*2.3. Maintenance Data*

On each component/element for the OWTs, the maintenance model relays one or more maintenance activities and rounds. To accomplish each maintenance activity, the model takes into account three kinds of assets:


Spare parts and consumables are included in the model by assigning them a delivery time and a cost linked to it. It is also necessary to detach all the maintenance tasks involving traveling to the offshore wind farm, which requires an offshore vessel to transport the maintenance technicians. However, some maintenance tasks require specific capabilities, such as high load capacity. For these maintenance tasks, vessels with additional capacity are required, which creates an additional cost for the LCC chain. All these factors have to be considered within the developed model [5].

#### *2.4. Online Health Monitoring*

OWTs demand appropriate online monitoring, in order to measure the industrial assets in real-time. Therefore, online data management for maintaining OWT is needed, to be exported to monitoring systems (i.e., SCADA, CMS, etc.). This online data measures reliability, availability, and maintenance from the control monitoring room of the OWTs' OEM.

As we can see from the figure above, online asset management data gives robust health monitoring, allowing continuous monitoring of OWTs as well as ensuring that the OEM controls and operates in a cost-efficient and reliable manner, in order to guarantee the lowest LCC of the OWTs.

Description of the Case Studies.

We assume two different maintenance contracts, both lasting 20 years, including transport systems. Each contract carries out a hypothetical O&M strategy.

O&M Strategy 1:

The transport of maintenance crews offshore uses a light vessel (CTV) without access systems (MCA class 2), with 20 knots of cruising speed, a catamaran hull design, 12 personnel and 2 crews needed to operate, and suitable for 10–20 km offshore travels. This vessel has a limitation of 1.5 m in significant wave height, since availability cannot be over 98%.

#### O&M Strategy 2:

The transport of maintenance crews offshore uses an oilfield support vessel (FSV) with 12 knots of cruising speed and 18–68 personnel and crews needed to operate, suitable for long stays offshore up to 5–7 weeks. FSVs have dynamic positioning and access systems suitable for transferring heavier equipment to the OWT, so they can do heavier repair operations than Alternative 1. This vessel has a limitation of 4 m in significant wave height, so it is suitable for year-round maintenance. Hourly operation costs can be summarized as follows [10].

#### **3. Analysis Review**

#### *3.1. Life Cycle Cost Analysis*

Life cycle costs regarding O&M activities related to a general configuration can be calculated considering the following terms:

LCC = Capital Costs (*Ccap*) + Operating Costs (*Cop*) + Cost of Deferred Production (*Cpr*).

These terms have to be calculated yearly and corrected with a discount rate that accounts for inflation, interest rate, and investor risk, as is usual in economic analyses. A more general approach can be formulated as:

$$\text{LCC} = \sum\_{\text{i}}^{\text{N}} \frac{\left(\mathbb{C}\_{cap} + \mathbb{C}\_{\mathcal{O}^p} + \mathbb{C}\_{pr}\right)\_{\text{i}}}{\left(1 + r\right)^{\text{i}}} \tag{4}$$

where "N" is the life of the project in years (20).

This equation also complies with NORSOK O-CR-001 (for systems and equipment) and O-CR-002 (for production facilities). However, since this is an example comparing two different strategies for O&M in offshore wind power, not for equipment or production facilities, an optimum alternative solution will be used.

Now we compute the LCC for two alternatives, that is, for two different O&M strategies and two different transport concepts for maintenance crews:

Alternatives:

The two different maintenance contracts each last 25 years (the minimum life cycle of the OWT). Both alternatives are for an offshore wind location at a distance to the shore of 20 km (10,7238 NM) from where the wind farm is placed (i.e., WindFloat). Each O&M strategy will include different transport systems [11–16]:


**Table 1.** Input data for Alternative 1 using a light vessel (CTV) without access systems.




**Table 2.** O&M Strategy 2 (Alternative 2) using an oilfield support vessel (FSV).

#### *3.2. Assumptions*

The upcoming analysis requires a list of assumptions. The two different strategies will be compared based on the following assumptions. The preventive maintenance program shall be done every 3500 h (2 times/year), taking 2–3 days/WT per year. In this case study, we have N = 20 years of duration of the transport contract and, since this transport alternative is externally hired, capital costs are 0, so:

$$\text{LCC} = \sum\_{\mathbf{i}=1}^{20} \frac{\left(\mathbb{C}\_{op} + \mathbb{C}\_{pr}\right)\_{\mathbf{i}}}{\left(1 + r\right)^{\mathbf{i}}} \tag{5}$$

Operating costs will be divided between preventive and corrective, since both are mandatory and the LCC of each needs different treatment:

$$\text{LCC} = \sum\_{\mathbf{i}=1}^{20} \frac{\left(\mathbf{C}\_{op}^{prv} + \mathbf{C}\_{op}^{corr}\mathbf{C}\_{pr}\right)\_{\mathbf{i}}}{(1+r)^{\mathbf{i}}} \tag{6}$$

Relevant operating costs for comparing both alternatives are due to transportation strategies (including energy/fuel consumption) and man-labor hours. Spare parts, insurance, and other operating costs are considered constant for both alternatives.

We will consider a failure rate that changes with time to be more realistic, since WTs are more likely to fail the older that they get, following the bathtub curve approach:

> *λyear* <sup>04</sup> = 1 ; *λyear* <sup>510</sup> = 0.75 ; *λyear* <sup>1116</sup> = 0.5; *λyear* <sup>1720</sup> = 0.75 *λyear* <sup>2125</sup> = 1 in failures/year.

This equation comprises minor and major failures (needing minor and major repair). In this context, we define failure as an event that prevents the WT from producing energy at all.

Preventive maintenance will be based on planned maintenance rounds, which are also assumed to change with time, and according to the feedback from each settled and applied maintenance program, in order to better optimize O&M strategies with the failure rates:


The cost of man-labor in offshore conditions is considered to be 250 \$/h.

During corrective maintenance, minor failures on each WT will take 1 day to repair (9 h of offshore labor by 1–4-man crews); major failures will take 3 days offshore with accommodation, in 4 shifts (4 × 4-man crews, 8 h each) [12].

#### *3.3. Operational Cost Results*

The hourly costs (costs of operation) of each alternative are shown in the Tables 3 and 4 below:

**Table 3.** Hourly cost for Alternative 1.


Alternative 1 cost of transportation per hour of O&M work is equal to 58.42 \$/h.

**Table 4.** Hourly cost for Alternative 2.


Alternative 2 cost of transportation per hour of O&M work is equal to 5738.8 \$/h.

Both case studies (Alternatives 1 and 2), are calculated using the same distance to the shore, at 20 km, from where the wind farm (i.e., WindFloat) is placed.

*3.4. Cost of Deferred Production*

According to NORSOK O-CR-001 and O-CR-002, the costs of deferred production can be calculated, in general form, as:

$$\mathbb{C}\_{pr} = \lambda \ast p \ast D \ast L \tag{7}$$

where *λ* is the failure rate per year (which is assumed to be varying with time, as stated above), *p* is the probability of interrupted production reduction, D is the duration of production reduction (downtime), and L is the production loss per time unit.


The downtime (D) is the main difference between the two alternatives. Alternative 2 can have a much higher availability and lower downtime. For this, we follow some of the concepts and procedures indicated by [11].

In general, the failure rate during a season (year) can be divided into failure needing major repair (change of rotor blades) and minor repair (change of lubricating boxes):

$$
\lambda^s = \lambda\_m^s + \lambda\_M^S = \frac{1}{MTBF} \tag{8}
$$

We will assume *λ* = *λ<sup>m</sup>* + *λ<sup>M</sup>* = 0.75*λ* + 0.25*λ* failures/year, so 75% of failures are solved with minor repair operations, while 25% need major repair. When considering both major and minor repairs, the repair time per failure MTTR can be calculated as (this downtime includes waiting for the weather window, but does not include queuing, when maintenance crews are not available to repair the failures, or logistics, such as waiting time for spares; these are supposed to be constant in both alternatives):

$$d\_{CM}^{s} = \frac{\lambda\_m^S \ast d\_m^s + \lambda\_M^s \ast d\_M^s}{\lambda^S} = \frac{1}{\mu^S} = MTTR \tag{9}$$

Where *d<sup>s</sup> <sup>m</sup>* is the mean downtime due to failure needing minor repairs, *d<sup>s</sup> <sup>M</sup>* is the mean downtime due to failures needing major repairs, and *μ<sup>S</sup>* is the average repair rate.

For Alternative 1, we will assume that *d<sup>s</sup> <sup>m</sup>* is around 3 days/turbine and *d<sup>s</sup> <sup>M</sup>* is large, in the order of 20 days/turbine, since no major repairs can be done with these vessels. Notice that in this case, we would need another vessel for that purpose (major repairs), which is outside of the scopes of the contract. So, considering the time varying failure rate per year:

$$d\_{CM}^{alt1} = \frac{0.75 \ast 3 + 0.25 \ast 20}{1} = 7.25 \frac{days}{failure} = \frac{1}{\mu^{alt1}} \tag{10}$$

For Alternative 2, we will assume that *d<sup>s</sup> <sup>m</sup>* is around 1.5 days/turbine, since 24 h shifts can be considered, and *d<sup>s</sup> <sup>M</sup>* is in the order of 10 days/turbine, since major repairs can be done with the FSV vessel.

$$d\_{CM}^{alt2} = \frac{0.75 \ast 1.5 + 0.25 \ast 10}{1} = 3.625 \frac{days}{failure} = \frac{1}{\mu^{alt2}} \tag{11}$$

With these assumptions, we can finally obtain an estimate for the costs of deferred production. A more detailed calculation on downtimes, including queuing issues, is discussed in [10], by means of Markov chain models.

The expressive summary for the whole life cycle of the project, comparing the given O&M options, is showed in Table 5 and Figure 4:


#### **Table 5.** Comparison between Alternatives 1 and 2.

**Figure 4.** Cumulative lifecycle costs.

Alternative 1 shows lower overall LCC (less than a million USD); this is mainly because corrective maintenance due to minor repairs is less costly due to the characteristics of the chosen transportation (CTV). The penalization in the costs of major repair operations (120%) is not enough to compensate for the high costs for minor repair of Alternative 2 (FSV).

Deferred production costs are not high enough to be decisive in the selection between alternatives. If this were an oil and gas project, this may have been different.

This way of obtaining LCC leads us to average values. In order to assess the variability of these assumptions and costs, a Monte Carlo simulation can be carried out on the decisive parameters (cost of man-labor, cost of fuel, costs due to major repairs, downtimes, failure rates), assuming a variance of those, with a certain distribution (usually a triangular one, with the mode at the center). After this simulation, we can obtain an estimate of the uncertainty and sensitivity of some assumptions, such as quantities for the obtained LCC or the probability that these are in a certain range, confidence intervals, or any other quantification of uncertainty. This though, is beyond the scope of this article.

#### **4. Conclusions**

The potential impact from maintenance at the operating and logistical level (flexibility, throughput time, quality management, etc.) is considerable, and, therefore, the financial impact of maintenance can be substantial.

This work analyzes decreasing the O&M cost depending upon failure rates, downtimes, the timing needed for each maintenance schedule work activity, and the associated spare part costs.

The O&M cost results proved a great variability in cost of transportation between each alternative. In Alternative 1, the cost of transportation per hour of O&M work is 58.42 \$/h, but for Alternative 2, it goes up to 5738.8 \$/h. In summary, the total O&M cost of transportation per hour of O&M work differs from Alternative 1 to 2 by 5680.38 \$/h, showing that a reachable decrease in O&M cost is highly dependent upon the technical assumptions set into the initial alternative/strategy and on the development of O&M requirement values (parameters and variables), which are key to recreating and covering the full spectrum of each case study.

Availability rises with a higher degree of accessibility and faster transportation times from support organizations. In contrast, the availability itself depends upon the O&M principles (effective working hours scheduled and number of technicians) set in each O&M strategy (Alternative 1 vs. Alternative 2).

In addition, as the cumulative lifecycle cost proves, for almost half of the life cycle (25 years), the costs-discounted are higher for Alternative 2 (using FSV) than for Alternative 1. Therefore, the long-term life cycle (25 years) is more suitable for implementing Alternative 1, as it is more cost-effective. In contrast, it is more suitable to switch to Alternative 2 in order to guarantee major capabilities, as well as the advantage of achieving the access levels needed to efficiently operate.

Increasing the size of OWTs demands a higher robustness of the O&M implementation, in comparison with traditional and conventional offshore wind farms.

Finally, the optimal O&M strategy maximizes availability at the lowest cost by ensuring safety and the best access to offshore wind farms, minimizing unscheduled maintenance activities, and carrying out scheduled maintenance tasks as efficiently as possible, ultimately resulting in the lowest possible LCOE.

**Author Contributions:** Writing—original draft preparation, J.V.T.; reviewing and editing, V.D.-C.; supervision, X.Y. All authors have read and agreed to the published version of the manuscript.

**Funding:** This paper has been partially funded by the Xunta de Galicia ED431C 2021/39 program, and for the open access charge, the Universidade da Coruña/CIUG.

**Institutional Review Board Statement:** Not applicable.

**Informed Consent Statement:** Not applicable.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**

