LIT_072: Luthander, Widén, Nilsson & Palm (2015) — Photovoltaic Self-Consumption in Buildings: A Review

LIT_072: Luthander, Widén, Nilsson & Palm (2015) — Photovoltaic Self-Consumption in Buildings: A Review

Source

https://doi.org/10.1016/j.apenergy.2014.12.028 — original source (opens in a new tab; the file is not redistributed)

Luthander, Widén, Nilsson & Palm (2015) — Photovoltaic Self-Consumption in Buildings: A Review

The canonical review that formally DEFINES PV self-consumption (share of PV production consumed on-site, Eq. 1/5) and self-sufficiency (share of building load met on-site, Eq. 2/6) for a single grid-connected building, and their conversion relationship (Eq. 7: sc/ss = total load ÷ total production). It then synthesises ~15 studies on the two levers that raise self-consumption: battery storage (+13 to 24 percentage points at 0.5-1 kWh per installed kW PV) and demand-side management / load shifting (+2 to 15 pp), with battery+DSM combined reaching +29 to 32 pp.

This is the RT_170 target — the single-building formula that LIT_034 (Afzalan & Jazizadeh) extends to the community scale and that the NI energy skill builds on. Peer-reviewed (Applied Energy), primary read via pdftotext → data_quality: verified. ⚠ The uplift figures are a review synthesis of others’ work (each traceable to a named study in Tables 4-5), not new measurements; and the abstract’s ‘13-24 pp’ battery range is written ‘10-24 pp’ in the Conclusions.

Summary

This review paper summarises the existing research on photovoltaic self-consumption in grid-connected residential buildings and the two main options for increasing it: energy storage (chiefly batteries) and demand-side management (DSM, i.e. load shifting). Its foundational contribution — and the reason it is in this corpus — is Section 2.1, which gives a clean, formal, widely-adopted pair of metrics: self-consumption = the share of total PV production consumed directly on-site (C/(B+C), Eq. 1), and self-sufficiency = the degree to which on-site generation fills the building’s energy needs (C/(A+C), Eq. 2). Both are defined rigorously as time integrals of the instantaneous overlap M(t) = min{L(t), P(t)} of load and generation (Eq. 5-6), extended to include storage via M(t) = min{L(t), P(t)+S(t)} (Eq. 4), and linked by Eq. 7 (sc/ss equals the ratio of total load to total production), which allows conversion between the two metrics when the load/production ratio is known. The typical integration period is one year.

The review then compiles quantitative results from ~15 primary studies (Tables 4-5, almost all European — Germany, Sweden, Spain, Italy, Switzerland, plus one Japanese) showing how much each lever raises relative self-consumption: batteries sized at 0.5-1 kWh per installed kW of PV deliver +13 to 24 percentage points; DSM alone +2 to 15 pp; and DSM combined with battery storage the highest, +29 to 32 pp in three of four results. It also flags the interpretive traps NI must respect — self-consumption falls as PV is oversized relative to load, coarse (hourly) time resolution overstates self-consumption, storage round-trip losses must NOT be counted as self-consumed energy, and the metric depends heavily on the total-load-to-total-production ratio (Eq. 7). It sits directly upstream of the community-scale self-sufficiency formula in LIT_034, and complements the NZ off-grid microgrid cases already in the corpus (LIT_033 / LIT_068) by supplying the definitional backbone rather than site data. It resolves RT_170.

Key claims

- claim: "Self-consumption is defined as the self-consumed part of PV production relative to the total production: Self-consumption = C/(B+C) (Eq. 1), where C is the PV power utilised directly within the building (the instantaneous overlap of generation and load) and B+C is the total on-site PV generation. Self-sufficiency is defined as the same overlap relative to the total load: Self-sufficiency = C/(A+C) (Eq. 2), 'the degree to which the on-site generation is sufficient to fill the energy needs of the building', where A+C is the total electricity demand (the total load; a gross total — A alone is the paper's 'net electricity demand', the non-overlapping part)."
  source_location: "Section 2.1 Basic definitions, Eq. (1)-(2), postprint p.5"
- claim: "Formally, with instantaneous building load L(t) and on-site PV generation P(t), the on-site-utilised power is M(t) = min{L(t), P(t)} (Eq. 3); with energy storage this extends to M(t) = min{L(t), P(t)+S(t)} (Eq. 4), where S(t) is storage power (S<0 charging, S>0 discharging). Self-consumption sc = ∫M(t)dt / ∫P(t)dt (Eq. 5) and self-sufficiency ss = ∫M(t)dt / ∫L(t)dt (Eq. 6), integrated over t1 to t2."
  source_location: "Section 2.1, Eq. (3)-(6), postprint p.5"
- claim: "The relationship between the two metrics is sc/ss = ∫L(t)dt / ∫P(t)dt (Eq. 7) — self-consumption divided by self-sufficiency equals total load divided by total production. This allows conversion between self-consumption and self-sufficiency if the total load and production (or at least their ratio) are given. The typical integration period is one year, which is long enough to take seasonal variations into account and to minimise short-term random fluctuations. For a Net Zero Energy Building (perfect annual balance between generation and demand) self-consumption always equals self-sufficiency."
  source_location: "Section 2.1, Eq. (7) + surrounding text, postprint p.5; NetZEB special case §2.3, p.6-7"
- claim: "Battery storage: most studies used a PV-battery system with a storage capacity of 0.5 to 1 kWh times the installed PV capacity in kW (short-term, sub-daily storage); for these the increase of self-consumption is between 13 and 24 percentage points (abstract and §5). The Conclusions (§7) state the increase 'is between 10 and 24 percentage points for the systems referred to' (⚠ the lower bound differs from the abstract's 13)."
  source_location: "Abstract, postprint p.2; Section 5, p.11-12; Section 7 Conclusions, p.14"
- claim: "Demand-side management (DSM) alone: it is possible to increase the relative self-consumption by between 2 and 15 percentage points with DSM (load shifting), compared to the original self-consumption rate. Results diverge substantially across the studies (Sweden, Spain, Italy) due to climate, household consumption, and how many appliances are treated as shiftable."
  source_location: "Abstract, postprint p.2; Section 5 + Figure 9, p.11-12"
- claim: "DSM combined with battery storage gives the highest uplift: of four results from three papers, three show an increase between 29 and 32 percentage points, whereas one presents an increase of 15 percentage points. The combined approach always improves on battery-only or DSM-only."
  source_location: "Section 5, postprint p.12 (Table 4 / Table 5)"
- claim: "Table 5 (battery-only, uplift vs battery capacity normalised by PV peak power, kWh/kW): Bruch & Müller 0.33→18 pp and 0.67→22 pp; Braun et al. 0.46→10 pp and 0.92→15 pp; Vrettos et al. 0.56→17 pp; Williams et al. 0.64→18 pp; Femia et al. 0.76→17 pp (32 pp combined with DSM) and 0.76→21 pp (29 pp combined); Waffenschmidt 0.91→19 pp; Li & Danzer 0.94→24 pp; Castillo-Cagigal et al. 0.97→20 pp (29 pp combined with DSM); Widén & Munkhammar 1.0→13 pp (15 pp combined); Weniger et al. 1.38→30 pp; Schreiber & Hochloff 1.80→41 pp; Thygesen & Karlsson 4.62/9.25 (24/48 kWh storage)→33 pp. The relationship between normalised battery capacity and self-consumption increase is non-linear but a clearly distinguishable increasing trend."
  source_location: "Table 5, postprint p.16; trend described Section 5, p.11-12"
- claim: "Illustrative single-study results (Table 4): Braun et al. (Kassel, 5 kW PV, lithium-ion) ~35% self-consumption without storage, ~45% with 2.3 kWh, ~50% with 4.6 kWh, 'economically interesting with battery costs below 350 €/kWh'; Bruch & Müller (S. Germany, 6 kW PV, lead-acid) 29% no storage → 47% with 2 kWh → 51% with 4 kWh; Weniger et al. (NE Germany, 3.2 kW PV) ~35% → ~65% with 4.4 kWh; Schreiber & Hochloff (Germany, 4.1 kW PV) 31% → 72% with 7.4 kWh storage."
  source_location: "Table 4 (Braun [57], Bruch & Müller [58], Weniger [67], Schreiber & Hochloff [20]), postprint p.16"
- claim: "Storage costs and efficiencies: a residential battery storage system costs 'from a few hundred up to more than thousand dollars per kWh of storage capacity' (2014-era) — one of its main drawbacks. Batteries have high conversion efficiency but relatively high self-discharge, so are best for balancing daily fluctuations; hydrogen storage (electrolyser → high-pressure tank → fuel cell) has a round-trip electricity-to-hydrogen-to-electricity efficiency of about 36 percent, 'remarkably lower than for batteries', but near-zero self-discharge, suiting longer storage periods."
  source_location: "Section 4.2.1 Residential battery storage, postprint p.9 (cost); Section 4.2 Storage technologies, p.9 (~36% H2 round-trip)"
- claim: "Interpretation cautions for the metrics: (a) increasing PV generation relative to demand always DECREASES self-consumption while self-sufficiency increases or is unchanged (metrics are sensitive to system sizing, Eq. 7); (b) lower (e.g. hourly vs sub-hourly) time resolution ALWAYS overestimates self-consumption because averaging evens out mismatch — sub-hourly data are needed for individual buildings; (c) storage charging/discharging and self-discharge LOSSES must not be counted as self-consumed energy, since that would inflate self-consumption without increasing useful energy."
  source_location: "Section 2.3 Important factors, postprint p.6-7; Section 6 Discussion, p.12-13"
- claim: "Market/context figures (2013): more than 37 GW of PV was installed worldwide in 2013 for a cumulative ~137 GW; the European share of the world PV market fell from more than 70% in 2011 to 28% in 2013, with Asia now the largest share of new installations; average PV system price declined 6 to 7 percent per year since 1998; the global module price index is now less than $1 per W. The vast majority of PV installations are grid-connected, so production need not match local consumption instantly (unlike off-grid PV)."
  source_location: "Section 1 Introduction, postprint p.3"
- claim: "Evidence base and gaps: almost all reviewed papers are from Europe (Germany, Sweden, Spain, Italy, Switzerland; one from Japan), even though Asia is now the largest PV market; almost every study simulates/measures one year at one-hour resolution or finer. The total number of papers is limited; the authors call for more comparative studies with representative building/end-user samples, and for further study of behavioural responses and the aggregate impact of self-consumption on distribution-grid hosting capacity."
  source_location: "Section 5, postprint p.11; Section 6 Discussion + Section 7 Conclusions, p.12-14"

Neobiome Intelligence relevance

This is the document behind RT_170 and it is a definitional / methodological source rather than a data source. Its value to NI is concentrated in Section 2.

• The canonical single-building self-consumption / self-sufficiency formula (D01, foundational). Eq. 1-7 are exactly the metrics the NI energy skill computes, and they are the upstream definition that LIT_034 cites when it lifts self-sufficiency to the community scale. NI should treat this pair as its base definitions: self-consumption sc = ∫min{L,P}dt / ∫P dt (share of PV production used on-site) and self-sufficiency ss = ∫min{L,P}dt / ∫L dt (share of load met on-site), with the storage-aware overlap M(t)=min{L(t), P(t)+S(t)}. Confirming this formula was the entire point of RT_170.
• Eq. 7 is the load-bearing identity for NI’s sizing logic (D01). sc/ss = (total load)/(total production). This means a self-sufficiency number is meaningless without the PV-to-demand ratio, and it lets NI convert between the two metrics — the same dependency LIT_034 stresses (self-sufficiency collapses toward evening; the ratio and time window must always be stated). NI must fix the integration period at one year for headline figures (the paper’s stated standard) and use sub-hourly resolution where possible, because coarser resolution systematically overstates self-consumption.
• Two calibrated uplift levers for the storage/DSM cells (D01 storage). For a grid-connected building, a battery of 0.5-1 kWh per installed kW PV raises self-consumption +13 to 24 pp (Table 5 gives the full non-linear curve vs battery/PV ratio: ~+10 pp at 0.33 kWh/kW rising to ~+30-41 pp above 1.4 kWh/kW), DSM alone +2 to 15 pp, and battery+DSM +29 to 32 pp. These are order-of-magnitude design anchors for how much on-site consumption each lever buys — European, sub-hourly, single-building, so use them as method-illustrative ranges, not NZ baselines.
• Storage-technology reality checks (D01 storage). Battery cost ‘a few hundred up to more than thousand dollars per kWh’ (2014-era; a EUR figure of ‘350 €/kWh’ is the economic-viability threshold in one study); hydrogen round-trip ≈ 36% vs high battery efficiency — so batteries suit daily balancing and hydrogen only long-duration storage. The explicit warning that storage losses must not be counted as self-consumption is a modelling rule NI should encode directly.

I07 (Fulfilment of basic needs) — light feed. Self-sufficiency as defined here (‘the degree to which on-site generation is sufficient to fill the energy needs of the building’) is the metric that operationalises how far a community meets its own energy need on-site; it is the definitional anchor for the energy component of basic-needs fulfilment, though this paper supplies the formula, not fulfilment data.

Scope caveat: every quantitative result is for grid-connected residential buildings in Europe, and self-consumption is only a meaningful objective when there is a grid to export to. For NI’s off-grid / remote scope the self-sufficiency metric (Eq. 2/6) and the storage-sizing curve transfer directly; the self-consumption framing (which is about not exporting to a grid) applies only to grid-tied community configurations. The uplift ranges are non-NZ, non-calibrated.

Research targets

Resolved

• RT_170 (RESOLVED → this page): Luthander, Widén, Nilsson & Palm (2015) “Photovoltaic self-consumption in buildings: A review”, Applied Energy 142:80-94 — retrieved and read in full. Delivers exactly what the target sought: the formal single-building self-consumption (Eq. 1/5) and self-sufficiency (Eq. 2/6) definitions and their conversion relationship (Eq. 7) — the canonical formula that LIT_034 extends to community scale and that NI builds on.

Research gaps

• The paper is entirely European/grid-connected; a national NZ PV self-consumption / self-sufficiency benchmark at sub-hourly resolution is still needed to calibrate the NI energy skill to NZ irradiance and load shapes. This gap is already tracked on LIT_034’s research-targets section (candidate sources: EECA / TIMES-NZ load data, smart-meter datasets) — not duplicated here.
• The fuller Net-ZEB load-matching / grid-interaction indicator taxonomy (Salom, Widén, Candanedo et al. 2011, ref [21], Table 1 — categories I-IV) sits behind this review’s category-I metric choice. Deliberately not backlogged as a research target: NI uses only the self-consumption/self-sufficiency pair, and the wider indicator set (grid-interaction/peak metrics) is out of scope. Retrieve only if NI later needs those additional indicators.

Notes

Peer-reviewed review article, Applied Energy 142:80-94 (doi:10.1016/j.apenergy.2014.12.028); the raw filed here is the Linköping University Electronic Press postprint (identical text, Elsevier copyright, different pagination). Page references above use the postprint’s own table of contents (Abstract p.2, Intro p.3, §2 p.5, §3 p.8, §4 p.9, §5 p.11, Discussion p.12, Conclusions p.14, References p.16), since that is the pagination a reader of the raw will see — the journal version paginates 80-94. Read in full via pdftotext -layout → data_quality: verified (primary read, every figure traceable).

• The uplift figures are a review SYNTHESIS, not new measurements. The +13-24 pp (battery), +2-15 pp (DSM) and +29-32 pp (battery+DSM) ranges are Luthander et al.’s summary of ~15 primary studies, each named and individually traceable in Tables 4-5 (e.g. Braun et al. [57], Weniger et al. [67], Widén [19]). The definitions (Eq. 1-7), by contrast, are the authors’ own canonical formulation. Treat the ranges as method-illustrative European values.
• ⚠ Internal figure discrepancy (recorded verbatim, not reconciled): the Abstract states the battery uplift as ‘13 to 24 percentage points’ while the Conclusions (§7) state ‘10 and 24 percentage points for the systems referred to’. Both appear in the paper; the abstract’s 13-24 is the headline, the 10 lower-bound reflects Braun et al.’s 0.46 kWh/kW → 10 pp row in Table 5.
• ⚠ A+C is a GROSS total, not ‘net’: verified against the raw (§2.1, postprint p.5), the paper writes “The areas A and B are the total net electricity demand and generation, respectively” — so it is A alone (the non-overlapping part) that is the ‘net’ demand; A+C is the total load (the gross demand). The Eq. 1-2 key claim above therefore reads A+C as the total (gross) electricity demand, consistent with Eq. 7 (“when the total load equals the total production”) and the Figure 1 caption (“daily net load (A+C)” is the paper’s loose label for the full daily load curve).
• Currency: the single costed figure (‘economically interesting with battery costs below 350 €/kWh’, Braun et al., Table 4) is in EUR; the general ‘few hundred up to more than thousand dollars per kWh’ is unspecified ’$’ (2011/2014 references, likely USD). Neither is an NZ cost input.
• Prefix/store decision: filed as lit_ / source_type: literature (peer-reviewed journal review), store source-files/01_literature/. Single-file raw → canonical name lit_072_luthander-2015-self-consumption.pdf.

Connections

Links to

Sources (3): LIT_033 · LIT_034 · LIT_068

Referenced by

Sources (2): CR_002 · LIT_077

EDT domains (1): D01: Renewable Energy & Storage Systems

SSI indicators (1): I07: Fulfilment of Basic Needs