LIT_119: Evans et al. (2015), cold-store specific energy consumption (IIR ICR2015, intl proxy)

LIT_119: Evans et al. (2015), cold-store specific energy consumption (IIR ICR2015, intl proxy)

Source

https://iifiir.org/en/fridoc/31592 — original source (opens in a new tab; the file is not redistributed)

Summary

This IIR ICR2015 conference paper (Evans et al., London South Bank University and partners) reports specific energy consumption (SEC) benchmarks for refrigerated food cold stores drawn from an internet-based survey of a large cross-country dataset (428 stores analysed, or 758 counting a UK sub-survey individually). For the Neobiome project its value is narrow and specific: it is the international peer-reviewed benchmark for the electricity a produce cold store consumes per unit volume, expressed as kWh per m³ per year, split by store type (chilled vs frozen/mixed). It is the international bracket around the NZ-specific coolstore load figures held alongside it (the apple/cold-store 79 kWh/m³/yr anchor and the kiwifruit 0.44 kWh/TE figure).

The headline figures are the chilled all-stores mean of 55.7 kWh/m³/yr (range 4.4 to 250.4) and the frozen/mixed all-stores mean of 71.5 (range 6.0 to 391.6) LIT_119, with much tighter central bands once outliers are trimmed. The paper also states that 60-70% of a cold store’s electricity may be used for refrigeration LIT_119 (attributing Asano & Mugabi 2013) — the coefficient that lets the model split a whole-store meter into refrigeration versus ancillary load. Its Table 1 records a New Zealand cold-storage-industry survey range of 26 to 379 kWh/m³/yr (Werner et al. 2006) LIT_119, the same NZ research group (Merts, Cleland) behind the Frater anchor, so the NZ and international evidence cross-corroborate. Data quality is high but flagged PROXY — the dataset is international (UK/European-weighted), not NZ.

Key claims

- claim: "REFRIGERATION SHARE OF COLD-STORE ELECTRICITY = 60-70%. Verbatim (Abstract): 'Within cold storage facilities 60-70% of the electrical energy may be used for refrigeration.' Verbatim (Introduction): 'Asano & Mugabi (2013) stated that within cold storage facilities, 60-70% of the electrical energy may be used for refrigeration.' ORIGIN: the figure is attributed to Asano & Mugabi (2013); Evans et al. carry it. Lets a model split a whole-store electricity figure into refrigeration vs ancillary (lighting/fans/defrost/controls). [verified verbatim against the raw]"
  source_location: "Evans et al. (2015), Abstract and §1 Introduction (PDF p.3), citing Asano & Mugabi (2013)."
- claim: "CHILLED COLD-STORE SEC (all stores, n=167). Mean 55.7 kWh.m-3.year-1; Minimum 4.4; Maximum 250.4; Standard deviation 34.7. (Chilled stores modelled/measured around a 2 degC setpoint.) [verified verbatim against Table 3]"
  source_location: "Evans et al. (2015), Table 3 'Range in SEC values for cold stores examined', column 'Chilled (kWh.m-3.year-1)', row 'All stores' (PDF p.6)."
- claim: "FROZEN AND MIXED COLD-STORE SEC (all stores, n=262). Mean 71.5 kWh.m-3.year-1; Minimum 6.0; Maximum 391.6; Standard deviation 40.6. (Frozen stores modelled at -23 degC; 'mixed' = chilled and frozen run off one refrigeration system.) Frozen/mixed SEC is significantly higher than chilled (ANOVA P<0.05), while frozen vs mixed were not significantly different (P=0.44), so the two were combined. [verified verbatim against Table 3 and §3.1]"
  source_location: "Evans et al. (2015), Table 3, column 'Frozen and mixed (kWh.m-3.year-1)', row 'All stores' (PDF p.6); combination rationale in §3.1 Survey data (PDF p.6)."
- claim: "TRIMMED (OUTLIER-REMOVED) SEC BANDS. With the upper AND lower 20% of values removed, chilled SEC spans 31.3 to 70.6 kWh.m-3.year-1 (mean 50.2, SD 10.3) and frozen/mixed 40.0 to 93.0 (mean 66.2, SD 14.8). With the upper AND lower 10% removed, chilled 21.7 to 86.1 (mean 50.9) and frozen/mixed 31.1 to 119.4 (mean 67.2). NOTE: Table 3 labels these rows 'Upper 10%/20% removed', but §3.1 states the ranges are 'for all stores with the 10% and 20% upper and lower values removed' (both tails). The full-set range was 246.0 (chilled) / 385.6 (frozen/mixed); trimming the top and bottom 10% cut this to 64.4 / 88.2 - i.e. a small number of extreme stores drive the wide full ranges. [verified verbatim against Table 3 and §3.1]"
  source_location: "Evans et al. (2015), Table 3, rows 'Upper 20% removed' and 'Upper 10% removed' (PDF p.6); trimming interpretation and range-reduction figures in §3.1 (PDF p.6)."
- claim: "NEW ZEALAND CROSS-REFERENCE (in this paper's literature table). Table 1 lists a NEW ZEALAND cold-store SEC range of 26 to 379 kWh.m-3.year-1, source Werner et al. (2006). Full Table 1 (published survey SEC ranges, kWh.m-3.year-1): UK 4.4-250.4 (Evans et al. 2014a); USA 15-132 (Singh 2006); USA 19-88 (Elleson & Freund 2004); New Zealand 26-379 (Werner et al. 2006); Europe 30-50 (Duiven & Binard 2002); UK 34-124 (ETSU 1994); Netherlands 35 (Bosma 1995); Germany 101-1520 (Carlsson-Kanyama & Faist 2000, reporting BELF 1983). The NZ line is Werner, S.R., Vaino, F., Merts, I. & Cleland, D.J. (2006) 'Energy use in the New Zealand cold storage industry', IIR-IRHACE Conference, Auckland - the same NZ group (Merts, Cleland) behind the Frater/Merts & Cleland apple-coolstore anchor. [verified verbatim against Table 1 and the reference list]"
  source_location: "Evans et al. (2015), Table 1 'Publications listing energy used in food cold stores' (PDF p.4); reference 16 (PDF p.10)."
- claim: "ACHIEVABLE SAVINGS + DATASET/MODEL SCOPE. Verbatim (§1): earlier surveys 'demonstrated that energy savings of around 30-40% were achievable by optimising usage of the stores, repairing current equipment and by retrofitting of energy efficient equipment.' The analysed dataset = 167 chilled + 187 frozen + 75 mixed stores; the mathematical model ran 8 usage scenarios (Scenario 1 = all heat loads minimised, high efficiency; Scenario 8 = worst feasible) over store volumes 10 to 350,000 m3, with a chilled store at 2 degC and a frozen store at -23 degC. 13% of chilled and 5% of frozen/mixed stores performed WORSE than Scenario 8 (the worst modelled configuration), and ~2% performed below Scenario 1. [verified verbatim against §1, §2.2 and §2.4/§3.1]"
  source_location: "Evans et al. (2015), §1 Introduction (30-40% savings), §2.2 Mathematical model + Table 2 (scenarios/volumes/temps), §3.1 Survey data (dataset counts), §2.4/Figures 7-8 discussion (scenario over/under-performance) (PDF pp.3-8)."

Neobiome Intelligence relevance

Feeds the NI energy model (edt energy-demand layer) as the INTERNATIONAL cold-store electricity benchmark for sizing a community’s produce cold-storage load. Concretely:

1. Per-volume coefficient (kWh/m³/yr), split by store type. For chilled produce (fruit/vegetables near 0-5 °C) use the chilled band: all-stores mean 55.7, robust central band (20%-trimmed) 31.3 to 70.6 LIT_119. For frozen/long-term use the frozen/mixed band: mean 71.5, central band 40.0 to 93.0 LIT_119. Frozen/mixed runs ~28% higher than chilled at the mean, so the model should carry separate chilled vs frozen coefficients, not one blended figure.
2. Refrigeration-share splitter: 60-70% LIT_119. Where only a whole-store electricity total is known, apply 0.60-0.70 to isolate the refrigeration load (or, inversely, gross up a refrigeration-only figure to a whole-store demand). This is the coefficient that reconciles “refrigeration system” figures against “whole cold store” figures across the corpus’s coolstore sources.
3. International bracket around the NZ anchors. Evans’ chilled range comfortably contains the NZ apple/cold-store figure of 79 kWh/m³/yr (35-151) (Merts & Cleland 2004 via Frater), and Evans’ own Table 1 carries the NZ 26-379 survey line (Werner et al. 2006) LIT_119 from the same NZ research group. So the NZ-specific and international evidence corroborate rather than conflict: NZ values sit inside the international distribution. Prefer the NZ figures for NZ calibration; use Evans for the chilled/frozen structure, the refrigeration-share split, and international robustness.
4. Small-store direction of error. The dataset and model span large industrial stores (surveyed/modelled up to 350,000 m³); Evans notes 13% of chilled stores performed worse than the worst modelled configuration LIT_119. A community-scale store (tens to hundreds of m³) has a worse surface-area-to-volume ratio, higher relative infiltration/door and part-load losses, so it should be modelled toward the upper part of the chilled band (mean-to-max), not at the mean. Recorded as a modelling caveat (see Research targets), the same small-store caveat that attaches to the Frater NZ anchor.

PROXY flag (data_quality high, but not NZ). The dataset is international and UK/European-weighted; location (country/continent) had minimal effect on chilled SEC (temperature zone mattered only for frozen), which supports transferring the ranges to NZ, but the figure remains a proxy. It is high-quality (peer-reviewed, large n, verbatim-verified) and cross-corroborated by the in-corpus NZ evidence, but for a hard NZ model cell the NZ-specific coolstore sources take precedence.

Underlying / related sources

• Asano, H. & Mugabi, N. (2013) — origin of the 60-70% refrigeration-share figure Evans et al. carry. Not separately retrieved; recorded for provenance.
• Werner, S.R., Vaino, F., Merts, I. & Cleland, D.J. (2006) Energy use in the New Zealand cold storage industry, IIR-IRHACE Conference, Auckland — the NZ 26-379 kWh/m³/yr line in Table 1; same NZ group (Merts, Cleland) as the Frater/Merts & Cleland (2004) apple-coolstore anchor. Not separately retrieved; the NZ range is already held via this paper’s Table 1 and, at finer detail, via the Frater primary.
• Foster, A. et al. (2013) Freely available cold store energy models and Evans et al. (2014a) Specific energy consumption values for various refrigerated food cold stores, Energy and Buildings — the model and the prior (294-store) dataset this paper extends. Contextual; not ingested.

Notes

• No printed page numbers. The paper carries only a running footer, so source_location citations anchor on the table/section number and give the physical PDF page (paper body = PDF pp.3-11; covers = pp.1-2). Proceedings pagination differs; the figures are verbatim regardless of which pagination is used.
• Verified-source-only. Every figure above was extracted with pdftotext -layout and confirmed verbatim against the raw. The sha256 (cd31a0a0…) is of the paper PDF (single file).
• Trim-label subtlety. Table 3’s rows read “Upper 10%/20% removed,” but §3.1’s prose confirms the tighter ranges are after removing the upper and lower tails (both ends). The claim above records this so the bands are not mis-cited as one-tailed.
• PROXY caveat. International dataset, not NZ; the NZ-specific coolstore sources (apple/cold-store, kiwifruit) are the NZ calibration; Evans is the international bracket + the refrigeration-share splitter + the chilled/frozen structure.

Research targets

Documents to retrieve

• Werner et al. (2006) / Merts & Cleland NZ cold-storage surveys — the NZ origin of the 26-379 (and 35-151/79) ranges. Recorded here, NOT raised as a new RT: the NZ range is already held via this paper’s Table 1 and the Frater primary; retrieving the underlying NZ surveys would add distribution detail only, not a new headline number.

Research gaps

• Small community-scale coolstore SEC (uplift factor). Evans’ dataset/model skews to large industrial stores; a community store (tens to hundreds of m³) will sit above the mean kWh/m³ (surface-area-to-volume, infiltration, part-load). Recorded as a modelling caveat, not a new RT — the same caveat attaches to the Frater NZ anchor, so it is a single cross-source modelling note, not a per-source retrieval target.

Connections

Referenced by

Sources (2): CR_058 · OT_206

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