# 1. Introduction

In the latter part of the 19th century, there was a bitter controversy between Edison’s direct current (DC) and Tesla’s alternating current (AC) [1]. In the end AC technology won out since it could lend itself well to being transformed to lower or higher voltages and shipping massive quantities of energy over long distances. The AC system tentatively emerged the victor due to the fact that there was a lack of applicable power conversion technologies at that time. Recently, the DC power is having a surprising comeback for delivering electricity to customers since the possibility to shift the DC voltage from one level to another and transmit powers at very high voltages via the DC link is most likely to reach the sufficiently reliable stage with the latest energy-harnessing technologies.

It is of interest to identify what benefits can be taken advantage of by being able to adopt the modern innovations in DC end-use technologies [2]. The most outstanding merit would be further improvement of power quality using the fast control characteristics of power converters in voltage and current. Today, telecommunication facilities, semiconductor electronics, computers, delicate motor drive circuits, and the light emitting diode (LED) require the low voltage DC (LVDC) so that they have power supplies which change the high voltage AC to the low voltage DC when first connected to AC. On the contrary, the direct supplies of DC power may ultimately make it possible to eliminate or simplify the power supply circuits used for those loads. That is, reducing losses by the conversions from DC to AC and vice versa may contribute to enhancing the efficiency of both the overall grid and the individual devices [3]. According to Electric Power Research Institute (EPRI), residential, commercial, and industrial facilities in the United States are expected to drastically increase digital electrical loads from 13% of the total energy consumed in 2000 to 50% of the total energy utilized by each sector in 2020 [4]. This trend will obviously provide the DC distribution scheme with a potential to be explored in certain aspects of power systems.

A limited number of references in power systems are available to validate the effectiveness of an LVDC distribution concept. In [5], the authors compared a traditional 20/0.4 kV system with a proposed ±750 V bipolar LVDC distribution system in regard to total system costs.They reported that the DC system would have an overwhelming advantage over a rival by realizing the 12% cost savings. Also, in [6], the authors addressed issues of using the LVDC distribution system to improve the reliability of electricity distribution and the power quality experienced by the customers. In [7], the authors investigated the efficiencies of AC and DC power systems and concluded that further benefits of a DC distribution system would be expected when more than 50% of the loads are DC and DC-DC converters with efficiencies of 95 % or higher are used. In [8] and [9], a DC microgrid system for residential complex and a generic DC distribution system within a building have been envisaged, where in both cases it was emphasized that the energy efficiency would be remarkably fulfilled by replacing the current AC distribution system with the DC distribution system. It was observed in [10] that the implementation of a DC distribution architecture would be a favorable option for offices and commercial facilities because of the significant cost savings incurred by the backup system using a battery block.

This paper mainly focuses on the economic analysis on the assumed scenarios when a medium voltage AC (MVAC) branch is replaced by the LVDC distribution system. Most of all, the viability, flexibility, and applicability of the LVDC distribution network will be clearly demonstrated in terms of economic considerations. This paper starts with an overview of a general-purpose economic assessment tool, net present value (NPV) method, together with an introduction of cost components, i.e. capital investment and operation costs, in Section 2. Some parameters, which are collected to justify portfolios or alternatives of a small distribution network, are briefly outlined in Section 3. The economic evaluation in a distribution planning project is carried out concurrently with the sensitivity analysis to cope with uncertainties of input data and then discussed in Section 4. The last part of this paper is devoted to concluding remarks as well as future research strategies pertinent to the proposed implementation.

# 2. Economic Analysis Method

Two possibilities are compared based on the method of economic analysis for the distribution system between MV main line and coupling points of the customers: MVAC vs. LVDC. To begin with, two cost components for planning projects are discussed: investment and operation costs. The fundamental concept of engineering economics is then reviewed to extend it to applications in distribution planning.

## 2.1 Cost components of projects

There are two essential cost components to determine which one shows an economic supremacy over the other. Depending on the inherent feature of an investment problem, the unreliability cost may be incorporated in the economic analysis model. The investment costs and the annual costs for both technologies are detailed in the following subsections [11].

2.1.1 Capital investment cost

The capital investment of a project includes the equipment costs, costs for transportation, installation, and commissioning of equipment, and land use costs, etc. For an MVAC, the investment cost is typically broken down into the cable and the transformer costs whereas, in case of an LVDC, the investment costs of a distribution transformer, a rectifier, a DC distribution line, DC/AC inverter units, and filters may be taken into account.

2.1.2 Operation cost

The operation cost is represented on an annual basis and encompasses the operation, maintenance, and administration (OMA) cost, taxes, and network losses.

The annual OMA costs for both technologies are estimated at specified percentages of the investment costs. Annual taxes allow for property and income taxes. It should be noted that the taxes are varied according to the tax policies. Plus, the annual network energy losses are given in kWh. To assign a monetary value to these losses, the electricity price on power exchanges is needed in the calculation. Though this value is not fixed, depending heavily on weather conditions or costs of the marginal generating units, the yearly average system marginal price (SMP) may be normally chosen as a base value. For an MVAC distribution system in Fig. 1, the energy losses in the transformer and the AC cable are derived by:

**Fig. 1.**The equivalent circuit of an MVAC distribution network for a loss calculation

where and are the losses of a transformer and an AC cable, ηtx is the efficiency of a transformer, RAC is the resistance per unit length of an AC cable (in Ω/km), is the voltage at a sending end (i.e. it can be specified to be 22.9 kV in Korea) of an MV branch line, and Pload is the real power consumed by end-users.

In a similar fashion, the energy losses for an LVDC distribution system in Fig. 2 are formulated as follows

**Fig. 2.**The equivalent circuit of an LVDC distribution network for a loss calculation

where , , , and are the losses of a DC/AC inverter, a DC cable, an AC/DC converter, and a transformer, respectively. ηinv, ηconv, and ηtx denote the efficiencies of an inverter, a converter, and a transformer, RDC is the resistance per unit length of a DC cable (in Ω/km), and is the DC voltage transformed by the AC/DC converter.

## 2.2 Economic assessment of projects

The methodology adopted for an economic comparison is a discounted cash flow (DCF) analysis [12]. The result from this computation leads to the derivation of the difference between the net present values (NPVs) of the investigated candidates. The NPV is defined as the value of an investment calculated as the sum of its initial investment cost and the discounted value of expected future operation costs. Provided that the NPV must be an ordinary annuity, it has two assumptions: 1) the discount rate does not change and 2) the first operation cost is one period away. When each cash outflow is discounted back to its present value, the calculation of taxes as well as risks of technical obsolescence need to be tackled. To put it more concretely, a net cost of tax effect is known as after-tax cost. After-tax cost can be computed using the following formula:

And depreciation is a non-cash tax deductible expense that saves income tax by reducing taxable income. The amount of tax that is saved by depreciation is known as depreciation tax shield. The formula to compute the depreciation tax shield is as follows:

The mathematical formulation for computing the discounted value of investment’s cash outflows may be expressed by:

where the present value factor (PVF) for an annuity cash flow is defined as [1 − (1 + discount rate)(−lifetime)]/(discount rate) and the asset wears out during its lifetime until its book value reaches the salvage value by the concept of depreciation. In this paper, the straight-line method is employed for the process of depreciation and the risk-free nominal discount rate, including the effect of inflation, is involved in the calculation of the present value factor (PVF) for an annuity cash flow [11].

# 3. Experimental Setup of Hypothetical LVDC Distribution Networks

It is, for a financial analysis, supposed that the total power of an MV branch line is set at 100 kVA and the total length of the MV branch line is fixed at 5 km. Four different MVAC and LVDC distribution system topologies are now chosen to compare, which are depicted in Fig. 3 and Fig. 4, respectively. In all scenarios, the AC/DC conversion is made at beginning of the MV branch line. Case 1 and case 2 consist of one LVDC link which interconnects two separate AC grids. In either case, the customers are connected to a common LVAC network. In contrast, case 3 and case 4 show wide LVDC distribution districts where the DC/AC conversions are made at the customer-ends. For case 1 and case 2, the LVDC distribution structure is all the same, except that singlephase power is delivered to the customers in case 1 and three-phase power in case 2.

**Fig. 3.**Schematic diagram of each MVAC distribution network topology

**Fig. 4.**Schematic diagram of each LVDC distribution network topology

Similarly, only the difference between case 3 and case 4 is that in case 3 single-phase power loads are connected and in case 4 end-users receive three-phase power. Table 1 presents the commonly recognized calculation parameters for economic comparisons. The specifications and costs of all power electronic devices for each scenario are displayed in Table 2 [13].

**Table 1.**Input parameters for economic assessment

**Table 2.**Specifications and costs of power electronic devices

# 4. Simulation Results

The purpose of this section is to provide the decisionmaking information in working on the final determinations of an LVDC distribution system planning project. Furthermore, the sensitivity of the result for each scenario with respect to the variations of the parameters, such as the total load demand and the total length of the refurbished distribution line, is accomplished.

## 4.1 Discounted Cash Flow (DCF) analysis

The investment costs and losses of all power electronic devices for each scenario are found in Table 3. The data for investment cost and annual costs caused by maintenance and losses are combined to elicit the outcome from a DCF analysis using (9), which can be seen in Table 4.

**Table 3.**Investment costs and total losses for each scenario(Unit: million KRW, kWh)

**Table 4.**NB: EBITDA = Loss cost + OMA cost and present value factor (PVF) for an annuity cash flow @ 20 years and 7% = 10.59

It is worthwhile to stress that some general postulates have been made for an accounting activity:

The salvage values of all power electronic devices are zero. The differences in reliability and insurance costs between MVAC and LVDC distribution systems are neglected. The present value factor (PVF) for an annuity cash flow is about 10.59 given that the lifetime and the riskfree nominal depreciate rate are 20 years and 7.0%, respectively.When the difference in NPVs between MVAC and LVDC distribution systems is designated by ΔNPV, it implies that the LVDC distribution system is advantageous if ΔNPV ＞ 0.Since a replacement of converters and inverters in LVDC distribution systems will be performed in 10 years from the initial installment, the present value discounted from total investment costs of converters and inverters during a study period of 20 years is calculated by:

The losses in both technologies are monetized with a yearly average system marginal price (SMP) in 2012 of Korea, 160 KRW/kWh.

The DCF analysis results for each scenario are summarized as follows:

The losses in an LVDC link distribution are the same as those in a wide LVDC distribution district. That is, the loss cost in case 1 is the same as that in case 3 whilst the loss cost in case 2 is the same as that in case 4.The difference in investment cost between the LVDC distribution network and the MVAC distribution network is much greater in an LVDC link type than that in a wide LVDC distribution district. For instance, the difference in investment cost between the LVDC configuration and the MVAC configuration is 88.27 (in million KRW) in case 1 and 80.03 (in million KRW) in case 3. In the MVAC configuration, the investment cost in case 3 increases when the number of transformers increases and the rated capacity of transformers decreases, compared with that in case 1. However, in the LVDC configuration, the investment cost in case 3 is virtually the same as that in case 1 even though the number of DC/AC inverters increases and the rated capacity of DC/AC inverters decreases. Even in case 4, the investment cost of the MVAC configuration exceeds that of the LVDC configuration.The investment, loss, and OMA costs of the LVDC distribution network in case 1 and case 3 are all the same. Nevertheless, the investment cost of the MVAC distribution network in case 3 is relatively higher than that in case 1 and the same is true of the OMA cost. Thus, in the MVAC distribution network, the NPV of case 3 should be evidently higher than that of case 1 since the loss cost in case 3 is equal to that in case 1. Accordingly, the absolute magnitude of ΔNPV in case 1 is greater than that in case 3 since the MVAC distribution network is an economical solution in case 1 and case 3. In fact, ΔNPV is defined as the NPV of the MVAC distribution network minus the NPV of the LVDC distribution network.In the MVAC distribution network, total costs are higher in case 4 than that in case 2. In the LVDC distribution network of case 2 and case 4, total costs are nearly the same. Consequently, the magnitude of ΔNPV in case 4 is undoubtedly greater than that is case 2 since the ΔNPVs in case 2 and case 4 are all positive.When the three-phase power loads are connected, the use of the LVDC distribution network is more cost efficient than the MVAC distribution network, which is certainly observed in the positive ΔNPV of case 2 and case 4.A wide LVDC distribution district layout is less expensive than an LVDC link distribution structure, which is confirmed in the greater ΔNPV of case 4, compared with that of case 2.

## 4.2 Sensitivity Analysis

In this subsection, the influence of further changes in the total length of the MV branch line and the total amount of transmitted power on the preference of an LVDC distribution system is detected. The financial data are adapted to these changes and the losses are recalculated as well.

On one hand, the length of the cable is varied from 5 km to 3 km and 7 km to come up with an idea of the sensitivity of the result in relation with the variation of the distance in the MV branch line.

As portrayed in Fig. 5, the economic value (DCF result) of an LVDC distribution system compared with that of an MVAC distribution system increases with the cable length. In case 2, the LVDC distribution is not a feasible solution for connection of a distribution line shorter than 4.8 km to the MV network because of the negative ΔNPV in this range (break-even distance ≈4.8 km).

**Fig. 5.**Sensitivity of ΔNPV to variation of the length of line

On the other hand, a reduction in the load demand supplied would bring down the loss cost in an LVDC distribution system and have a favorable effect for the LVDC distribution system on the DCF result. In other words, an increase in load demand has a negative effect on the DCF result for an LVDC system compared to an MVAC system. The explanation is the higher losses of an LVDC system, which become more important when the load demand is higher. Besides, the total investment cost for an LVDC system become sharply higher with the transmissible power. The load demand is therefore reduced to 50 kVA and varied up to 200 kVA in each case.

The DCF values in all scenarios are illustrated in Fig. 6. As could be expected, the reduction in load demand consumed by end users has a major positive effect on the DCF result since a growing trend of ΔNPV is noticed as the load demand is decreased. It is even more remarkable that case 4 as a competent enabler of an LVDC system will start to lose much of its economic benefit when the load demand reaches approximately 120 kVA.

**Fig. 6.**Sensitivity of ΔNPV to variation of the load demand consumed

# 5. Conclusion

In this paper, the possibility of replacing an existing MVAC distribution system by an LVDC distribution system has been contemplated on an economic basis. Four types of the LVDC distribution system topology have been suggested to exploit the portfolios of the 5 km distribution network for the 100 kVA load demand. From this perspective, a wide LVDC distribution district configuration connected with three-phase loads is deemed most suitable for the promotion of an LVDC distribution system. Meanwhile, the sensitivity analysis for the uncertainty of technical parameters such as the total length of the MV branch line and the total load demand has been executed. Eventually, it was concluded that an LVDC distribution system must be a competitive option as the length increases and/or the load demand delivered to end users decreases.

It has to be reiterated that the derived DCF result is not a finalized version for the construction of an LVDC distribution network. The determination of unreliability cost, taking account of the outages in both technologies, should not be ignored. In addition, special caution may be paid to the benefit / cost ratio method when the power quality of both schemes is really acknowledged. As renewable resources become more prevalent at a household level and DC gadgets as well as LEDs are widespread, the installation of a localized DC network along with the ongoing evolution of power electronics technologies may finally be a promising solution in the near future.

Aside from the economics of an LVDC distribution framework, operation, protection, reliability, and safety issues should be comprehensively envisioned to feature more prominently.