimated return period (1/Return period). In this respect, the RAM approach differs from the CRAM approach in that it bases the likelihood of events happening on observed/realized consequences of events (posteriori estimations), not on probabilities based on essentially forecasting models (a priori estimations).

3.2. Classification of Consequences

In risk management, consequences (or impacts) are defined as a quantifiable measure of the damages, disruptions and performance degradation of assets/resources, or societal functions, in a wider context. In the study presented in this paper, for the specific categories of the consequences of extreme weather events we used the conventional RAM approach [24,37], as it was being applied by the case organization. In this line,

Table 2. Τhe four consequences categories: people, environment, assets, and reputation (PEAR).

Impact Categories	People	Assets	Environment	Reputation
Severity Level
1	Very mild injury	Slight damage	Barely noticeable	Barely noticeable
2	Mild injury	Minor	Minor	Minor
3	Serious injury	Moderate	Moderate	Moderate
4	Permanent total disability	Major	Major	Major
5	Fatality	Massive	Massive	Massive

presents the consequences’ severity levels on a scale of 1 to 5, for the four impact categories of people, environment, assets, and reputation (PEAR).

3.3. Representation of Risks: The Risk Matrix

With the likelihood and consequences being classified into five distinct categories/levels each, for both RAM and CRAM, clearly, a matrix consisting of 25 cells (five levels of likelihood by five levels of severity/impact) can be built for representing the different categories/intensities of overall risk (

Table 3. Risk Matrix approach along with their risk levels classified into 3 levels, low risk (blue), medium risk (yellow), and high risk (red).

	Likelihood
Consequences		1-A	2-B	3-C	4-D	5-E
	1	1	2	3	4	5
	2	2	4	6	8	10
	3	3	6	9	12	15
	4	4	8	12	16	20
	5	5	10	15	20	25

). As is common practice in the focal organization of our research, the matrix was partitioned into three areas, each area representing a specific level of risk: low risk (blue), which is considered acceptable; medium risk (yellow), which may, or may not, be acceptable; and high risk (red), which is considered unacceptable. By placing it in the appropriate cell, this matrix allows every critical asset and process to be characterized with respect to risk.

4. Climate Risk Assessment of an Oil Processing Organization: Application of the Methodology and the Use of CRAM

4.1. The Focal Organization

The site of the action research was a typical privately-owned oil processing and distribution company with a refining capacity of around 3.5 million m³ crude oil per year. The facilities are located on the western coastline of Attica, 35 km from the center of Athens. The company is one of the leading energy groups in South East Europe, with activities spanning across the entire energy value chain. It has a presence in six countries, and supplies energy resources to light and heavy industry, marine industry, power plants, agricultural facilities, households, as well as commercial buildings and infrastructures.

The company’s operations infrastructure consists of refineries with a significant number of main distillation units, conversion and cooling units, production, storage and distribution facilities and systems, a large wholly-owned port, oil terminals, an extensive crude oil distribution pipeline network from and to the crude oil unloading and storage installation, a distribution pipeline for finished and semi-finished products, and a distribution pipeline to the International Airport of Athens.

Previous studies have already investigated the impacts of climate change on the global and regional level (in particular, the West Attica region where the case company is located), indicating that the oil sector’s critical processes and assets are highly exposed [5,51]. Drawing on these studies, a list of the most critical processes/assets of the case company, with regard to the quantitative critical values (thresholds) of the climate parameters was constructed and is depicted in

Table 4. Design thresholds of the most exposed critical operations and assets of the case company as listed in literature and as provided by the company’s engineers.

Assets/Processes	Design Thresholds
Pipelines (plastic)	TX = 38 °C
Cooling system/Tower	TG = 32 °C/TX = 38 °C/RH = 90%
FCC unit	TX = 38 °C
Refinery processes	TX = 40 °C/TN = −5 °C
Water/Wastewater biological treatment	TX = 45 °C
OSH (occupational safety and health)	HI = 41 °C/WSavg = 17 m/s
Coolant lines/Pipelines	TN = 0 °C
Valves	TN = 0 °C
Hydrogen plant	TN = 0 °C
Distillation equipment	TN = 0 °C
Vessels	TN = 0 °C/TN = −28 °C
Fluidic Catalytic Cracking (FCC) Unit	TN = −7.2 °C
Pipelines (plastic)	TN = −28 °C
Electricity lines	WSmax = 30 m/s
Industrial buildings	WSmax = 30 m/s/WSmax = 50 m/s
Ship docking platform	WSmax = 12 m/s
Tower crane	WSavg = 20 m/s
Storage tank	WSavg = 20 m/s
Transit operations	WSmax = 22m/s/PR = 100 mm
Shipping	WSmax = 26 m/s
Ports	WSmax = 26 m/s
Roads	PR = 150 mm

and in Section 4.2. The climate parameters and indicators are explained in Section 4.2 and

Table 5. Likelihood and impact analysis of the most exposed operations and assets of the case company under the climate change context for the past (1980–2004) and future (2025–2049) periods, for RCP 4.5 and 8.5.

		Return Period			Likelihood of Occurrence				Consequences
Facility Assets	Threshold	Past	RCP 4.5	RCP 8.5	RAM	Past	RCP 4.5	RCP 8.5	Description	Max Level
Pipelines (plastic)	TX = 38 °C	2.2	1.5	1.8	5	4	5	5	Above 38 °C, there are sanctions and ban of use of plastic pipes [58]	2
Cooling system/Tower	TG = 32 °C	3.1	2.2	2.6	5	4	4	4	The capacity to cool down the cooling water is decreased 1 In the past, installations decreased their activity level1	1
	TX = 38 °C	2.2	1.5	1.8	5	4	5	5		1
FCC unit	TX = 38 °C	2.2	1.5	1.8	5	4	5	5	Reduced capacity of 5–10% may occur with air temperatures > 38 °C 1	1
Refinery processes	TX = 40 °C	5.1	4.4	5.9	4	4	4	4	High temperatures cause a need to reduce capacity by 1–2% 1	3
Water/Wastewater biological treatment	TX = 45 °C	42	61	114	4	3	3	3	Efficiency might be decreased by reducing bacterial floc formulation 1	3
OSH	HI = 41 °C	1.3	0.5	0.8	5	5	5	5	Dangerous conditions with great discomfort and health issues [59]	2
Refinery processes	TN = −5 °C	6.2	41	11	4	4	3	4	In the past, refinery installations decreased their activity level1	3
Coolant lines/Pipelines	TN = 0 °C	1.0	2.2	2.1	4	5	4	4	Low temperatures are responsible for freezing and interrupting services [58,60,61,62]	1
Valves		1.0	2.2	2.1	4	5	4	4		1
Hydrogen plant		1.0	2.2	2.1	4	5	4	4		1
Distillation equipment		1.0	2.2	2.1	4	5	4	4		1
Vessel	TN = 0 °C	1.0	2.2	2.1	4	5	4	4	Water at low temperatures, heavy seas (relative to the vessel), and wind that splashes or sprays on a vessel might freeze, resulting in the accumulation of ice on decks decreasing the vessel’s stability [60]	2
FCC unit	TN = −7.2 °C	14.2	148	21	3	4	3	3	Structural damage, loss of equipment, and decreased production [63]	3
Pipelines (plastic)	TN = −28 °C	*	*	*	2	**	**	**	Low temperatures present numerous challenges related to operation of equipment, systems, structures, pipelines, vessel maintenance, and safety equipment, as well as performance of personnel [58,62]	4
Vessel		*	*	*	2	**	**	**		4
Electricity lines	WSmax = 30 m/s	41	60	51	4	3	3	3	Either items thrown into the air, or wind above the structural design threshold of poles can lead to power losses [64]	3
Refinery processes	WSmax = 33 m/s	41	60	51	4	3	2	3	Operations should be avoided beyond this point [65]	3
Industrial buildings	WSmax = 50 m/s	*	*	*	3	**	**	**	Total failures might occur > 50 m/s	3
Ship docking platform	WSavg = 12 m/s	0.2	0.1	0.1	5	5	5	5	Ships might not be able to approach to dock or moor on platforms 1	2
OSH	WSavg = 17 m/s	0.9	0.9	1.1	5	5	5	5	Review of work at height and outdoor work to ensure safety [66]	2
Tower crane	WSmax = 20 m/s	0.8	0.6	1.1	5	5	5	5	High risk of loss of asset [67]	2
Storage tank	WSmax = 20 m/s	0.8	0.6	1.1	5	5	5	5	Failure of the insulation of the storage tank roof might occur 1 [68]	2
Transit operations	WSmax = 22 m/s	1.7	1.6	2.3	5	5	5	4	The critical thresholds for most transit operations are 13.5 m/s for moderate risk and 22 m/s for severe risk and significant impact to safety and transit operations. Roadway damage, flying debris, traffic slowdowns, congestion, and operational delays might occur [27,58]	2
Shipping	WSmax = 26 m/s	8.3	10	11	4	4	4	4	Difficulty of tankers to approach, interruption of fuel deliveries, postponement of pumping, interruption of construction and maintenance [58]	2
Ports		8.3	10	11	4	4	4	4		2
Transit operations	PR = 100 mm	13	65	66	4	4	3	3	Reduced visibility and flooding of passageways [69]	2
Roads	PR = 150 mm	89	1234	1229	3	3	2	2	Road constructions might fail, and roads might be inundated. In 2007, the refinery faced an extreme precipitation event (with associated effects in its activity levels), which caused an unexpected increase in the water level, with inundation depth in the facilities >1.5 m 1 [67]	3
Cooling system/Tower	RH = 90%	3.1	4.4	7.8	5	4	4	4	The cooling capacity of the tower might be decreased 1	1

* The calculated number of the return period is too high. ** The likelihood of occurrence of that event is not expected as a probability. ¹ Oil facility installations’ info and technical specifications.

below.

4.2. Data Sources and Climate Parameters Analysis for CRAM

For constructing CRAM, climate simulations with the Advanced Weather Research and Forecasting (WRF-ARW) (v3.6.1) model [52], forced by the European community Earth-System Model (EC-EARTH), which was dynamically downscaled to the region of Greece at a scale of 5 × 5 km² [39,40], were performed for the periods 1980–2004 and 2025–2049, for the two RCPs (4.5 W/m² and 8.5 W/m²). The efficacy of the WRF to estimate satisfactory climate conditions in Greece has already been assessed and documented [39,40,53,54,55]. Further statistical analyses concerning the performance of these simulations was performed using observations of the nearest station of the Hellenic National Meteorological Service (HNMS). They produced high correlation coefficients, index of agreement (IOA), and statistical significance of the Pearson correlation using a t-test at the 95% confidence interval [51].

The site-specific climate parameters and indicators considered, which are based on previous studies [5,51] and concern directly the assets, processes, and human resources of the oil industry, are presented in Table 4. Climate analysis with the use of WRF model, for the period 2025–2049, showed an increase in the values of maximum (TX), minimum (TN), and average (TG) temperatures. It also depicted a minor increase in the maximum (WSmax) and average (WSavg) wind speeds and a decrease in the daily accumulated precipitation values (PR).

In addition, projections of the Humidex Index (HI) were calculated to assess the comfort/discomfort conditions of human resources. The HI index is a nominally dimensionless number based on air temperature and humidity [56]. The calculations showed an increase in the daily HI index values over the future period reaching values in the dangerous zone, 40 ≤ HI < 55 [56].

Moreover, in order to determine the probability of occurrence of the particular climate parameters and indicators and their likelihood class (Table 5), the changing frequency of the future extremes was quantified. The method for calculating return periods and the probability of exceeding a value was based on the Extreme Value Theory (EVT) [57], using the peaks-over-threshold method, with the use of a high specified threshold. As an example, a graphical solution of the TX EVT is presented in Figure A1 (Appendix A).

4.3. Design Thresholds and Impact Analysis

Table 5 depicts the data used for the analysis of the impacts: the return periods of the design thresholds of climatological parameters, their likelihood (probability of occurrence) in certain climate conditions, indicators that reveal threshold phenomena, and the consequences in case of exceedance of thresholds as found in the literature and as provided and finally qualified by the focal organization through the validation process. Among the several climate hazards acting together, those of the highest impact were selected. The likelihood classification of Table 1 was used to match the estimated initial probability of a risk scenario to occur on a scale of 1 to 5 (from very low (A) to very high (E)). Moreover, the four impact categories (people, environment, assets, and reputation (PEAR)) of Table 2 were used to rank the consequences of the climate hazards in the oil facility on a severity level of 1 to 5.

Having determined the likelihood of occurrence and the consequences of threshold exceeding climate events for every asset and for each approach, risk values were calculated using equation 1. Risks were then characterized (Low, Medium, and High) according to the convention depicted in Table 1 (values and symbols in the eight rightmost columns of

Table 6. Multi-hazard analysis and risk score of the oil facility located in Greece. The risk level corresponds to a different color classified into 3 levels, low risk (blue), medium risk (yellow).

Oil Facility		Likelihood of Occurrence				Consequences	Risk Score (Likelihood × Consequences)
Assets/Processes	Threshold	RAM	History	RCP 4.5	RCP 8.5	MAX level	RAM	History	RCP 4.5	RCP 8.5	RAM	History	RCP 4.5	RCP 8.5
Pipelines (plastic)	TX = 38 °C	5	4	5	5	2	10	8	10	10	M	M	M	M
	TN = −28 °C	2	**	**	**	4	8	NR	NR	NR	M	NR	NR	NR
Cooling System/Tower	TG = 32 °C	5	4	4	4	1	5	4	4	4	M	L	L	L
	TX = 38 °C	5	4	5	5	1	5	4	5	5	M	L	M	M
	RH = 90%	5	4	4	4	1	5	4	4	4	M	L	L	L
FCC unit	TX = 38 °C	5	4	5	5	1	5	4	5	5	M	L	M	M
	TN = −7.2 °C	3	4	3	3	3	9	12	9	9	M	M	M	M
Refinery processes	TX = 40 °C	4	4	4	4	3	12	12	12	12	M	M	M	M
	TN = −5 °C	4	4	3	4	3	12	12	9	12	M	M	M	M
	WSmax = 33m/s	4	3	2	3	3	12	9	6	9	M	M	L	M
Water/Wastewater biological treatment	TX = 45 °C	4	3	3	3	3	12	9	9	9	M	M	M	M
OSH	WSavg = 17 m/s	5	5	5	5	2	10	10	10	10	M	M	M	M
	HI = 41 °C	5	5	5	5	2	10	10	10	10	M	M	M	M
Coolant lines/Pipelines	TN = 0 °C						4	5	4	4	L	M	L	L
Valves		4	5	4	4	1	4	5	4	4	L	M	L	L
Hydrogen plant							4	5	4	4	L	M	L	L
Distillation equipment							4	5	4	4	L	M	L	L
Vessel	TN = 0 °C	4	5	4	4	2	8	10	8	8	M	M	M	M
	TN = −28 °C	2	**	**	**	4	8	NR	NR	NR	M	NR	NR	NR
Industrial buildings	WSmax = 50 m/s	3	**	**	**	3	9	NR	NR	NR	M	NR	NR	NR
Ship docking platform	WSavg = 12 m/s	5	5	5	5	2	10	10	10	10	M	M	M	M
Tower crane	WSmax = 20 m/s	5	5	5	5	2	10	10	10	10	M	M	M	M
Storage tank	WSmax = 20 m/s	5	5	5	5	2	10	10	10	10	M	M	M	M
Shipping	WSmax = 26 m/s	4	4	4	4	2	8	8	8	8	M	M	M	M
Ports							8	8	8	8	M	M	M	M
Electricity lines	WSmax = 30 m/s	4	3	3	3	3	12	9	9	9	M	M	M	M
Transit operations	WSmax = 22 m/s	5	5	5	4	2	10	10	10	8	M	M	M	M
	PR = 100 mm	4	4	3	3	2	8	8	6	6	M	M	L	L
Roads	PR = 150 mm	3	3	2	2	3	9	9	6	6	M	M	L	L

** The likelihood of occurrence of that event is not expected as a probability, NR, not estimated risk.

below).

5. Construction of Risk Matrices, Assessment, and Comparison of Methods

Following the procedures described above and after carrying out the calculations as depicted in Table 6, a risk matrix, using the extant RAM methodology and two climate risk matrices (CRAM) considering the likelihood categories of the climate events, was developed. One of the two CRAM referred to the past period (1980–2004) for comparison with RAM, and one referred to the future (years 2025 to 2049) to identify risks due to climate change dynamics. The latter matrix was constructed on the basis of climate change projections as calculated by the two models (RCP 4.5 and RCP 8.5). A comparison of the RAM and CRAM methodologies is shown in

Table 7. Comparison of the risk levels (low risk (blue), medium risk (yellow), and high risk (red)) of the exposed assets between RAM and CRAM for the past period.

Oil Processing and Distribution Organization RAM (Past)							Oil Processing and Distribution Organization CRAM (Past)
	Likelihood								Likelihood
Consequences		1	2	3	4	5			1	2	3	4	5
	1				Coolant lines/Pipelines Valves Hydrogen plant Distillation equipment	Cooling system/Tower	Consequences	1				Cooling system/Tower	Coolant lines/Pipelines Valves Hydrogen plant Distillation equipment
	2				Vessel	Pipelines (plastic)		2				Pipelines (plastic)	Vessel
	3			FCC unit Industrial buildings	Water/Wastewater biological treatment Electricity lines			3			Water/Wastewater biological treatment Electricity lines	FCC unit
	4							4
	5							5

.

Table 7 presents only the assets whose risk score differs between RAM and CRAM, both applied using historical data. In general, the comparison of the two matrices depicted:

• A decrease in the number of the exposed assets belonging to the medium (M) risk level when the climate approach (CRAM) was used with data of the past (one asset), compared to the existing RAM approach (four assets).
• The Water/Wastewater biological treatment processes, the assets of Pipelines (plastic) (under high temperatures), and Electricity lines (under extreme wind) exhibited the highest risk score in RAM, whereas Vessels (under low temperatures) and FCC units (under low temperatures) were additional assets with high risk scores according to the climate approach (CRAM).
• Based on CRAM, increased risk (from L to M) was anticipated for the assets of Coolant lines/Pipelines, Valves, Hydrogen plant, Distillation equipment, and Vessel (under low temperatures).

Overall, lower risk scores and rankings were estimated using CRAM for the assets of Pipelines (plastic), Water/Wastewater biological treatment, Electricity lines, and the Cooling system/Tower. This was mainly due to the decreased likelihood of occurrence of associated disturbing events. In the RAM approach, there was overestimation of the probability of occurrence of rare hazards that impact assets. Extreme events that “already had been heard of” happening in the global industry, or particularly in the oil industry, even if their return period is not expected as a probability, and (logically) were not expected under the climate risk approach, were evaluated as more probable to occur under the RAM approach. Therefore, industrial buildings, which have been found to be highly exposed under extreme wind speeds (>50 m/s) [64,70], were considered as an optional risk only by RAM. This is due to the different calculation process, which is based on the likelihood of the particular consequence under consideration, and not on the likelihood of the hazard being realized or the incident occurring. Hence, it overestimates the probability of hazard happening.

In general, both approaches indicated an absence of high (H) risks for the present and the future, for the organization (Table 6 and Table 7). The analysis indicated that the FCC unit, Pipelines (plastic), as well as the operations of the Refinery processes, are the most exposed assets, having the highest risk score of multi-hazard consequences, particularly under high and low temperatures and extreme winds [66]. In these conditions, particular attention should be paid to the outdoor working conditions of employees.

Employees are the most important asset exposed to the climate change effects in industrial sites [71], in danger of occupational illnesses and injuries. For instance, heat stress can lead to heat exhaustion, heat cramps, and even to heat stroke. Extreme heat can also decrease significantly employees’ morale and their overall performance. The climate-based risk analysis, performed for both past and future periods, estimated risk of the same level, medium (M), for working conditions in high temperatures (Table 6).

As far as the future is concerned, climate analysis indicated an increase in the likelihood of occurrence of average and high temperatures and wind speeds, and a moderate decrease in the minimum temperatures and accumulated precipitation values. These changes may have a marginal real impact on the risk scores for the future, as risk values that are based on the highest possible impact to a particular asset, or the actual changes in their return periods, are too small. If we consider that the impact of exceeding the climate threshold of an asset is more or less constant, the change in risk is mostly related to the changes in the likelihood of occurrence. Thus, risk levels are expected to remain the same for many exposed assets and processes, or even decrease slightly (Table 6). In general, in both cases, refinery processes (under high and low temperatures) belong to the category with the highest risk score. Similarly, an increase in the risk level concerning the Cooling system/Tower and the FCC unit (under high temperatures) is anticipated. On the other hand, a decrease in the risk level, from medium to low, is expected in the future for the assets of Refinery processes (under extreme wind), Coolant Lines/Pipelines, Valves, Hydrogen plant, Distillation equipment (under low temperatures), Transit operations, and Roads (under extreme precipitation events).

6. Climate Change and the Adaptation of the Oil Industry

In recent years, environmental sustainability has been introduced into business discourse; the oil industry in general, and the case company, in particular, has focused their environmental strategies towards mitigation by using more energy-efficient cleaner technologies in their operations. Along the same lines, their energy needs have been fulfilled by the development and use of renewable sources. Regarding continuity and resilience to extreme events stemming from climate change, until now, as the management of the case company indicated, efforts have been concentrated on operational and technological interventions that guaranteed operations continuity, even with reduced capacities. However, the inherent dynamic nature of risks induced by climate change necessitate a review of adaptive measures, as extreme weather events are expected to be more frequent and more intense in the future.

As already mentioned, the climate risk assessment method (CRAM) developed for the specific oil facility indicated an absence of high risks for the present and the future until 2049. However, there are a number of assets and process whose risk is significant. Analysis indicated that the FCC unit, the pipelines (plastic), the refinery processes, along with the occupational safety and health of human resources will be exposed to the multi-hazard consequences of extreme weather, particularly high and low temperatures and extreme winds. Hence, climate-related risk management and adaptation actions for the above, as well as for similar risks, necessitate increased awareness. It also requires increased funding for modifying existing systems and practices, along with measures to identify and assess risks, establish plans to mitigate and cope with risks, and to implement risk management plans [72]. In this direction, monitoring of global good practices and technologies used in the oil industry for avoiding climate hazards is necessary [73].

So far, several technical proposals have been put forward to augment the resilience of oil infrastructure at the global level. They include reviewing of the design of installations located along the coastline, building concrete walls and dikes, taking flood defense measures, improving drainage systems, developing alarm safety flood systems, increasing the storage capacity for vital equipment and supplies, etc. [5,30]. As there is uncertainty concerning climate stability in projections beyond the period 2040–2050 [72], the development and fast deployment of early warning systems, efficient responders’ engagement, as well as establishment of collaboration between stakeholders, along with flexible and robust designs, are of great importance for managing climate risks and adapting to the consequences of climate change [5]. Moreover, monitoring processes, tools to process climate data and understand extremes, as well as active engagement of the scientific community and industry stakeholders are important in the adaptation process [74]. In this direction, the oil industry, for instance, must implement initiatives for continuously monitoring the thermal discomfort in outdoor working conditions and to prevent the adverse impacts of climate change on the occupational safety and health of human resources. As the action research study revealed, additional steps required towards adaptation to the dynamics of climate change could be:

• Relocation of critical assets, such as electromechanical systems from areas that are at high risk of a climate hazard to safer locations.
• Avoidance of construction in areas that are at high risk of a climate event.
• Planting of trees in industrial areas to decrease the urban heat island effect.
• Use of desalination technologies to maintain adequate water supply in refineries.
• Modification, redesign, and improvement of assets and operations to withstand higher values of climate indicators. In that direction:
  • the capacity of storage tanks could be increased;
  • the drainage systems could be further improved;
  • the energy efficiency standards could be upgraded to manage increased demand;
  • the future grid planning should be developed under extreme temperature scenarios;
  • materials that are durable and highly resistant to leaks and ruptures should be used; and
  • road materials, rail equipment, and drainage systems should be replaced with improved, eco-friendly, and resilient to climate extremes materials.

Finally, knowledge exchange with other industries, governments and society in general will contribute to better understanding of climate change risks and will lead to more effective adaptation measures [30].

7. Conclusions

The paper reported the process and results of an action research effort to incorporate future climate scenarios in the risk mapping and assessment of an oil refinery and distribution company’s structure and infrastructure. Climate scenarios developed from simulations of the WRF-ARW (v3.6.1) model for the specific area where the case company is located were used to assess the likelihood of the occurrence of extreme events that could impact the operations of the company. A list of critical assets with their threshold operational values was constructed and the impact of extreme events with respect to these values (exceedance) was determined. A Climate Risk Assessment Matrix (CRAM) based on the likelihood–impact logic was defined to categorize assets with respect to risks associated with climate-induced hazards. The CRAM approach relies on future climate scenarios and provides a framework for taking proactive measures for the consequences of extreme weather events, which are most likely to occur in the future.

The comparison of CRAM with the previous categorization developed empirically, based on the observation/occurrence of weather events and depicted in the Risk Assessment Matrix (RAM), for a past period of twenty-five years, indicated that CRAM is more conservative and sensitive to events, as it is based on “hard” probability calculations not events that actually happened. This is more obvious for assets of medium risk. Assets at high risk were not identified in either method. However, in CRAM applied for the future periods, increased risk was calculated for the refinery processes and the cooling system/tower, as well as for the occupational safety and health issues of human resources.

Even though, at present, the resilience of the focal company, at the corporate level, seems adequate, it has to be clearly re-evaluated in the future, based on scenarios and conditions that will be changing, integrating future climate hazards into its strategic and business continuity plans, in order to improve its sustainability. Thus, additional research and periodic revision of risk maps and recovery processes are required until climate conditions are stabilized at a level that guarantees sustainability. In addition, as this research clearly indicates, in similar organizations in different geographical locations, there is a need to engage multidisciplinary teams of climate specialists, engineers, and organization specialists, to forecast future climate conditions, assess the risks of structure and infrastructure, including human resources, and design appropriate technical and organizational interventions to mitigate these risks.