The Economic Case for Digital Transformers

In today’s evolving electricity grid, transformers face increasing complexity due to renewable energy integration, industrial load variability, and aging infrastructure. Digital transformers, equipped with real-time monitoring and analytics, promise to address these challenges. However, their higher upfront costs raise questions about economic viability. This article explores the cost-benefit analysis of digital transformers, drawing from recent research to highlight their long-term advantages.

The Need for Dynamic Asset Management

Transformers are critical for converting high-voltage electricity into usable power. Yet, modern grid dynamics—such as fluctuating demands in industries like oil and gas, chemicals, and mining—strain traditional management approaches. Without real-time data, operators risk higher failure rates, unplanned outages, and costly repairs. Bhabha P. Das notes in the article Investigating cost-benefit analysis for digital distribution transformers that "the need for dynamic asset management strategies is very evident," especially in applications with large load variability.

Understanding Digital Transformers

Digital transformers are equipped with sensors and processing units that continuously monitor parameters like temperature, hydrogen, moisture, and current (load conditions). Unlike traditional transformers, which depend on periodic manual inspections, digital transformers provide real-time data analysis, early failure warnings, and optimized maintenance (condition-based maintenance strategy), enabling actionable intelligence, e.g., below types of analysis and measurements:

• Thermal,

• Load,

• Aging,

• Harmonic distortion,

• Hydrogen and moisture detection and trending,

• Watch alarms – oil level, tank pressure, voltage, current, oil and winding temperature,

• Ambient temperature.

These capabilities can reduce failures and extend equipment life, but their economic benefits depend on a careful analysis of costs and savings that we will explore through the article.

Economic Variables

Several variables are critical to evaluating the investment in digital transformers:

1. Failure Rates

Transformers follow a "bathtub curve" failure pattern:

• Infant Stage: High failure rates due to manufacturing or installation issues.

• Normal Stage: Low, consistent failure rates during mid-life.

• Wear-Out Stage: Rising failure rates as the transformer ages.

Figure 1. Transformer failure curve [1].

The “normal life” stage for industrial transformers is around 16 years, after which the failure rate starts to increase, as shown in Figure 1. The average failure rate for transformers is 0.5-3% annually. Digital transformers reduce this rate by enabling predictive maintenance, which identifies and addresses issues before they cause failures. The efficiency of a digital transformer can be estimated to be 60%. The calculated improvement is listed in Table 1 and the typical failure rate distribution is shown in Figure 2.

Figure 2. Breakdown of failure probability [2].

Table 1. Improvements made by digitalization.

	Non-digital	Digital	Absolute relative improvement
Failure occurring	70%	28%	Decrease by 86%
Failure prevented	30%	72%	Increases by 82%

"The most recognized benefit of early detection of incipient faults is the major savings that can be achieved on repair costs. In this regard, the purpose of an on-line monitoring system is to prevent major (or catastrophic) failures and convert them into failures that will be repaired at a reduced cost during a planned outage [2]."

2. Reduced Maintenance Costs

The condition-based maintenance strategy enabled by the digital transformer provides early warning signals, allowing optimized operations and data-driven decisions. Digital monitoring cuts dissolved gas analysis (DGA) costs from an estimated R$ 24,000 to R$ 9,600 over 48 years. This specific saving is attributed to the ability of digital transformers to perform continuous oil quality assessments, reducing the need for manual sampling and laboratory analysis. With digital transformers, manual oil sampling can be extended to once in 3 years on average.

3. Reduced Repair or Replacement Costs

When a transformer fails, repair or replacement costs can range from thousands to millions, depending on the damage, which can be non-catastrophic or catastrophic. Digital transformers lower these costs by reducing failure frequency through proactive monitoring and maintenance.

The Annual Replacement Costs (ARC) method is an approach used to estimate the annualized cost of replacing transformers or other equipment over their expected lifespan, factoring in failure rates, replacement costs, and the time value of money. The investment decision in digital transformers can be made by comparing the ARC of a non-digital and a digital transformer. The ARC method calculates the average annual cost of replacing transformers based on:

• Failure Rate (λ): The probability of a transformer failing in a given year, which varies over the transformer’s lifecycle (e.g., 2% in the infant stage, 1% in the normal stage, up to 6.35% in the wear-out stage per Table 7 in [1]).

• Replacement Cost (Cr​): The cost to replace a failed transformer, including purchase price, installation, and downtime costs.

• Number of Transformers (N): The size of the transformer fleet under consideration.

• Expected Life: The nominal lifespan or the effective life adjusted for failures.

• Discount Rate: To account for the time value of money, reflecting the cost of capital or inflation.

The method annualizes the replacement cost by considering the probability of failure each year and discounting future costs to present value, then averaging over the evaluation period (typically the transformer’s life or a standard horizon like 30 years).

The ARC can be calculated using one of two approaches: a simplified annual cost model or a present worth method. We will use the simplified ARC:

Figure 3. ARC simplified formula

The first step is to calculate the current replacement cost (Cr) and account for inflation each year of the expected life by using the Inflation-Adjusted Replacement Cost formula:

Figure 4. Inflation-Adjusted Replacement Cost

Where:

• Cr​ = Current replacement cost.

• i = Annual inflation rate (e.g., 2% or 0.02).

• t = Year of replacement (1 to 30).

4. Downtime Costs

Failures often lead to operational downtime, costing production loss/h in industrial settings, with an average downtime to replace a transformer for non-catastrophic and catastrophic failures of 8 hours and 72 hours, respectively. By preventing failures or enabling faster responses, digital transformers can significantly cut these expenses.

5. Cost of Digitalization

The upfront cost of adding digital features is approximately R$ 90,000 per transformer, as estimated by the research. This additional investment must be offset by the savings from reduced failures, repairs, and downtime to justify the expense. Although not only the above-listed costs are part of the "business case", there are two other important factors: improved loading and life extension.

6. Improved Loading

Deferred upgrade capital costs due to load growth, enabled by the improved real-time transformer loading capability. Therefore, allowing transformers to operate at higher loads (e.g., 2 MVA vs. 1 MVA) based on real-time conditions. This maximizes utilization, reduces capital costs, and maintains reliability. The hot spot temperature and the aging rate at 50% and 60% loading, for a 2 MVA transformer, are shown in Table 2:

Table 2. Hot spot temperature and the aging rate at different loading [1].

Load %	Hot spot temperature (°C)	Aging rate	Expected transformer life*
50%	69.6	0.038	395 years
60%	77.3	0.091	136 years

*Under "free from air" and 0.5% moisture values

The typical desired lifespan of a transformer is 30 years. Therefore, even with an additional loss of paper life resulting from an extra 10% loading, the technical end of the transformer’s life will still be the desired lifespan of 30 years. Consequently, the cost attributed to the loss of life from improved loading will only be applicable if the loading decreases the lifespan below 30 years, or 50 years as a safety margin, i.e., approximately 85°C.

"A digital transformer comprised of load, temperatures, hydrogen, and moisture sensors; can support the end-user to calculate the maximum safe load the transformer can carry from a thermal perspective [1]."

7. Life Extension

On-line monitoring provides visibility of issues earlier (early detection of incipient faults), reducing the risk of unexpected failures and unscheduled outages, thus raising the reliability to a level acceptable. If half of the major failures can be avoided and converted to minor failures, reliability is enhanced and the transformer can be allowed to serve for a few additional years before an unacceptable level of unscheduled outage probability is reached (see Figure 3).

Figure 5. Effects of monitoring on transformer life duration [2]

Under the failure rate curve for non-digital distribution transformers, the acceptable age is close to 25 years, whereas for a digital transformer with 50% efficiency this age is around 30 years. Digital transformers extend life by 5 years (25 to 30 years), delaying replacement costs.

Therefore, there are two types of financial benefits in extending the transformer life: one is the deferred capital expenditure (CAPEX) allowing the capital investment in other areas instead of buying a new transformer; i.e. capital gains; and the other is the extended depreciation period, reducing the annual cost of the asset; i.e. the book value.

"The benefit from deferred replacement is directly proportional to the current interest rate and the capital cost of a new unit [2]."

---

Our Solutions

HV Assets has a complete solution for monitoring transformers using an advanced water, hydrogen, temperature and pressure sensor, the Basic Care Sensor, and the Early Warning Sensor, measuring hydrogen and temperature. Click here to check the technical data.

Basic Care Sensor

Early Warning Sensor

---

Economic Metrics

To analyze the investment, standard metrics are used:

• Net Present Value (NPV): Measures the profitability by comparing the present value of savings to costs. A positive NPV indicates a good investment.

• Internal Rate of Return (IRR): The discount rate at which NPV equals zero. An IRR above the cost of capital suggests viability.

• Payback Period: The time required for savings to recover the initial investment.

• Future Value (FV): The value of a current sum of money or investment at a specified future date, based on an assumed interest rate, accounting for the effects of compounding over time.

• Present Value (PV): The current worth of a future sum of money, discounted back to the present using a specific interest or discount rate, accounting for the time value of money.

These metrics help quantify the long-term economic impact of digital transformers.

Case Study: Economic Impact in Action

Consider a scenario with the following assumptions:

Table 3. Calculation Parameters*

Transformer Data	3 MVA 69kV/13.8kV
A - Cost (manufacturing and installation)	R$ 1,250,000.00
B - Corrected Cost (inflation correction)	R$ 11,921,659.70
C - Expected Life	40 years
D - Extended Life	8 years (total 48 years)
E - Discount rate	12%
F - Inflation rate	5.8%
G - Annual Average Revenue	R$ 800 million
H - Production Hours/day	16 hours

*The parameters should be adjusted to reflect the local economic variables. In this case, Brazil’s costs, in Reais, will be used for evaluation.

• Non-Digital Transformer:

  • Annual normal failure rate: 1% (variable during the infant and the wear out stage). To model the changes in the failure rate during the transformer’s lifetime, a time-varying scaling factor is used based on the proposed model in [5] and failure statistics in [4] and are listed in Table 4.

  • Repair/Replacement costs: R$ 200,000.00-R$ 666,666.67 per failure (non-catastrophic and catastrophic, respectively). The non-catastrophic repair/replacement costs are based on the Corrected Cost (B) of the transformer, considered to be 20% of the Corrected Cost, and the catastrophic repair/replacement costs are based on a Predictive Repair of R$ 26,666.67, estimated by dividing the repair/replacement costs by a 7.5 factor, multiplied by a 25 factor, both factors recommended in [2].

  • Downtime: R$ 1,666,666.67-R$ 15,000,000.00, 8 to 72 hours (non-catastrophic and catastrophic, respectively). The production loss is calculated based on the average annual revenue (G) from a industry or a power utility, related to the transformer outage, divided by the total production hours per day (H) and multiplied by 8 and 72 hours.

• Digital Transformer:

  • Monitoring and installation cost, including the following sensors, hydrogen, moisture, temperature, voltage, current, gateway and energy meter (voltage, current and harmonics): R$90,000,00.

  • Failure rate reduction: 60%

    
    

Table 4. Failure Rates per Year

Year	Stage	Failure Rate
1	Infant	2.00%
2 to 16	Normal	1.00%
17	Wear out	1.03%
18	Wear out	1.06%
19	Wear out	1.10%
20	Wear out	1.13%
21	Wear out	1.17%
22	Wear out	1.20%
23	Wear out	1.24%
24	Wear out	1.28%
25	Wear out	1.32%
26	Wear out	1.36%
27	Wear out	1.40%
28	Wear out	1.45%
29	Wear out	1.49%
30	Wear out	1.54%
31	Wear out	1.59%
32	Wear out	1.64%
33	Wear out	1.69%
34	Wear out	1.74%
35	Wear out	1.80%
36	Wear out	1.85%
37	Wear out	1.91%
38	Wear out	1.97%
39	Wear out	2.03%
40	Wear out	2.09%
41	Wear out	2.16%
42	Wear out	2.23%
43	Wear out	2.30%
44	Wear out	2.37%
45	Wear out	2.44%
46	Wear out	2.52%
47	Wear out	2.60%
48	Wear out	2.68%
49	Wear out	2.76%
50	Wear out	2.85%

The ARC must be calculated for non-digital and digital transformers based on the projected failure rates from Table 4 for each year during its extended life. The Table 5, below, demonstrates the calculation example for the normal stage, using a 1% failure rate. The ARC formula (figure 3) uses the failure probability to calculate the non-catastrophic and catastrophic costs, we will use the failure rates based on the figure 2. Therefore, the following calculations were made:

• Non-Digital

  Non-catastrophic: 1% x 70% x 99% = 0.693%

  Catastrophic: 1% x 70% x 1% = 0.007%

• Digital

  Non-catastrophic: 1% x 70% x 60% x 99% = 0.416%

  Catastrophic: 1% x 70% x 60% x 1% = 0.004%

After the failure rates probability calculation, the total non-catastrophic and catastrophic costs (including production loss and repair/replacement) for non-digital and the total non-catastrophic and catastrophic costs (including production loss, repair/replacement and predictive repair) for digital must be calculated, as below, and multiply by the failure rate:

ARC Non-Digital = Total non-catastrophic costs x 0.693% + Total catastrophic costs x 0.007%

ARC Digital = Total non-catastrophic costs x 0.416% + Total catastrophic costs x 0.004%

Finally, the difference between the Non-Digital and Digital ARCs is the annual benefit provided by installing a monitoring system:

ARC Non-Digital - ARC Digital = Annual Benefit

Table 5. Non-Digital and Digital ARCs

Costs	Non-catastrophic	Catastrophic
Production loss	1,666,666.67	15,000,000.00
Repair/Replacement	200,000.00	666,666.67
Predictive Repair	26,666.67	26,666.67
Non-Digital			Total
Failure Rate	0.693%	0.007%
ARC	12,936.00	1,096.67	14,032.67
Digital			Total	Benefit/year
Failure Rate	0.416%	0.004%
ARC	7,872.48	659.12	8,531.60	5,501.07

In addition, the calculated ARCs (failure reduction benefit) by year must be inflation-adjusted, as below. Afterwards, the next step is to project a Cash Flow, as below Table 6, including the Failure Reduction Benefit, Oil Sampling Reduction, Deferred CAPEX Gains and Book Value. All the yearly cash flows must be converted to Present Value (PV) by using a discount rate to accurate evaluation of the investment.

• Inflation-Adjusted ARC / Failure Reduction Benefit

  The ARC of each year must be calculated using the ARC Non-Digital and ARC Digital to determine the future valeu of the costs, by using the below formula, e.g. second and third years:

  Annual Benefit = ARC Non-Digital x (1 + i)^t-1 - ARC Digital x (1 + i)^t-1

  Annual Benefit (2) = R$ 14,032.67 x (1 + 5.8%)^2-1 - R$ 8,531.60 x (1 + 5.8%)^2-1

  Annual Benefit (2) = R$ 14,846.56 - R$ 9,026.43

  Annual Benefit (2) = R$ 5,820.13

  Annual Benefit (3) = R$ 14,032.67 x (1 + 5.8%)^3-1 - R$ 8,531.60 x (1 + 5.8%)^3-1

  Annual Benefit (3) = R$ 15,707.66 - R$ 9,549.97

  Annual Benefit (3) = R$ 6,157.70

• Oil Sampling Reduction

  R$ 600.00 by samples analysis (including dissolved gas analysis and oil quality analysis)

  Non-digital = R$ 600.00 x 40 = R$ 24,000.00

  Digital = (R$ 24,000.00 x 48) / 3 = R$ 9,600.00 (3 years interval)

  R$ 24,000.00 - R$ 9,600.00 = R$ 14,400.00

  R$ 14,400.00 / 48 = R$ 300.00

  Note: The oil sampling cost should also be inflation-adjusted.

• Deferred CAPEX Gains

  It is calculated by using the Future Value (FV) formula by each year and the accumulated value, e.g., the first and second year, the 41 st and 42 nd years, using the value of a new transformer in the future (parameter B - Corrected Cost). All the years must be calculated until the transformer end of life.

  FV = PV × (1 + r)^n

  FV (41) = 11,921,659.70 x (1 + 12%)^1

  FV (41) = 1,430,599.16

  FV (42) = (11,921,659.70 + 1,430,599.16) x (1 + 12%)^1

  FV (42) = 1,602,271.06

  
  

• Book Value

  Cost of the Transformer = R$ 1,000,000.00 (without installation costs)

  (Extended years / Original years) x Cost of transformer

  8 / 40 x R$ 1,000,000.00 = R$ 200,000.00

  R$ 200,000,00 / 48 = R$ 4,167.67

  
  

• Cash Flow

  Cash Flow = Failure Reduction Benefit + Oi Sampling Reduction + Deferred CAPEX Gains + Book Value

• Present Value

  The Present Value (PV) of each year must be calculated using each Cash Flow to determine the payback, by using the below formula, e.g. first and second years:

  PV = FV / (1 + r)^n

  PV (1) = R$ 15,468.80 / (1 + 12%)^1

  PV (1) = R$ 13,811.43

  PV (2) = R$ 10,304.20 / (1 + 12%)^2

  PV (2) = R$ 8,214.44

  
  

Table 6. Cash flow for economic metrics calculations

Year	Failure Benefit	Oil Sampling	Deferred CAPEX	Book Value	Cash Flow	PV	Cumulative Cash Flow
0	0.00	0.00	0.00	0.00	-90,000.00	-90,000.00	-90,000.00
1	11,002.13	300.00	0.00	4,166.67	15,468.80	13,811.43	-76,188.57
2	5,820.13	317.40	0.00	4,166.67	10,304.20	8,214.44	-67,974.13
3	6,157.70	335.81	0.00	4,166.67	10,660.17	7,587.70	-60,386.43
4	6,514.84	355.29	0.00	4,166.67	11,036.80	7,014.08	-53,372.35
5	6,892.70	375.89	0.00	4,166.67	11,435.26	6,488.68	-46,883.67
6	7,292.48	397.69	0.00	4,166.67	11,856.84	6,007.04	-40,876.63
7	7,715.44	420.76	0.00	4,166.67	12,302.87	5,565.19	-35,311.43
8	8,162.94	445.16	0.00	4,166.67	12,774.77	5,159.52	-30,151.92
9	8,636.39	470.98	0.00	4,166.67	13,274.04	4,786.75	-25,365.17
10	9,137.30	498.30	0.00	4,166.67	13,802.27	4,443.96	-20,921.20
11	9,667.26	527.20	0.00	4,166.67	14,361.13	4,128.48	-16,792.72
12	10,227.97	557.78	0.00	4,166.67	14,952.41	3,837.91	-12,954.81
13	10,821.19	590.13	0.00	4,166.67	15,577.99	3,570.07	-9,384.74
14	11,448.82	624.36	0.00	4,166.67	16,239.84	3,322.99	-6,061.74
15	12,112.85	660.57	0.00	4,166.67	16,940.09	3,094.89	-2,966.85
16	12,815.39	698.89	0.00	4,166.67	17,680.95	2,884.15	-82.71
17	13,982.88	739.42	0.00	4,166.67	18,888.97	2,751.07	2,668.36
18	15,256.73	782.31	0.00	4,166.67	20,205.71	2,627.54	5,295.91
19	16,646.63	827.68	0.00	4,166.67	21,640.98	2,512.66	7,808.57
20	18,163.14	875.69	0.00	4,166.67	23,205.50	2,405.64	10,214.21
21	19,817.82	926.48	0.00	4,166.67	24,910.96	2,305.75	12,519.96
22	21,623.23	980.21	0.00	4,166.67	26,770.11	2,212.35	14,732.31
23	23,593.12	1,037.06	0.00	4,166.67	28,796.85	2,124.86	16,857.17
24	25,742.47	1,097.21	0.00	4,166.67	31,006.35	2,042.76	18,899.93
25	28,087.62	1,160.85	0.00	4,166.67	-46,584.86	-2,740.28	16,159.66
26	30,646.42	1,228.18	0.00	4,166.67	36,041.27	1,892.92	18,052.57
27	33,438.32	1,299.42	0.00	4,166.67	38,904.41	1,824.37	19,876.94
28	36,484.57	1,374.78	0.00	4,166.67	42,026.02	1,759.60	21,636.54
29	39,808.34	1,454.52	0.00	4,166.67	45,429.53	1,698.30	23,334.84
30	43,434.90	1,538.88	0.00	4,166.67	49,140.45	1,640.21	24,975.05
31	47,391.84	1,628.14	0.00	4,166.67	53,186.65	1,585.05	26,560.10
32	51,709.27	1,722.57	0.00	4,166.67	57,598.50	1,532.62	28,092.72
33	56,420.01	1,822.48	0.00	4,166.67	62,409.16	1,482.70	29,575.42
34	61,559.90	1,928.18	0.00	4,166.67	67,654.75	1,435.11	31,010.53
35	67,168.05	2,040.02	0.00	4,166.67	73,374.73	1,389.68	32,400.22
36	73,287.09	2,158.34	0.00	4,166.67	79,612.10	1,346.26	33,746.48
37	79,963.59	2,283.52	0.00	4,166.67	86,413.78	1,304.72	35,051.20
38	87,248.32	2,415.97	0.00	4,166.67	93,830.95	1,264.91	36,316.11
39	95,196.69	2,556.09	0.00	4,166.67	101,919.45	1,226.74	37,542.86
40	103,869.16	2,704.35	0.00	4,166.67	110,740.17	1,190.10	38,732.96
41	113,331.70	2,861.20	1,430,599.16	4,166.67	1,550,958.73	14,882.00	53,614.96
42	123,656.28	3,027.15	1,602,271.06	4,166.67	1,733,121.16	14,848.14	68,463.10
43	134,921.43	3,202.72	1,794,543.59	4,166.67	1,936,834.41	14,815.54	83,278.64
44	147,212.85	3,388.48	2,009,888.82	4,166.67	2,164,656.82	14,784.14	98,062.78
45	160,624.02	3,585.01	2,251,075.48	4,166.67	2,419,451.18	14,753.87	112,816.64
46	175,256.96	3,792.94	2,521,204.54	4,166.67	2,704,421.11	14,724.66	127,541.30
47	191,222.97	4,012.93	2,823,749.08	4,166.67	3,023,151.65	14,696.46	142,237.77
48	208,643.49	4,245.68	3,162,598.97	4,166.67	3,379,654.81	14,669.23	156,907.00

At a 12% discount rate, the payback period is 17 years. This is when the cumulative savings from fewer failures, lower repair costs, and reduced downtime equal the R$ 90,000 additional investment. Depending on the sensor technology, there would be different additional costs, e.g., sensor replacement cost of R$ 80,000 in the middle of the transformer's life (year 25) that was added in this case study. The calculated IRR is 17.72%, above the discounted rate, and the NPV results in R$ 156,907.00, showing that the digital transformer is more cost-effective over its lifespan.

Sensitivity Analysis

The payback period varies with different assumptions:

• At a 6% discount rate, the payback period drops to 10 years.

• At a 15% discount rate, it extends to 31 years.

• With a 2% failure rate, the payback period shortens, as higher failure rates amplify the savings from digitalization.

This analysis highlights how sensitive the investment’s viability is to factors like discount rates and failure rates, helping stakeholders assess risks and benefits under different conditions.

Conclusion

Given the evolving dynamics and escalating complexity of contemporary electricity grids, monitoring becomes paramount to mitigate revenue losses. Even in industrial settings, where the widespread adoption of variable-speed drives and other power electronic-driven machinery underscores the necessity of self-monitored transformers. Digital transformers offer compelling economic advantages by reducing failure rates, replacement/repair costs, and downtime, despite their initial investment. Finally, they improve load management and extend life.

The case study demonstrates a payback period of 17 years at a 1% failure rate and 12% discount rate, with positive NPV thereafter. Sensitivity analysis further shows that higher failure rates or lower discount rates can enhance these benefits. As grids grow more complex, the economic case for digital transformers strengthens, making them a smart investment for reliable, cost-effective operations.

---

Our Solutions

The HV Assets Care Platform is a complete solution for data analysis and diagnostics. It includes all the recommended methods from the IEEE standard, including the Duval Triangle and the advanced Combined Duval Pentagon, integrated in an Asset Management dashboard. It provides a Health Index with individual scores to create an asset ranking. For more information, click here.

---

References

1. Das, Bhaba. (2021). Transformers Magazine, Vol8, Issue 4, 2021 - Cost-benefit analysis for digital distribution transformers - part I and II.

2. "IEEE Guide for Application for Monitoring Equipment to Liquid-Immersed Transformers and Components," in IEEE Std C57.143-2012 , vol., no., pp.1-83, 19 Dec. 2012, doi: 10.1109/IEEESTD.2012.6387561.

3. CIGRE, (2022). Life Extension of Oil-Filled Transformers and Shunt Reactors (TB887).

4. CIGRE, (2024). Analysis of AC transformer reliability (TB939).

5. B. Retterath, S. S. Venkata and A. A. Chowdhury, "Impact of time-varying failure rates on

   distribution reliability," International Journal of Electrical Power & Energy Systems, vol. 27, pp.

   682-688, 2005.